The Center is led by Derek Chollet, a veteran of the State Department, the Pentagon, and the White House who held senior posts in both the Obama and Biden administrations: Assistant Secretary of Defense, under Secretaries Leon Panetta and Chuck Hagel, Counselor of the State Department, and Chief of Staff to the Secretary of Defense. In his early career, Chollet assisted in researching and writing the memoirs of prominent foreign policy practitioners, including former Secretaries of State Warren Christopher and James Baker, and Deputy Secretary Strobe Talbott. He is the author of several books of his own, including The Middle Way, The Long Game, and The Road to the Dayton Accords.

I sat down with Chollet to discuss the impact of ongoing wars in Ukraine and the Persian Gulf, the future of American military alliances, and JPMorgan’s approach to geopolitical risk. What follows is a transcript of our conversation. 

CB: Why did JPMorganChase decide to establish the Center for Geopolitics last year? It's not the only bank that has done this. What inspired this decision? And then, how are you interfacing with your clients?

DC: Jamie Dimon, both last year when we did our rollout and then just in the last few weeks as we've put out our one-year-anniversary compendium of “greatest hits,” wrote a note that talks about the rationale he brought to it. It came from a conversation that started with him and other senior leaders here at the end of 2024, when they saw an insatiable demand from clients in the United States and around the world — from all of our lines of business, whether it's the private bank, the commercial and investment bank, or the consumer bank — to better understand what's happening in this complicated time we're in.

It seemed pretty compelling back then, at the end of 2024, and events since then have made it more so. I've been struck by how, in surveys our bankers do with their clients, geopolitics ranks number one, two, or three among their concerns across the board. And it makes sense, given that we've got two important strategic arenas for the United States, for our economy and our security — Europe and the Middle East — where we have sustained conflict in both places. And in Asia and around the world, we have a defining relationship between the US and China. On any given day, there's something in the news that's geopolitical in nature and impacting markets. We're living at a uniquely historic moment — we haven't dealt with this combination of issues all happening at once since the 1930s. So it's understandable why, for many clients in the United States and around the world, it's on their minds: where is the world headed, what does it mean for us, for our countries, for our companies, for our investments? Our job is to work alongside our banking colleagues to help these clients navigate the landscape.

CB: And what specifically is the Center producing for your clients? Are you strictly producing the reports you've made public, or is it more involved than that?

DC: It's engaging with clients across the board. Since we started, we've had hundreds of engagements with clients all around the world — one-on-one engagements, large group engagements, large conferences we participate in. In addition, we have written products, some of which are available on our website, and some of which go only to our clients — not widely distributed publicly, though they still reach many clients. So it's a combination of those two things. What we try to do is not just give a news digest or tell people what happened, but tell them why something is happening, why it matters, and offer some forecasting on what's coming next.

In doing that work, we're a small team, so we work closely with our colleagues in research as well as in the risk operation to help serve clients' needs. Internally, I think of us as a force multiplier — like special operators in the military. You're there working alongside your colleagues to enhance their efforts, giving them added bandwidth and resources to serve clients. J.P. Morgan is a company with global reach and a tremendous amount of resources, assets, and expertise, so we're here to tap into that and be a central node for the firm.

One closing observation: as I've gone around the firm meeting almost all the senior leaders here, for most everyone, geopolitics has been their second or third job in addition to their primary job. Our team goes to bed at night and wakes up in the morning thinking about geopolitics. So we're here as a service to our colleagues, but also, importantly, to our clients as they're seeking more insight into what's going on.

CB: I always like to ask people who consume large quantities of information and have to forecast and extrapolate from it — how do you identify quality information from low-quality information? Where do you go for your information?

DC: It's a great question. One of the things we're helped by is that I and our team have a lot of experience — we've been in government, in the trenches, working these issues. Obviously we no longer have access to that information, but our ethos is: we read everything so our clients don't have to.

I consume a lot through the press — following very closely in the US and around the world, in all its forms, whether Substacks or mainstream media. There are several good Substacks, though many of them are more niche than broad-brush "here's what's happening in the world." It's more like: I want to follow someone who's really smart on China, who speaks Chinese, and I learn a lot from them about China, which I can then put together with everything else I'm absorbing.

We also spend a lot of time talking to people in the US and around the world — myself and the team are still very much on the policy circuit, learning from former colleagues and other experts about how they see things. And then there's what J.P. Morgan itself brings to bear. I'm humbled by the amount of information and resources J.P. Morgan puts out — much of which I paid far too little attention to when I was in government; I should have paid more attention. In all our work, we try to leverage the good work of our colleagues who put out deep analyses of certain markets or sectors of the economy, and lift up what they're doing. I think we serve clients best when, say, in a discussion on Iran or events in the Middle East, it's not just us contributing — it's also our research colleagues doing commodities analysis alongside our experts on the US economy, so clients understand what's happening geopolitically, in energy markets, and in the US economy all at once.

CB: Scott Bessent told the Wall Street Journal earlier this year that the US-China relationship will define the success of the Trump administration. Where are we at this stage in that relationship?

DC: I think he's absolutely right, and it won't just define this administration — it will define geopolitics, not just the future of the US but the futures of many other countries around the world, which will be impacted by the trajectory of the US-China relationship.

At the end of September, President Xi is scheduled to come to the United States. So the US-China relationship has a healthy foundation to work from right now, as a result of the President's visit to China in May and President Xi's planned visit here in September. That follows a pretty rocky back-and-forth last year in the wake of Liberation Day and the US-imposed tariffs and China's retaliation against those tariffs — China was one of only two countries in the world to retaliate against the United States, the other being Canada. We're at a point of truce in that back-and-forth now. I don't think we've resolved any of the underlying issues or differences between the two countries, but we've at least arrested the back-and-forth over tariffs and counter-reactions.

CB: In terms of leverage — how far has China come in mitigating the leverage the US has over it, and vice versa?

DC: My assessment is that Beijing was among the least prepared for the first Trump administration, and they learned a lot of lessons from that experience. Then the Biden administration came in and largely continued many of the first Trump administration's policies on tariffs, export controls, and the strategic outlook of competition with China, including its military investments. So I think the Chinese were among the best prepared, if not the best prepared, for Trump 2.0, and they showed that last year in their retaliation — which also had the effect of being something of a Sputnik moment for the United States.

It wasn't a secret, at least to those of us who'd worked on the relationship from 2021 to 2025, that the US had built up a lot of vulnerabilities to China over the years, particularly in critical minerals and rare earths, and that China had effectively created an ability to weaponize those dependencies. They used those weapons last year. That's been a wake-up call, certainly for the US government and the private sector, in trying to reduce those dependencies, and I think progress is being made. It's an area where J.P. Morgan is contributing, helping finance and facilitate some important deals to help build reliable supply chains here in the United States, whether through resourcing, reshoring, or friend-shoring with more reliable partners. We're not there yet, but it's a fixable problem — there's money for it, there are government policies for it — it's just not going to be fixed in a year. It will take some time to reduce those dependencies.

CB: What would you say we've learned about China over the last few months as the war with Iran has progressed? Anything meaningfully new or surprising in how China responded — how they assisted Iran in evading sanctions, for example?

DC: For me it wasn't so much a surprise as a reaffirmation of what I've observed over the last several years: China is not particularly interested in actively playing a role in solving problems around the world. It's laser-focused on its goals of dominating global manufacturing and building dependencies — making the world more reliant on China and China less reliant on the world. There were some modest attempts, phone calls to try to de-escalate things, but you didn't see China playing much of a role at all in trying to resolve the crisis or bring it to an end.

There could be a couple of reasons for that. I think it's reflective of their strategic outlook, and also of the fact that they'd built up significant energy reserves that let them economically weather the storm in the Middle East more than expected, so they didn't have a real need to get in there and fix things. But more importantly, it's their strategic outlook — this just isn't something they see as their role in the world, now or for the foreseeable future. It's not consistent with their overall strategic goals for the country.

CB: On the impact of that conflict on the American alliance system — two points. First, our allies and partners in East Asia, who are much more dependent on Gulf energy suppliers, like Japan and South Korea — what lessons have they taken from this? Would you say those lessons translate to reduced trust in the United States? Second, the Gulf countries, where we have very involved security arrangements — we've already seen Zelensky visiting and signing parallel security agreements with the Gulf countries. What's to stop them from going to China for interceptors or other defense assets, given we've demonstrated we don't necessarily have the capacity to look out for their interests at all times?

DC: In those two regions, which have been directly and significantly impacted by the Iran war, I haven't seen evidence yet that it's fundamentally frayed alliances — with the huge caveat that we're still in the first half of this Iran conflict; we've got a ways to go, and a lot will depend on how it ends, whenever that is.

In Asia, I think the biggest lesson is less about US alliances and more about building greater resilience and diversification in energy sourcing, so countries are less reliant on Gulf energy. I haven't seen anything to suggest a meaningful impact on US alliances in the region, which remain pretty strong. That matters for US-China competition, because that competition looks quite different, and more favorable to the US, when measured as the US-plus-allies-and-partners versus China-plus-its-allies-and-partners, of which there are very few, if any. The Chinese understand that — they have a sophisticated understanding of the totality of American power, not just what the US brings on its own but what it does alongside its allies and partners.

In the Gulf, the picture is more complicated. This is a war the Gulf countries did not wish for — none of them were friends of Iran, but none wanted the conflict to unfold this way. In many ways it confirmed their fears that they'd end up targets of Iranian retaliation, which is what we saw. At certain points during the war, they've also felt they weren't adequately consulted — I'm not in government, so I can't say who's right or wrong there, but there have been some hiccups in those relationships. That said, they clearly still rely on the US for the bulk of their military power, and I expect military relationships across the board to only get stronger.

What we've seen in terms of the Gulf's ability to defend itself throughout this crisis is, in many ways, the fruit of something that's been growing for 15 years — deepening military-to-military cooperation between the US and Gulf partners, including integrated air and missile defense, with the US military as a central node. That's a big part of why the Gulf states haven't taken more damage than they have.

Is it true they're diversifying? Yes — if you want state-of-the-art counter-drone systems right now, you go to Ukraine, because the Ukrainians have been shooting down Iranian-made drones for three or four years. I think it's a good thing the Gulf countries are seeking to diversify some of their defense relationships. But overall, across the Middle East, I think the relationships, despite some hiccups, are fundamentally strong, and I expect US defense relationships to maintain that strength — and probably get even stronger — as this crisis plays out, which it will for a while.

CB: On the defense technology side — who do you see as the big winners, not just from the Iran conflict but from all the recent developments in warfare over the last five or six years: drones, ballistic missiles, and the response to them? I've been talking to a lot of people from companies like Anduril — the name of the game seems to be scale and increasing production volumes. What else do you see as a major upside in the defense sector as a result of these conflicts?

DC: Scale is a big piece of it — being able to produce attritable systems at scale. The unmanned space writ large is going to be critically important, along with the integration of autonomous technology and AI into that space. We've seen some of that play out in the Middle East and certainly in Ukraine — both theaters are, in many ways, incubators of cutting-edge defense technology. There's been a tremendous amount of work and thought given to how the US and its partners can promote greater innovation in the defense sector, improve procurement, move more quickly, and allow companies to take greater risk in their production lines. This is another area where J.P. Morgan is trying to do its part — through financing, facilitation, and advisory work, as well as in the policy space — to help militaries around the world, the US and its allies, get the systems they need when they need them, and bring that to scale.

There are a lot of real-time lessons we're learning, and more we'll learn in the coming years as we sift through how we got here. In many ways, Covid and Ukraine were the wake-up calls for the defense industry. Covid exposed the vulnerability of supply chains, known issues but which we all started actually experiencing. Ukraine exposed the challenges the US defense industrial base has had in replenishing the stockpiles we sent there. It uncovered vulnerabilities and weaknesses in the system that we'd had for decades but hadn't fully realized. When we burn through the bulk of a certain munition, we can't rebuild it at a satisfactory rate. We're learning that in real time today with Iran: when you burn through a thousand Tomahawks and it takes three years to build that back, that's not an acceptable result for a military facing the challenges it does.

CB: I wanted to pivot slightly to China's relationship with Europe. Where do you anticipate the trade relationship between China and Europe going in the next few months?

DC: I think the theme of de-risking that Europe has pursued for the last several years is, in many ways, a fulfillment of something the US was seeking for years. I remember going to Europe 10 to 15 years ago, meeting with European colleagues, and it was hard to have a common conversation about China, because the strategic debate here in the US about China was in a fundamentally different place than where Europe was. The US was still doing a lot of business with China — still does — but the sense of China as an emerging strategic competitor, the possibility of confrontation, just wasn't a conversation Europeans were having then. That started to change around that time, and changed fundamentally in the last five years. Europe hasn't really moved off that.

You still see significant internal debates all around Europe about overreliance on China, alongside recognition that China is a major market with important economic ties to Europe. It's not an on-off switch — either doing business with them or treating them as enemies and doing nothing. It's more nuanced than that, similar to the situation here in the United States. Is it possible China will seek to take advantage of turbulence between the US and Europe down the road? Perhaps — but I don't know that they'll necessarily succeed, given where most European governments currently stand on China.

CB: On that point — what kind of incentives does China have to offer Europe in that scenario, trying to pull them away from the US?

DC: Some of it would come into play if there's an increased tariff war between the US and Europe — China could take advantage of that, perhaps. But at the same time, Chinese manufacturing dominance is becoming a real issue for many European countries, and China's increasing dominance in electric vehicles is becoming a real issue for European car companies trying to compete. So China has leverage, and it has carrots to offer, but it's also presenting a set of challenges that I think will make many Europeans think twice about deepening their dependencies and reliance on China.

CB: In terms of the long-term US-Europe relationship, I guess there are roughly three scenarios: one where the status quo is more or less kept in place, one where there's a pivot to a more offshore-balancing posture, and one involving a hemispheric-defense-style withdrawal from Eurasia — over the long term, which of those do you think is most likely?

DC: I think the US-Europe relationship is actually better than it sounds. Just in the last few weeks there was a largely successful G7 summit hosted by President Emmanuel Macron, and the NATO summit accomplished a lot of good things on defense investment. Is there drama around the relationship? Yes. Will there be fundamental questions moving forward in Europe about American reliability? Yes. But Europe is spending more on its defense than it has in the past, for a number of reasons, and whatever those reasons, that's a good thing from my perspective. Europe is also looking to be more innovative in its defense spending. Despite the US largely pulling back from direct support of Ukraine over the last year and a half, Europe has filled the gap and done so with a large degree of success — where Ukraine stands today is first and foremost a testament to Ukraine's own innovation and resilience, but also to our European partners for stepping up in ways that may have been somewhat unexpected.

The US-European trade relationship, despite tariffs and the back-and-forth we've seen, remains the most important relationship the United States has. Will the US force posture in Europe change? Sure — that's something we'll continue to debate, and it's natural for any government to periodically assess whether it has the right capability in the right place. But I don't see offshore balancing, and I don't see hemispheric defense. I don't see a scenario where the US pulls out of Europe completely.

I think the real question, going back not just the last year and a half but really the last 20 years, is whether the US is reliable — whether we're going to be there for our partners. I'm old enough to remember debates in Europe 20 years ago about the US not caring about them because it was mired in wars in the Middle East, or during the Obama administration, about pivoting to Asia meaning pivoting away from Europe. There are always anxiety points in the US-European relationship, and we're at one of those points now. Some elements of it may be worse today than before, but I still fundamentally think the relationship remains quite strong. That said, you never take these things for granted — there's no divine right to success, and the alliances the US has weren't just ordained; they're relationships you have to work on. Maybe we're in a bit of marriage counseling right now, but we're still in the relationship, and still working at it.

CB: We rolled out extensive sanctions against Russia in 2014 and have expanded them considerably since. How much more room is there to give on US sanctions as a tool of coercion? How much further could we go before we start to undermine our own system?

DC: I think there's quite a ways to go. Depending on when this runs, we may by then have what's being informally called the Lindsey Graham bill move through Congress, which would further tighten the pressure on Russia and give the President greater authority to ratchet it up. I'd argue that one of the less fortunate byproducts of the Iran war was a modest boost it gave Russia — in order to keep energy markets stable, there was a temporary loosening of some sanctions that allowed more Russian oil onto the market, putting more money into Putin's coffers. But that was temporary and has since been rolled back.

I think there are more tools available, and they'll likely be used — probably by the time this runs. I don't foresee an investment environment in Russia becoming favorable anytime soon. Even if one were to wave a magic wand and resolve the Ukraine war in an acceptable way, providing grounds to lift some sanctions, Russia is in a pretty difficult place now and will likely remain there for some time.

One thing I'm keeping an eye on geopolitically, that isn't really on most people's minds right now, is a post-Putin Russia. There's more and more talk of internal pressure on Putin and questions about the Ukraine war. I don't foresee any near-term issue, but it's worth keeping in mind — Putin has been in power, one way or another, for 26 years now, and there isn't a clear sense of who comes next once he's gone. It wasn't that long ago — three years — that there was an attempted coup, when the Wagner Group made its "thunder run" toward Moscow before being repelled. So I think we could expect more uncertainty emanating out of Russia over the medium to long term.

This transcript has been edited for length and clarity.

That's starting to change. Katie Miller, wife of White House deputy chief of staff Stephen Miller, former DOGE aide, and a reliable weather vane for MAGA sentiment, posted that solar is "the energy of the future." The sun, she added, is the "giant fusion reactor up there in the sky." Fabrizio, Lee & Associates, Trump’s own chief pollster, found that 70% of Republican voters back utility-scale solar when the panels are American-made; Kellyanne Conway’s firm found 75% of Trump voters across Arizona, Florida, Indiana, Ohio, and Texas want solar used to strengthen the US energy supply. In 2025, 73% of all new US solar capacity went up in red states. Even Energy Secretary Chris Wright (who told a House subcommittee in June 2025 that an intermittent source like solar is "just a parasite on the grid") insists he's "pro-solar," and says the technology should compete on its own merits.

Solar's new trendiness has an obvious explanation. Data centers now need over 100 gigawatts of new capacity by 2030, which is roughly the output of 100 large nuclear reactors. Residential electricity rates climbed about 10% nationally over the past year, with more than a dozen states seeing double-digit jumps. China installed 315 gigawatts of solar in a single year — more than most countries have ever built. Solar made up 85% of new capacity added to the American grid in Q3 2025; solar paired with battery storage is projected to be responsible for 79% of new capacity in 2026.

Solar's case is being made on deployment speed, cost, and geopolitical urgency — all real concerns — but not on solar as a technology. The underlying electrical architecture inside a solar installation has barely changed since the early 2000s. Worse, utility-scale farms routinely underperform their own production models by eight percent on average, leaving billions in lifetime value on the table.

One company has spent nine years betting that solar can be smart. Optivolt, founded by Thiel Fellow Rohit Kalyanpur, embeds aerospace-grade power electronics directly inside panels to intelligently modulate power flow in response to changing weather, light, and demand conditions. They claim enormous improvements, like up to 6x better panel performance in shaded conditions, compared to legacy installations.

Optivolt further bets their technology tips both sides of the levelized cost of energy (LCOE) ratio, total system cost divided by the total energy a system produces over its life. If they’re right, they’ll fundamentally change solar’s unit economics. 

Solar cells within a panel are wired in series, so electricity flows through every cell in sequence, like bulbs on a string of Christmas lights. If cells get shaded (by a tree, dirt, bird poop, snow) they choke the current flowing through their section of the panel. The industry's usual fix is the bypass diode: panels are divided into a few zones, each with a diode that trips when shading drags its zone down, routing current around the whole zone so the rest of the panel keeps working. But the diode is only a safety valve. The energy the bypassed zone could still have produced is thrown away, not recovered. Because a single diode switches off a third of the panel at a time, a single shaded cell in a 108-cell panel can erase more than a third of its output. Plus, when shading is light enough that the diode doesn’t trip, the shaded cells simply drag their whole zone down anyway. Diodes also sometimes fail, and they don’t stop hotspots: roughly one panel in 10 develops a hotspot or diode failure over its life. Hotspots cook the cells around them and, on a roof or in a field, can start fires.

And shade is everywhere. Most American rooftops have some mix of shade, dust, or debris; every utility farm has soiling and interrow shading. The problems compound year over year as cells age at different rates.

The average solar farm underperforms projected performance by eight percent, which works out to $20 to $40 million in lost lifetime revenue per site, and up to $4 billion across a 10-gigawatt portfolio. To manage all these vulnerabilities, the industry has tacked on a series of improvised solutions that never resolve solar’s underlying design flaws, like drone flyovers with thermal cameras, manual fault isolation, and repair crews sent out with almost no data. This is how the industry “maintains” gigawatt-scale energy assets.

Plus, all energy optimization currently happens outside the panel. Solar installations come with power electronics, meaning components that manage, convert, and route the raw direct-current (DC) electricity panels generate. (This includes inverters, microinverters, optimizers, charge controllers, and, yes, even the bypass diode.) Together, they determine how much of what a panel makes actually reaches the grid. 

Unfortunately, panel manufacturers (mostly Chinese, commodity-focused, running on razor-thin margins) and power electronics manufacturers (higher-margin, IP-driven, usually American and Israeli) almost never collaborate. Enphase and SolarEdge built the category of module-level power electronics (MLPE) in the past 15 years, which made solar meaningfully better; they deserve credit. But today’s solar electronics are still just band-aids bolted onto a flawed design. 

A microinverter (the small box that converts and tunes the output of a single panel) can decide how much power to pull from that panel, but it can't move energy between the cells inside it. When a leaf covers three cells and drags a whole zone down to zero output, the microinverter just works with what’s left.

And electronics are expensive. Tesla published a white paper analyzing roughly 13,000 residential solar sites and found that for 80% of homes (those with moderate to good sun) module-level electronics delivered only one to two percent more energy than a plain string inverter, despite costing thousands of dollars more per install. You can imagine what that does to system-level LCOE.

No one tried to balance power at the level of cells themselves because the industry agreed it was impossible. 

Kalyanpur was 19, a computer engineering student at the University of Illinois, when he started working on solar energy. Every project just kept pushing him to solve what the rest of the industry had written off: redesigning how a solar panel is built from the ground up, with better power electronics that are embedded inside the panel itself. 

In 2017, Kalyanpur founded Optivolt and built what he calls the Power Balancer. It redistributes energy inside the solar panel at the substring level in real time, letting weaker cells borrow from stronger ones; instead of a shaded cell dragging down its neighbors, the strong cells carry it. Active redistribution replaces passive bypass, so bypass diodes vanish entirely. So do two of every three junction boxes, four of every six connectors, trunk cables, and combiner boxes. Folding the electronics into the panel makes the entire solar install simpler. Optivolt holds 13 patents on the architecture, with 25 more pending.

Convincing solar manufacturers to abandon their previous designs to try something completely new was not easy. Three barriers explain why everyone thought this was impossible:

Heat. Power electronics generate heat; solar panels degrade from it. Putting one inside the other was considered a non-starter. Optivolt’s answer was to make the electronics barely generate heat at all: the Power Balancer runs at 99.4% efficiency in typical conditions, so only about 0.6% of the energy passing through it is lost as heat, and the panel's thermal, materials, and mechanical design absorb that small amount gracefully.

Making power electronics generate almost no heat required aerospace-grade power-systems engineering: the kind done on satellite programs, where electronics survive decades of thermal cycling with no viability of repair. To do so, Optivolt recruited chief architect Linda Irish. She holds more than 40 electronics patents and previously designed power systems for a Jupiter satellite program and wireless-power systems at Qualcomm.

Regulation. No safety standard existed for module-integrated power electronics at UL, the private lab whose certifications electrical codes and insurers defer to; there was no easy way to certify the safety of Optivolt’s products. Optivolt invested four years in defining new safety test standards with UL, work funded in part by the US Department of Energy.

Cost. Integrated electronics had to replace the panel’s existing junction-box components, not add to the bill of materials. Optivolt’s supply-chain work brought the Power Balancer to a net simplification: fewer parts, lower total system cost.

Years of work later, Optivolt has a device that survives the same thermal cycling as the panel it lives in, backed by a 30-year warranty. The Power Balancer is built in Mexico and integrated into panels assembled in California. Kalyanpur, now 28, has been building toward this for nearly a third of his life.

Most solar innovations move just one side of the LCOE ratio. A cheaper panel lowers the numerator, the system's cost; a more efficient cell raises the denominator, its lifetime output. Either way, the gains are incremental.

The electric-vehicle market was stuck in a similar problem for years: electric cars cost too much and couldn’t go far enough. Finally, improved battery chemistry, vertical integration, and manufacturing scale pushed on cost and range at once. Kalyanpur hopes his technology unlocks a similar transformation. On the cost side, Optivolt’s Power Balancer replaces many existing hardware components rather than adding new ones, and no additional electronics are needed to retrofit in the field. (Optivolt panels are also far cooler when something blocks them. Hotspot testing has recorded shaded cells approaching 200 °C, hot enough to burn cells and fail encapsulation. This is why solar farms sometimes catch fire. Cooler panels can last longer and incur fewer maintenance costs.)

And as far as energy output, Kalyanpur says Optivolt delivers up to 50% more systems-level energy in real-world shaded conditions, and up to six times the yield in heavy shade. Utility farms see less shade than rooftop solar but still bleed energy to soiling (dust, debris, row shading). 

Put the two together and Optivolt projects a residential LCOE roughly 30% below a conventional system's (the exact figure varies from home to home). And for utility, their 250-megawatt simulation showed 13.8% more energy under heavy dust and soiling, worth $42 to $60 million in additional lifetime revenue per site. 

Kalyanpur is ready for solar to go mainstream.

Optivolt has shipped more than 20,000 off-grid units, secured US military and government contracts, and is launching its residential system in California in 2026. Their first utility-scale deployment is planned for later this year.

By figuring out how to embed power electronics directly into panels, Optivolt has also embedded sensors that monitor realtime panel output, temperature, and health. For utility solar, this visibility hopefully ends the need for drone inspections; at home, owners track these metrics through an app. Optivolt believes this information can proactively alert owners of maintenance issues, settle warranty claims against actual production curves, and help operators underwrite financing on verified output rather than projections.

Kalyanpur’s ambitions still manage to go further. He thinks fully off-grid data centers, run on solar and batteries, are only years away: cell-level balancing improves uptime enough to make them viable. Longer term, the team is looking at space-based solar (satellites that gather sunlight in orbit and beam the power down), where the same aerospace engineering behind the Power Balancer applies almost directly.

The technology to close the gap between solar’s performance and its potential is coming. It just involves mental resolve, tedious regulatory work, and teaming up with satellite engineers to solve some hard physics.

Sunlight is the most abundant energy source available to civilization yet we somehow capture almost none of what it offers. Solar isn’t intermittent the way people casually assume; it's predictable, cyclical, and, with enough capture and storage, continuous. Difficulties arise only between photons hitting solar cells and electrons reaching the grid. Fortunately, these are challenges ripe for new technology, not permanent fixtures of the hard laws of physics.

Deep beneath the Luzon Strait, the body of water connecting Taiwan to the northernmost island of the Philippines, a handful of fiber optic submarine cables were severed by the quakes. Within hours of the first tremors, cable operators reported that six of the seven cable systems in that corridor were failing. From the shoreline, nothing looked different. But online, the “cloud” suddenly buckled. International communication and digital activity on Taiwan and its neighboring nations was interrupted. Customers could not withdraw money from banks, traders could not fill orders, airlines could not check in customers.

By the next day, the two largest Chinese telecom companies reported that their capacity to connect users with the US and Europe was down over 90 percent, severely limiting international communications and business activity. Taiwan’s largest internet service provider (ISP), Chunghwa Telecom, lost all connection to Hong Kong and the rest of Southeast Asia, forcing firms to re-route traffic through satellites and alternative cable routes near Singapore. Taiwanese users, along with others across the region, were completely shut out from sites such as Yahoo, MSN, and Hotmail.

While the disruption is most apparent at the edge of the digital network, in this instance landing pages and email servers, it could not be solved by pushing a software update. Fixing this problem would require physical infrastructure: ships. On December 28, repair vessels started to arrive. Behind the scenes, industry investigators discovered that a submarine landslide likely compounded the damage, snapping and displacing cable segments in a way that made recovery slower and more complex than a single clean break.

People who lived in Taiwan and the surrounding region could access the internet within a few days, as operators rerouted traffic across longer paths until the seafloor could be stitched back together. By January 5th, Hong Kong reported that major ISPs were back to “reasonable levels” of service, even before the main cables were fully repaired. The episode’s lesson was startling: the modern internet can bend without breaking. But it still depends on a few slender strands of glass laid across a dark ocean trench — and on the rare ships, and the crews who can sail out, hook the seafloor, and knit the world back together.

What makes this tale so remarkable is that it was essentially invisible to the vast majority of people. You can climb the Hoover Dam, you can ride to the top of the Burj Khalifa, you can gaze in awe at the Hagia Sophia. But you cannot visit the pieces of what has become one of humanity’s greatest engineering achievements, because they lie in silence across continental shelves and abyssal plains, stitched together by repeaters, branching units, and the patience of crews who work with grapnels and ploughs in seas that do not care about your deadline.

You will likely never see these arteries of the global information flow. Maybe you’ll see a nondescript hut near a beach, a manhole cover in the sand, a warning sign on a fence that says “Danger — Buried Cable — Do Not Dig.” Through that unremarkable doorway runs the nervous system of modern life. The markets that open at dawn, the cloud that holds your photos, the calls between presidents, the texts you send without thinking — all of it relies on these cables. The world’s most consequential infrastructure is a bundle of glass fibers under miles of water, doing its work so quietly that we only remember it exists when it breaks.

Submarine Cables 101

For most people, the internet lives somewhere overhead. It is in the ether, the glowing nowhere from which emails emerge and into which tweets vanish. This is, of course, a charming fiction.

The reality is that the modern world is manufactured, dragged, bolted, buried, landed, maintained, and repaired. Submarine cables are one of those pieces of infrastructure that have a science-fictional character. The basic proposition is preposterous enough: take the most advanced communications technology of our age, wrap it in layers of metal and rubber, lower thousands of miles of it into the briny ocean, and trust that civilization can henceforth depend on it. Let us strip away the sci-fi for just a few moments.

As with any infrastructure deployment, the first step is planning where the cable will be laid. Mapping a route requires considerable diligence. There are constraints put upon these cables by man and the nation-state, such as security reviews of partners and financiers, and environmental permitting in and around the landing site. Cable projects span a spectrum of ownership models: a single company, a private consortium, a mixed consortium of private entities and state-backed telecommunications services or utilities, or a fully government-owned and controlled cable.

Then, there are the constraints imposed by the sea itself. Before a route is approved, engineers, oceanographers, and financiers need hydrographic surveys to identify topographical impediments or threats, geophysical surveys to identify hazards or threats on the seabed, and geotechnical investigations to determine the engineering required to secure and protect the cable.

Next, the cable must be prepared for deployment. Each segment can be more than 1,000 km long; the longest continuous cable in operation stretches over 45,000 km from the United Kingdom, around the Cape of Good Hope, and up to the Persian Gulf. The submarine cables in use today are marvels of layered engineering: thin strands of glass fiber-optics coated in a UV-cured acrylic material to prevent bending, bundled together with a copper conductor to enable connectivity. That fiber bundle is then encased in a copper tube to further protect the strands and conduct electricity from a landing station to “repeaters,” the copper components within the cable that amplify signals sent from the landing station along the cable, enabling consistent flows of energy and information. Over that comes a water barrier and polyethylene coating to ensure the cable is dry and secure. For particularly vulnerable routes, where fishing vessels or unforgiving submarine topography raise the risk of damage, such as the Japan-Guam-Australia South Cable and the EAR / Trans-Caribbean Fiber System, some cables receive armoring such as chains or metal shielding.

Just as with terrestrial fiber cables, deployment begins with trenching. But the submarine version is more onerous. First, a crew uses a drilling rig to put the cable in the ground and enable it to extend the cable into the shallow ocean. More modern cables leverage horizontal directional drilling, which allows a pipeline to pass through the coastline, emerging at a predetermined underwater location, through which the cable enters the water. From there, a cable-laying ship equipped with an underwater plow prepares the seabed, followed by a remotely operated vehicle (ROV), controlled by a crewmember, that buries the cable behind it. The depth and length of the burial depend on environmental factors such as local regulations or the potential for disruption by shipping activity. Once in deep water, most cables are simply laid on the seabed.

Once laid, cables can be disrupted and severed, requiring repair crews to retrieve the broken cable and repair it at sea. By international estimates, there are more than 200 cable disruptions requiring repair or replacement every year. The number has been steadily rising, and while many incidents are accidental, the growing scale of Chinese “gray fleets” — vessels operating outside normal regulatory oversight -— and the extensive history of Russian vessels purposefully severing cables mean that intentional disruption is becoming more common. As recently as 2025, a Chinese cargo vessel dropped its anchor six nautical miles offshore, dragging it along the seabed and severing the Taiwan-Penghu No. 3 cable in an act of blatant maritime sabotage.

Cable repair ships are among the most specialized vessels in operation, prepared to respond to cuts or disruptions within 24 to 48 hours — but repairs can take around a month on average, depending on where and when a disruption occurs. The International Cable Protection Committee (ICPC) puts the average cost of a cable repair between $1 million and $3 million. To detect and pinpoint the disruption, crews on land run various tests and analyses to diagnose the break, supported by ROVs to find the exact spot to begin repair.

But before a new cable can be laid, the old one is extricated from the ocean using a grapnel — essentially a grappling hook — that is positioned and dragged along the cable until it hooks and lifts it out of the water. Think of it as a higher-stakes version of an arcade claw game. The damaged portion of the cable is repaired, spliced, reattached, and sent back to its home under the sea.

Cable repair capacity remains a chokepoint in the market and life cycle for submarine cables. According to the ICPC, there are only 63 operational ships capable of laying or replacing cables worldwide. Between 2011 and 2020, only five new ships with this ability were constructed globally. The US government controls one cable repair ship, the USNS Zeus, but has contractual relationships with two more, and access to nine total, thanks to commercial entities based in America or controlled by US entities. China’s shipbuilding capacity dwarfs that of the US by 23,000 percent, measured by annual vessel output, but the country currently controls only six ships capable of laying and repairing cables. If that asymmetry translates into more Chinese ships at sea, the US could find itself dependent on its chief competitor to maintain essential infrastructure. This geopolitical asymmetry is built into the world of submarine cables, connecting the whole world but controlled only by the countries willing to invest in the infrastructure to lay them down and protect them.

Cables and Empire

It is tempting to tell this story as a triumph of private ingenuity — visionary businessmen such as Samuel Morse who bullied the future into existence. Those figures appear in this story, but, from the beginning, submarine cables were entangled with the state. They required diplomatic agreements, naval assistance, public subsidies, imperial logistics, and the conviction that instantaneous communication was worth a ridiculous amount of trouble. And trouble was what they got.

The first attempts at laying a transatlantic submarine cable did not go well. Cyrus W. Field established the Atlantic Telegraph Company in 1856 and promptly convinced the British and American governments to back the project. In exchange for free use of the line, the two governments agreed to provide an annual subsidy for the company and to provide the ships needed to lay the cable. When the USS Niagara and HMS Agamemnon set out carrying 2,500 tons of cable to connect Newfoundland to Ireland, the comedy of errors had already begun.

For one thing, the cable itself was both poorly designed and hurriedly manufactured. After only 350 miles had been laid, it broke and fell to the bottom of the sea. After months spent securing a new cable, the fleet set out again only to weather a ferocious storm before the cable snapped three more times, forcing them back to Ireland to reprovision. When the cable finally connected Newfoundland to Valentia Bay in 1858, it was met with enormous fanfare. Queen Victoria sent the first transatlantic telegraph message to President James Buchanan, and the Times compared it to the discovery of the New World. Less than a month later, the cable stopped working

Operators at the GPO’s Central Telegraph Office in London c. 1898

The man put in charge of the project, Dr. Edward Whitehouse, was sorely out of his depth. Whitehouse’s decision to use high-voltage induction to conduct energy within the cable had burned through the cable’s insulation, rendering it useless. Field and his Atlantic Telegraph Company promptly sacked Whitehouse in favor of an Irish professor of natural philosophy, William Thomson, better known to history as Lord Kelvin. After commissioning the largest ship afloat, reengineering the cable, and enduring a few more failed attempts, Atlantic Telegraph successfully laid the first functional submarine cable in 1866.

From then on, the operations took on a distinctly imperial character. Governments subsidized the firms to build them, navies protected them, post offices administered them, and imperial planners obsessed over their routes. Great Britain recognized the political and strategic value of cables. A telegraph cable, the earliest iteration of submarine cables, laid on the seabed, could bind colonies to the metropole more tightly than a governor’s speech or a frigate in the harbor ever could. By 1902, the British had completed the “All-Red Line,” a cable network stitched across imperial possessions so that messages could travel around the world while touching as little foreign territory as possible, denying colonial competitors any leverage over information flows. It was communications policy as statecraft.

Steam-powered empires ruled by telegraph. For nearly one hundred years after the British East India Company took control of India, dispatches from London to the subcontinent took around 10 weeks round trip, leaving senior British officials with a great deal of practical autonomy. As a result, when the Revolt of 1857 broke out north of Delhi, Governor General Charles Canning had to gather reinforcements, redirect troops (including a detachment of British regulars on their way to China), and manage the revolt before metropolitan oversight could meaningfully catch up. While some, including close associates, questioned whether Canning was up to the task, his deft handling of the revolt earned him the moniker “Clemency Canning” and the faith of both Parliament and the Crown. When ordered to preside over the reorganization of British rule in India, Canning was guided by the Indian Councils Act of 1861 and orders from Whitehouse, but was given plenty of room to act first and explain later.

That model of governance, which encouraged independent action by prudent, autonomous leaders, did not survive the arrival of the telegraph. Britain completed several overland and submarine telegraph cables connecting London and Calcutta between 1868 and 1870. The Suez Canal was completed around the same time, which allowed for greater commercial operations and communications between London and its colonial jewel. The combination dramatically shortened the feedback loop of colonial governance.

By the outbreak of the Second Afghan War in 1878, India’s frontier policy was orchestrated and conducted through rapid telegrams. Parliamentary records from this period reflect a very different tenor in the metropole’s oversight than during the Revolt of 1857. For example, in one exchange dated December 6, 1878, less than two weeks after the British invasion of Afghanistan, Lord Robert Montagu questioned the Chancellor of the Exchequer about a relatively minor discrepancy between the telegram dispatches — wired via submarine cables — of Lord Lytton, Major Cavagnari, and Sir Neville Chamberlain (not that one).

The cable solved coordination problems but quickly introduced a new vulnerability for statesmen and industrialists. A cable is miraculous right up until the moment it breaks. Then it becomes a very long, very expensive piece of damp string. For example, the 1859 failure of a cable built through the Red Sea and Arabian Sea to connect London and Karachi left both Whitehouse and Lord Canning bewildered amid the chaos of governmental reorganization. The cable fell victim to the rough environment within the Red Sea and was subsequently abandoned in favor of alternative over-land routes in the short run, and better engineering over the long-term.

The romance of the cable age was always shadowed by the less glamorous business of maintenance. While we remember the funders and boosters, the true heroes of submarine communications are the cable ships and their crews. Deployment and maintenance required charts, depots, shore stations, legal protections, trained crews, and standing arrangements for repair. In other words, they required institutions, a vast and unglamorous apparatus that made the miracle possible.

The state was the guarantor of the cables. Even when nominal ownership was mixed or private, the world’s submarine cables depended on government guarantees, government ships, government monopolies, and the kind of bureaucratic patience that can afford to think in decades. You could not build a global cable system the way one builds a fashionable app — licensing, laying rights, and the sheer upfront costs of construction and operation demanded something more patient than venture capital. You needed a navy. By the mid-twentieth century, the technology changed, but the political economy remained stable, and the role of the state persisted in planning, funding, and operating the infrastructure. The first transatlantic telephone cable, TAT-1, entered service in 1956 as a joint venture among New York-based AT&T, the British General Post Office, and the Canadian Overseas Telecommunications Corporation. TAT-1 inaugurated the modern era of reliable undersea voice communications, using coaxial cable, submerged repeaters, and an engineering standard that set the benchmark for the industry.

The remarkable thing is how long this clubby order endured. For decades, submarine cables were the domain of regulated monopolies and state-linked carriers, usually justified in the language of national interest rather than entrepreneurial disruption. Even as the internet took hold at the end of the 20th century, the network beneath the waves retained this old-regime flavor. The early internet, after all, began as a government project. The digital future was born, once again, in a world of public money, research institutions, and strategic priorities. The frontier mythology came later.

Long before Silicon Valley began speaking in the airy language of frictionless connection and information freedom, there were men trying to fish a broken empire out of the rough waters of the Atlantic. The submarine cable has always been both modern and archaic in this way: a machine for annihilating distance and time that is dependent on ships, sailors, steel, and states to keep the messages flowing.

The New Lords of the Seabed

By the 1990s, the old cable order was beginning to dissolve. The age of state-backed monopolies and national carriers had been broken up, privatized, deregulated, and been thrown into a storm of free(r) markets. Even some of the cables that were built and maintained by the USSR were absorbed by early telecommunications firms in the Russian Federation and upgraded with the help of Western technology and expertise. In the final decade of the century, cables increasingly were laid as privatized, speculative bets on a digital future that appeared to have no ceiling. The new cables were often financed by private consortia made up of telecom firms, carriers, investors, and, later, technology companies whose appetites for bandwidth would have seemed insane to an earlier generation. The state remained in the picture, naturally, but ownership and initiative were shifting. The modern network of cables would be built more like a syndicate than a ministry. Thus the internet, still young enough to feel slightly unserious, arrived at exactly the moment when the legal and financial architecture of global communications was becoming less imperial, less bureaucratic, and far more competitive.

The commercial reality was irresistible. Fiber-optic technology — pioneered by researchers in the UK and the Netherlands and made commercially viable by Robert Maurer, Donald Keck, and Peter Schultz at Corning Glass in New York — made undersea transmission vastly more powerful as internet traffic became a category of traffic unto itself. In the US, partnership between Corning and Bell Labs spurred iterative development of more advanced fiber cores, expanding capacity and reducing signal loss. New routes were laid, landing stations multiplied, and capacity exploded.

There was, for a while, something almost euphoric about this digital gold rush. The 1980s and 1990s produced one of those familiar modern spectacles in which financiers, engineers, and techno-evangelists all convince one another that this time demand really is infinite, and that information itself was the commodity. The dot-com boom led to a frenzy of traffic and cable construction. When the bubble burst, major cable builders and network operators went bankrupt, were forced to reorganize their firms, and in some cases, sell their recently-built networks to other operators. While many companies failed, the cables survived.

But the political constituency of the system changed. The old monopolies gave way to consortia, and the consortia eventually had to make room for a new class of actors altogether: the hyperscalers. As Amazon, Google, Meta, Microsoft, and Oracle expanded into cloud computing, global content delivery, and planetary-scale data storage, they started behaving more like infrastructure owners than like customers of ISPs. At first they bought capacity on other people’s lines. Then they bought more. Then, increasingly, they backed or financed entire systems themselves.

TeleGeography reported in 2025 that aggregate new-construction costs had averaged more than $2 billion annually over the previous nine years, and forecast more than $13 billion in cable investment for systems expected to come online between 2025 and 2027. In February 2025, Meta announced Project Waterworth, a multi-billion-dollar, multi-year system spanning five continents. Google, meanwhile, recently announced three major projects, including America-India Connect, anchored by Google’s five-year, $15 billion AI infrastructure investment in India. In the Americas, the MANTA consortium, composed of private entities, launched in March 2025 to connect Mexico, the United States, Central America, and Latin America with new low-latency routes.

This is one of the strange reversals of our time. We are accustomed to thinking of the internet as a force that dissolved old concentrations of power. In practice, its physical backbone has often moved in the opposite direction. What emerged was not the nationalized order of the cable empires that cyberlibertarians like John Perry Barlow called the “information railroad,” nor the frictionless commons the cypherpunks envisioned. It was a world in which a handful of very large firms, operating alongside carriers and consortia, came to possess extraordinary influence over the routes, capacity, and resilience of global communication. Some cables are wholly owned by a single hyperscaler — Google’s Dunant system, for example, connects the US to France. Others follow a consortium model, like MAREA, backed by Meta and Microsoft, where spare capacity is sold to third-party providers. The network became more distributed in some ways and more concentrated in others — the sort of paradox modern infrastructure tends to embody.

And because the complex system of submarine cables had become easy to ignore to their average beneficiary, its vulnerabilities acquired a new theatrical quality whenever they broke the surface. Thanks to redundancy, a severed cable no longer means the total communications blackout that a Victorian official might have feared. But local shocks can still feel unnervingly medieval. When cables connecting the Channel Islands were accidentally cut in 2016, banking, phone service, and ordinary online life all suddenly shut down — a sharp reminder that islands remain islands, however digital their economies may be.

The new crop of financiers are indeed private entities, but it would be wrong to assume this means that such infrastructure is now bereft of government influence. For any cable that lands on US soil, the owner or consortium must submit documentation to the Federal Communications Commission to comply with the Cable Landing Licensing Act of 1921. The current licensing process requires cable financiers and supporters to submit documentation to the FCC and “Team Telecom” — a cross-government team composed of national security officials to evaluate a project’s funding sources, cable path, environmental permits, equipment components and sources, among other criteria — to ensure cables cannot be used as vectors for foreign influence or sabotage. Financing, planning, and operations are increasingly privately driven, but not without the state’s permission.

Fault Lines

The most obvious problem besetting the world’s digital arteries is that everything is getting bigger. The traffic volumes are bigger, the data centers are bigger, the systems are bigger, the expectations of impact are bigger. That growth is not merely quantitative; it changes the engineering problem itself. Each new system is sold as faster, denser, more efficient, and more indispensable.

There is no easy “build more cables” button that will catch up with the growth. Behind every cable is a chain of dependencies: permits, landing stations, terrestrial backhaul, power equipment, marine surveys, maintenance contracts, spare parts, legal access, political consent. The challenge is building enough capacity without creating a world in which every increment of scale also widens the blast radius when something goes wrong.

Then there is the more unsettling problem: some breaks are not accidents.

Consider the Red Sea. Amid the disorder of the Houthi missiles in the Red Sea, it is one of the places where the submarine cable system is forced to reveal how much of its traffic still passes through narrow gates. What the Suez Canal is for shipping, the Gulf of Aden, Strait of Mandeb, and Red Sea corridor are for internet traffic between Europe and Asia: a chokepoint through which Europe, Asia, and the Gulf remain tied together by a bundle of vulnerable lines.

On February 24, 2024, Asia Africa Europe-1, Europe India Gateway, and SEACOM/Tata TGN-Eurasia suffered faults in the Red Sea. The disruptions were initially linked to the Houthi rebels, who earlier that year had circulated maps and images of submarine cables on Telegram channels. Investigators determined that the Houthis did not cut the cables directly; rather, the cables were damaged by the anchor of the Rubymar cargo ship, which was struck by Houthi missiles. The Rubymar subsequently dragged its anchor across the sea floor before sinking. After the disruption, the three cables were not repaired until July, five months later, with commentators suggesting repair capacity was constrained by the threat of more attacks.

That vulnerability is physical before it is geopolitical. Submarine cables follow the earth as it is, and, more often than not, the best route is also the one most exposed to politics, violence, fishing gear, landslides, and simple bad luck. In the Red Sea, those hazards do not sit neatly apart from one another. That is what made the Houthi-linked disruptions in 2024 so unnerving. Cable damage is one problem, but more importantly, the attacks tested the modern internet’s much-advertised redundancy in one of the least convenient places on earth. The global network could, and largely did, reroute around the trouble. But the episode was a vivid reminder that resilience, the ability for infrastructure to withstand stress, is not the same thing as invulnerability. Redundancy buys time and flexibility. But if enough cables in a narrow corridor are threatened at once, the problem stops looking like an isolated repair job and starts looking like a stress test for the architecture of globalization itself.

And so the geopolitics come back to the conversation. Governments once again began speaking openly about cable security, foreign ownership, surveillance risk, landing rights, and strategic dependence. Security services and legislators rediscovered an interest in who builds repeaters, who operates repair ships, and whose territory hosts key interchanges for data and power. The 21st century, built upon the optimism of the early internet and buoyed by the idea that information should be free, was forced to admit that information has a cost. It also requires a route, a permit, an insurer, a repair crew, and political certainty. Even now, as oil tankers and cruise ships become blockade runners in the Strait of Hormuz, a whole industry of IT and security professionals is monitoring the situation on the seafloor of the Persian Gulf.

This, too, is not entirely new. On August 4, 1914, immediately after Britain transmitted its declaration of war to Kaiser Wilhelm II, His Majesty’s Telegraph Ship (HMTS) Alert was sent into the English Channel to cut German telegraph cables on the seabed. This, one of Britain’s first acts of the war, was intended to force the German military onto radio and other communications systems that Britain could monitor more easily. The result was an unmitigated success, opening the door for the British Admiralty’s codebreaking unit, Room 40, to decrypt and read German naval messages. In fact, the United States’ entry into World War I can be tied directly to the work of the HMTS Alert on that summer’s eve.

When German State Secretary for Foreign Affairs Arthur Zimmermann sent his now-famous telegram in January 1917 proposing that Mexico join Germany in the event of war with the United States, Berlin’s direct links across the Atlantic had already been severed for nearly three years. As a result, the telegram traveled via Washington DC and was relayed onward by the German ambassador there — exactly the kind of detour that gave Britain an opening. Having already tapped many of these lines for surveillance, the British Admiralty easily copied the message and relayed it to Room 40 for decryption while the US Department of State maintained deniability and passed the message along to Mexico as requested. The real brilliance came afterward: Britain had to show the telegram to the Americans without revealing that it was reading diplomatic traffic and cracking German codes, so it masked the true source and let the message emerge under a carefully managed cover story. The result was one of the great intelligence coups of the war.

Perhaps more importantly, Britain’s deep understanding of the strategic significance of submarine cables and codebreaking during World War I set the stage for Bletchley Park and Alan Turing’s team during World War II. Nations understood perfectly well in the telegraph and telephone era that cutting cables could isolate an opponent, degrade communications, and divert traffic onto more favorable routes. What has changed is how appealing such action has become. In an age of gray-zone coercion, deniable sabotage, increasing technical sophistication, and infrastructural pressure below the threshold of open war, submarine cables are temptingly vulnerable. For an adversary, that is an attractive combination.

This leads to the least glamorous and perhaps most important challenge of all: cable repair. The process is costly, time-consuming, and constrained by the limited number of ships that can do the job. This creates a bottleneck with serious downstream consequences: reduced capacity, impeded digital communication, disrupted commercial activity, and sensitive traffic exposed to insecure infrastructure. If several disruptions occur at once, or if a major corridor is hit repeatedly, or if conflict turns one region into a hazardous operating environment, repair starts looking like triage. A small fleet means waiting. Waiting means economic cost, degraded service, nervous governments, and growing uncertainty about what comes next. As the cost of cable disruption — intentional or otherwise — continues to rise, so do opportunities for companies developing sensors and autonomous undersea vehicles to identify threats before they cause disruption or mitigate damage before it becomes physical.

Then politics takes the baton from engineering. A ship and her crew may be ready, but the repair can still stall because access is denied or the operating area becomes too dangerous to enter. It is much easier to damage a cable than to repair one. Which means the cloud still depends, rather embarrassingly, on whether the right ship can get to the right patch of water at the right time with the right permissions.

And yet the better lesson may be the harder one: the system of submarine cables has always survived despite the difficult world it serves. Submarine cables have endured wars, storms, earthquakes, bubbles, busts, and the constant indignity of the sea itself. Their survival has rested on redundancy, improvisation, and the incredible competence of the people who fix things when they break.

The Future in the Deep

It would be easy, after spending so much time with cable cuts, chokepoints, repair ships, and the general malice of geography, to end on a note of anxious realism. But that would miss the larger point. The future toward which submarine cables are carrying us is one of greater economic depth, broader technical possibility, and, if we are wise about it, a more open and resilient information environment. The best thing about this infrastructure is that, when expanded intelligently, it enlarges what society can become. Greater dependency drives demand for greater redundancy — which, in turn, drives interest in defending the digital arteries from clogs and cuts.

This is already visible on the balance sheet. More cables mean more capacity, but capacity is only the beginning. Research from the World Bank has found that when a submarine cable is laid or capacity is expanded, it is associated with a statistically significant decline in internet prices, between 14 and 21 percent per doubling of cable capacity. The World Bank also has found that cable expansion raises the probability of receiving services-sector foreign direct investment. Research published in the American Economic Review surveyed the effect of staggered deployment of submarine cables and additional terrestrial fiber backbone across twelve African nations, finding employment gains ranging from 3.1 to 13.2 percent across the study. It also found that nighttime light density increases 2.4 percent, a proxy for economic activity. The IMF similarly linked submarine scale expansion to increases in total-factor productivity and real per-capita GDP growth. Submarine cables are catalysts for economic growth, no matter where they land.

That matters because the next wave of value will come from enabling the next generation of digital services — globally distributed AI workloads, real-time industrial control across borders, more sophisticated scientific collaboration, and planetary-scale applications — that are only possible with dense, resilient, and high-capacity international networks. Google’s recent cable announcements are striking in part because they say this plainly: the new routes are constructed to support AI and computationally-heavy services that demand a robust physical architecture that spans the world.

This is where the politics of cables becomes especially interesting. In the cable world, more is simply better. More cables mean more routes. More routes mean more diversity. More diversity means fewer brittle chokepoints, less dependence on any one corridor, and greater room to reroute traffic when something fails or someone tries to make it fail. The OECD’s recent work on communications resilience is admirably blunt on this point: “Resilient communication networks are built on the principles of redundancy and diversity.”

A world with more routes and more operators is not a perfectly free world — the US, for example, would still be dependent upon a limited fleet of repair ships and cable networks running through hardware built or maintained by foreign adversaries. Still, it is plainly better than a world in which speech, commerce, and knowledge are forced through a few narrow maritime gates and a handful of vulnerable landing points. If liberal societies care about an open internet, they should care about route diversity with the same seriousness that earlier statesmen cared about sea lanes and coaling stations.

The ongoing cable boom is, at its core, a strategic public good being financed through a messy public-private mix (though, of course, it does benefit the hyperscalers and telecom firms). The government’s job is the unglamorous part: streamlining permits, protecting landing sites, avoiding self-defeating regulatory delays, expanding repair capacity, and supporting diverse routes. Cable security ought to be thought of as economic policy, not a national-security afterthought.

Governments can also follow Chile’s lead with Google, or the European Commission’s more recent example: acting as partners and co-investors when the strategic case for resilience exceeds commercial logic alone. In early 2026, the European Commission amended its digital infrastructure program to allocate €347 million for strategic cable projects, repair capacity, and “smart” cable systems, while the International Telecommunications Union and International Cable Protection Committee have similarly turned cable resilience into a sustained multilateral agenda, encouraging greater international collaboration on cable protection, route diversity, and investment.

There is something genuinely wondrous in the steady expansion of routes, financiers, and the services such infrastructure brings. The ocean, which once divided markets and empires, is becoming ever more densely sewn into the fabric of a common digital life.

If the submarine cable is the hidden architecture of our age, then the task ahead is to build more of it with confidence. Which is why submarine cables retain their peculiar capacity to inspire awe. Every age gets the sublime it deserves. Ours is not only in rockets, reactors, and retatrutide. It is also down there in the dark, where a network of glass threads holds together a planet that prefers not to think about how much it depends on them. The miracle under the sea is not finished. It is still being laid.

The story that follows is about the collision of two kinds of exponential progress. One is rooted in atoms, chemistry, and the science of advanced manufacturing. The other is driven by data, statistics, and algorithms. Each advanced along its own trajectory, yet their convergence produced capabilities that neither could have unlocked alone. The modern AI boom is best understood by viewing both forms of scaling as parts of one continuous technological evolution.

The first type of scaling, compute, is based on a simple on/off switch, the transistor. Transistors are the fundamental building blocks of digital logic. They use an electrical input signal to determine if current can flow through the transistor or not. This “on” or “off” current flow can store a binary state, a simple 0 or 1. Basic primitives — logic gates — can be constructed by wiring transistors together. Connecting two in series, with the output of one flowing to the input of another, forms a logical “AND” — the output is only a 1 if both inputs are 1. Similarly, a logical “NAND” — which stands for ‘not and’ — can be formed with 4 transistors wired in a complex pattern.

Amazingly, any digital circuit, from the simplest binary adder to the most complex AI accelerator, can be constructed with NAND gates. It’s an elegant and beautiful result of information theory: the NAND gate itself is universal and functionally complete as it can express every possible boolean function. Simple functions are intuitive: a logical AND — output 1 if and only if both inputs are 1 — can be formed with two NAND gates in series, the output of the first used as both inputs to the second. Logical OR can be produced with two ANDs and another NAND, for a total of five NANDs. And so on for other basic logic operations, NOT, NOR, XOR, and more.

Combining many simple logic operations in a circuit forms logic blocks, with simple functions like adding and subtracting. Logic blocks in turn can be combined into simple subsystems that load and store instructions. After that, microarchitectures, instruction-set architectures, and further until, after ten or so layers of abstraction, you are at the application level: what users see. This is a web browser, word processor — or now an AI performing humanlike tasks. But at the bottom of these systems are just simple on/off switches, wired together in the right way. There are billions of NAND gates in a chip the size of your thumbnail.

A long series of innovations were required to reach the point where billions of switches could be packed on a chip. Early computers used vacuum tubes which were fragile and power-hungry. Vacuum tubes relied on heated cathodes in a vacuum to emit electrons towards anodes, a mechanism that is slow, inefficient, and crude by modern standards. The transistor, invented in 1947, replaced clunky vacuum tubes with a solid piece of silicon that could switch reliably and efficiently. Rather than waiting for a cathode to heat up, in a transistor an electric field opens a “channel” for current to flow inside silicon. And instead of an airtight glass tube, a transistor needs just a literal sliver of silicon — which we now can manipulate at the nanometer scale. Transistors use far less power, can be made smaller, and operate at far higher speeds than vacuum tubes, switching on and off more than a billion times every second.

A decade after the invention of the transistor came the integrated circuit, which put multiple transistors on a single chip to form logic gates (like NAND). Soon after logic gates came logic blocks that could perform simple math or route digital signals , then microarchitectures executing instructions and interfacing with memory, and so on. And within twenty years of the creation of the first silicon transistor, simple operating systems were running applications on integrated circuits of 100,000 transistors.

Modern chips begin as highly purified silicon. Manufacturers convert silicon dioxide into material that is almost entirely free of defects, then melt it and form it into a single crystal cylinder called an “ingot.” The crystal is then sliced into wafers and polished until defects on the surface of these wafers are measured in fractions of an atom. This level of flatness is not cosmetic; even small imperfections would distort later steps and cause device failures, so the starting surface must be close to perfect.

Fabrication proceeds layer by layer atop the bare wafer. Transistors and wires are built through repeated cycles of material deposition, lithography, etching, and doping. Lithography, a process that uses light to map circuit patterns onto a silicon wafer, transfers the chip design — a series of simple lines, spaces, and circles that will be transformed into transistors, wires, and vertical connections from layer to layer. Using the pattern defined via lithography, other processes adjust the electrical behavior of the silicon itself. Ion implantation inserts controlled amounts of elements like boron or phosphorus to create regions where current can or cannot flow. Thin film deposition forms conductive wires and insulating layers only a few atoms thick. Chemical-mechanical polishing removes bumps, ensuring a pristine, atomically flat surface for the next layer. A leading-edge chip requires thousands of these steps, each executed with nanometer-scale precision.

After completing all layers — as many as one hundred in modern chips — the wafer is tested. Functional chips are cut out from the wafer, bonded to tiny solder bumps, and packaged so they can be connected to circuit boards. Although the end product may be small, construction of modern chips pushes physics, chemistry, and manufacturing equipment to their limits. The fact that these steps can be repeated millions of times per day with just one or two defective transistors in a billion is one of the most impressive achievements in modern industry.

Training a neural network is largely a matter of multiplying and summing vast tables of numbers. Although the results look almost like magic, this is just statistical pattern recognition at scale. Artificial intelligence took off once hardware could deliver enough raw computation. What made today’s AI possible was the ability to place enormous numbers of compute units side by side, close to memory, and to keep them all running in parallel. As chip capabilities grew, researchers built models large enough to capture language and vision. As models demanded more, the hardware advanced again. This feedback loop spurred the present AI boom.

Scaling laws add a second exponential to the story. Moore’s law concerned the number of transistors engineers could pack onto a chip. Scaling laws in machine learning describe how model performance improves as you increase three quantities: the size of the model, the amount of data it is trained on, and the compute used during training. In 2020, researchers studying large language models noticed something striking: when these three variables rose concurrently, performance improved following smooth power law curves. The models did not improve in fits and starts. And they did not level off unexpectedly. Instead, they followed clean mathematical relationships that predicted how much better the next, larger system would be. If you doubled the compute and data in the right ratio, you could predict the reduction in error before doing the training run.

The parallel between Moore’s law and scaling laws is not superficial. Both describe a world where steady, compounding improvements produce outcomes that seem impossible when viewed from the starting point. Moore’s exponential made processors a million times more capable over five decades. Scaling laws did something similar for AI, except the exponent has played out over a handful of years. Early language models were clumsy pattern matchers that struggled to keep track of long sentences. Their training datasets were a fraction of today’s, and their parameter counts were measured in millions. By the time ChatGPT arrived in November 2022, models had grown to hundreds of billions of parameters. Training runs consumed millions of GPU hours. The compute used just to train one model was larger than the total used to train every neural network in the world just a decade prior. Yet instead of chaos or diminishing returns, researchers found that bigger models almost always performed better, and that the gains were predictable.

There is an important difference between Moore’s law and scaling laws, though. The former was partly a roadmap. The semiconductor industry organized itself around Moore’s law, coordinating investments so that each generation of chips arrived roughly on schedule. But scaling laws in AI were discovered after the fact. They described what researchers were already stumbling into as they pushed models larger. Once recognized, they became a guide. Labs now forecast the performance of future models using these laws, estimate the compute budgets needed to reach certain capabilities, and design data collection efforts accordingly — the same as Moore’s law used to guide semiconductors. The laws do not guarantee success. But they do offer a rare quantitative framework that brings structure to a field long guided by intuition and luck.

Understanding why scaling laws work requires a brief look at what a large language model actually does. The model learns to predict the next word in a sequence by adjusting internal weights so that its predictions better match the patterns in its training data. Those weights give the model a sort of high dimensional memory of language structure. As models get larger, they can “memorize” and represent more subtle patterns. With more data, they can generalize across a greater span of contexts. With more compute, they can train on that data with a model size large enough to encode all of it. Scaling laws describe the quantitative relationship among these factors. Crucially, they show that models remain “underfitted” at every size we have tried. In plain terms, we have not yet reached a point where adding more data or compute ceases to help. Moore’s law eventually slowed when transistors ran into atomic limits in the 2010s. Scaling laws have yet to hit a similar plateau.

Remarkably, once models grew past certain scaling thresholds, their behavior improved in ways that went beyond steady reductions in prediction error. They began to show abilities that had never been programmed into them at all. Emergent abilities. Tasks like multi-step reasoning, translation between obscure language pairs, and writing coherent code, that were not explicitly programmed in new models. They just surfaced when models grew large enough to internalize the structure of these problems from data alone. These phase transitions echo the qualitative jumps seen in computing during the early years of Moore’s law — when added transistor budgets made entirely new classes of applications feasible. Just as the shift from thousands of transistors to millions enabled graphical interfaces and early networking, the leap from billions of parameters to hundreds of billions enabled systems that interact with humans in fluent language.

With the right architecture and training process, relentless scaling can carry a technology into regimes that feel qualitatively different from what came before. The semiconductor industry proved this over half a century of squeezing more transistors into less space. Machine learning is now showing the same pattern of relentless scaling in compressed form.

The foundation for all of this progress, the machines that “think," is itself extraordinary. Modern chips pack 100 billion transistors into the area of a thumbnail, each one patterned with atomic precision. A single misplaced atom can cause a defect, yet a finished chip typically has only a handful. Factories repeat this feat across wafer after wafer, every day, yielding hundreds of millions of nearly perfect devices each year. And this reliability and scale are what make the relentless expansion of AI models possible. The models might look miraculous when they cross new capability thresholds, but they depend on the second miracle of modern chip manufacturing. Taken together, these two forms of scaling have created systems whose behavior and reach would have been difficult to imagine only a decade ago. Only time will tell what the next decade will bring.

Anduril was founded on the assumption that the American defense model, with its guild of defense primes producing small numbers of exquisite, expensive, slowly built weapons platforms, would have to be fundamentally redesigned. Instead, Anduril would produce low-cost, autonomous, mass-produced systems in large numbers. Less than a decade later, as wars in Ukraine and the Persian Gulf throw the old assumptions of warfare into disarray, that wager looks prophetic. In May, Anduril closed a $5 billion Series H at a $61 billion valuation. The company more than doubled its revenue in 2025 to $2.2 billion. At Arsenal-1, its 5 million-square-foot manufacturing facility in Ohio, it is standing up high-rate production, transforming from technology startup into a bona fide defense production company. In time, Anduril plans to build similar facilities in allied countries worldwide.

Trae Stephens is Anduril’s co-founder and executive chairman, and a partner at Founders Fund. His path to the commanding heights of defense technology was an unlikely one. Raised in Ohio, Stephens spent part of the eighth grade in Gaza where he developed a fascination with the Middle East. After 9/11, which occurred during his senior year of high school, he abandoned ambitions in journalism and refocused his efforts toward a career in national service.

After studying Arabic and computational linguistics in college, Stephens’ first job was in intelligence. From there he joined Palantir Technologies as an early employee, then moved to Founders Fund, where he has built an investment practice around companies at the intersection of technology and national security like Varda Space Industries and Nominal. He has served as Anduril’s executive chairman since its founding. Stephens is one of the defense tech ecosystem’s most influential voices on the widening gap between American production capacity and the demands of modern war — a gap laid bare by the recent war with Iran, in which the United States burned through years of Tomahawk production, along with other crucial stockpiles, in just a few weeks.

I sat down with Stephens in San Francisco to better understand the lessons of modern war, the strategic challenge posed by China, and the future of the American defense industry. What follows is a transcript of our conversation.

CB: Have you read anything particularly valuable recently? I know you're a bit of a writer yourself.

TS: I would not claim that. Mike Solana is a brilliant writer. I’m just hanging on for dear life. The last book I read was about the San Francisco Presidio, called Bastion by the Bay — a cool history. Up until 1994, this was a continuously operated military base under three different countries: it was a Spanish military base, then a Mexican military base, and then a US military base. It was a cool story. I also read a lot of science fiction, which I think is helpful for venture capital. And then, obviously, all sorts of books concerning the history of defense.

CB: In your interviews you describe growing up in Ohio. One question I had from hearing your account was: How did you get so interested in the Middle East in particular, coming from Ohio? You ended up studying Arabic, and also Farsi?

TS: I did a bunch of computational linguistics stuff on Arabic script-based languages, which includes Farsi, things like that, but I was studying Arabic specifically.

When I was in eighth grade, my family went to the Gaza Strip on a service trip. My dad was doing some construction work and my mom was volunteering at a school for the deaf. I had a burgeoning interest in becoming a journalist doing international correspondence, and this really fueled that even more. I became the editor of my high school newspaper. Then, 9/11 happened my senior year, and so I pivoted from wanting to do foreign policy, geopolitical-type journalism to doing something in service to the country. I went down that path in college, and then went into the intelligence community from there.

CB: I’ve read a lot about the British Empire, their civil service, the Indian Civil Service, and so on. It was among the most prestigious jobs you could have. An intensive exam process. It drew the cognitive elite into that field. From your time in intelligence, did you get the impression that these American institutions were taking advantage of the brilliant minds we have here?

TS: No, definitely not. I mean, there were moments, like during the Cold War. I think we did a great job of attracting the right people then. If you look at the history of tech, most of the cutting-edge tech was coming out of research funding for defense, whether that was radar for commercial aviation, or GPS, or the internet. All of these were defense projects. The long history of Stanford's rise to the top of academia for science-related fields — that was also funded by the Defense Department. 

I think people understood the reality of the existential threat we were facing vis-a-vis the Soviet Union, and they were willing to spend their time and effort contributing to national service relevant to that potential conflict, or that related at least to the existential risk of that potential conflict. Once the Cold War ended, I feel like a lot of that just faded into oblivion. People were much more likely to graduate and go and work on Wall Street, or become lawyers.

CB: The more lucrative professions.

TS: Yeah, they’d go into tech, work on the early days of the internet — and who could blame them? There was this very confident feeling that we would have an impenetrable hegemony. The Soviet Union was dead. We had a global monopoly on power. It felt very safe. I remember living through the ‘90s and thinking that our biggest problem was whether or not the president had an affair with an intern. It seemed like there was nothing happening that would have led people to go and work in national security. 

I think the problem is that the mystique behind it remained, so the culture still believed that these were these incredibly high-powered, intellectual, aspirational jobs — media and entertainment played into that with popularity of the Bond movies and the Bourne movies — but the talent pipeline dried up. It's further than ever from being recovered, I would argue. So, I think this is a real problem that we haven't really reckoned with as a country yet.

CB: To touch on some of the themes that you mentioned in your Hill and Valley speech back in March: There are certain historical cases where the US was forced to respond to a technological or industrial challenge. A shock moment. In World War II, it led to us creating this massive assembly line where we could make 286,000 airplanes. A few years later, you had the Sputnik moment, which drove Silicon Valley’s development to a great extent. Do you think that we're on the verge of some kind of shock?

TS: It's probably less of a shock and more of a 30-year ignorance compounding into a major problem. This isn't even a partisan thing. Democrats and Republicans alike agreed that globalization was good. Offshoring to low-cost labor sources was totally acceptable. The rising tide was going to lift all boats. Everyone would be better off. Consumer prices went down, surely, but the belief was that economic growth was going to democratize potential international adversaries. 

I was in Paris in 2019 — right before Covid — and there was a debate at that time about Western Europe's reliance on Huawei for 5G infrastructure. Macron gave a speech saying, in effect: “I think people are overreacting to the Huawei conversation. Our close partnership with China is undoubtedly going to lead to better justice and outcomes for democracy to the Chinese people.” That was in 2019. That's not even that long ago. Now we look back on these twilight-zone moments from as recently as five or six years ago and we think: how could we have possibly believed that that was true as recently as 2019? I think there was a 30-year window where people thought, ‘this is just the expectation of globalization,’ and the Chinese have sort of hoodwinked us into getting complacent about natural resources and supply chain risk and the need for skilled labor, the trades. We're sort of waking up to it, but I don't get a sense that it’s going to be a single flash moment. It's more like we're the frog that has been boiling.

CB: The timing of Hill and Valley — this was in early March — was an interesting time, right? You had the Iran war kicking off, and then the whole Pentagon-Anthropic saga kicking off around the same time. On the Anthropic issue, do you think that indicates some sort of latent cold war between Silicon Valley and DC?

TS: I don't think there's a cold war. I think it's more like there are misunderstandings on both sides. There's been such a fracturing of communication. Obviously it's very different from Stanford in the Cold War, right? Those were very close ties. I think today the tech community sort of roughly believes that it operates with relative impunity, and the government believes that they don't really need tech, and I don't think either of those stories is true. Obviously, Microsoft found this out in a very painful way 20-some years ago, when it was bundling Internet Explorer and forcing consumers down that path, and it got smacked pretty hard by the government. So, I think there just needs to be closer engagement in order for that trust to be restored. But the trust certainly doesn't exist right now. Michael Kratsios, the director of the Office of Science and Technology Policy, released a policy document on the interaction between the Department of War and the AI labs. The work is happening. I feel like the chasm is much smaller than it was even a couple months ago.

CB: On the Iran war, there have been some interesting developments. On the one hand, the performance of the American aerospace industry — ISR, targeting systems, all of that — has been pretty good. However, the war has also proven that what we currently have is not enough to achieve our strategic objectives. The UAE fired $10 billion worth of interceptors from a very limited stockpile, at extremely high cost. If you were to break down the core lessons of this war, what have we learned about our shortcomings and vulnerabilities as a power?

TS: I think what you just hinted at is the exact thing that everyone's waking up to. The contract for Patriot missiles was first awarded in the 1960s. They were tested in 1969. They're incredibly capable. They can be used to shoot down a DJI drone or a maneuverable hypersonic weapon, and everything in between. They were built to counter MiGs, to shoot down fighter planes, and we're using them to shoot at $60,000 drones, Shahed-136s. So the economic arbitrage piece of this is clearly not the right solution. We need to have a layered air defense portfolio. There need to be lower cost ways of engaging less exquisite systems. Do you still need a PAC-3, a modern-block Patriot missile to shoot down hypersonic targets? Sure, that sounds great. If you're going to shoot a $3 million missile, it better be at something that's highly capable. 

The bigger problem is manufacturability. Whether it's a Tomahawk or an SM-6 or a PAC-3, these are multimillion-dollar weapon systems that take years to build at very low volumes. Right now our annual production capacity for Patriots is something like 685. Tomahawks were a few 100 per year, and we shot about 1,000 Tomahawks in the first few weeks of Epic Fury. That's years of production, and the most recent orders that are coming in won't actually be supplied until 2030. We shot over 200 THAAD interceptors, over half of our inventory. We fired more than half of our Patriot interceptors. So when you burn through your inventory, the resupply on that inventory boxes out our international allies and partners because they get pushed to the bottom of the waiting list.

Even if we're doing resupply and we don't engage in any major conflicts over the next three years that would further burn down those very limited inventories, we've told Kuwait and Qatar and the Emirates and the Saudis: sorry, you're not getting any more Patriots. What does that mean for them? They have to make hard decisions. Do we go to China for munitions? If we do, does that mean that we need to tighten our partnership there? Does that put US interests at risk in the region? I would argue that it does. So we really need to fix the manufacturing and production side of the equation, in addition to increasing capability. Production is a huge shortfall right now.

CB: If you were to apply the lessons of Iran to the production and innovative capacity of China, you'd probably find yourself with a much more frightening scenario. Have you thought about how that would look? 

TS: Yeah. It's not only drones — I mean, they have fully autonomous factories making cruise missiles. They also perceive that there is risk in becoming a critical supplier to Iran, right? China is not a provoking nation. Have you read The Hundred-Year Marathon?

CB: I've not.

TS: It's a brilliant book. It's by this guy, Michael Pillsbury, who was at the CIA in China House for a long time — their China office — and he basically ties in the fables, the stories we were told as kids in different cultures. For Western fables, we have Aesop and Shakespeare and all this stuff, and a lot of the moral lessons that we've learned come from the Western canon, even if we don't realize that that's where they come from.

CB: We’ve internalized it from childhood.

TS: Yeah, we've internalized the Western canon, and that's often why we do the things that we do. The Chinese canon is not the Western canon — it's very different. The most popular fable in Chinese culture comes from the Warring States period. There’s this story about a super powerful king that was the presumed hegemon, and he had a vassal state whose king he would summon to come and meet with him. And every time, the king would say, "Tell me about your growth — what's happening economically, what's happening militarily in your kingdom,?” And the vassal would say, “Oh, we're so weak. It’s very difficult. We’re no match for you — you’re far superior.” He said that every year while he was building up his power, and then the moment at which he surpassed the other king was the first time he even suggested that he had the power to take control.

That’s not how the Western canon works at all. We're provoking, that's how we interact. It's like peacocking, almost. Whereas the China story is: every time you meet with them, they're gonna remind you how weak they are, there's no shame in it. Whereas in Western culture, there would be a lot of shame to saying, “we’re no match for you” — but actually, that's what 's been happening for decades.

CB: There was this interesting speech Hu Jintao gave in 2003 — the “Malacca Dilemma” speech. It was a significant departure, because he was overtly addressing a weakness that China had: how economically dependent they are on trade that passes through this one strait. I was thinking about the advancements we have in missiles and coastal defense. Given the geography, isn’t it possible, in theory, to constrain China simply through aiming those weapons systems at these choke points and preventing any trade or any naval vessels from going through?

TS: Trade is the only real lever we have against China. There's a question of whether they’re self-sustaining? Could they actually survive as an economy without the Western world buying everything from them, having them manufacture everything? There are definitely agricultural shortfalls, although they have vassal states in the region that could help them fill that gap. I think trade is the lever that we hold, but obviously they know that, and that's exactly what they will be working to shore up.

CB: Anduril just raised its Series H in May. What are you doing with that capital to increase production volume?

TS: We have a bunch of production facilities that are coming online right now. We have 16 different production facilities across the world — and I do mean the world. We have submarine production in Australia. There's a lot of stuff internationally that will be coming online in the next year as well. Arsenal-1 in Columbus is the bulk of our investment as we go from 800,000 square feet to 5,000,000 square feet over the next five or six years. 

You can't just do all of that capex investment at once, because you have to recruit a lot of people and train them up. So we're staging everything out in pieces. So this summer we're going to start rolling aircraft — Furys — off the line in Ohio. Right after that, going into the fall, we'll start producing Barracuda, our low-cost cruise missile, in Ohio. Then we’re moving Roadrunner production from headquarters in California over to Ohio, and we'll kind of be following along as we build that out. We’re also planning on building shipyards for expanding our work on undersea: We’re a very large player with the DoW currently on undersea production. It's really all focused on production: realigning the company away from being a tech startup to being a defense production company.

CB: I’m very interested in the undersea warfare aspect, especially since you mentioned Australia. I was following AUKUS quite a bit. How do Anduril’s products augment undersea warfare, and specifically anti-submarine warfare?

TS: The AUKUS agreement is a big deal for Australia. But deliveries of nuclear subs are a long way away: They'll receive their first subs around 2035. So I think they have to have a bridging capability, and the Royal Australian Navy was very interested in what we were doing with Dive-XL — the big one, the school-bus-size autonomous submarine. There are all sorts of payloads you put on something like that. We’re really the fairing that carries a variety of payloads. Some of those are sensing payloads — for looking for mines, looking for other submarines — and you can put kinetic payloads in it as well, torpedoes and stuff like that. We have a whole class of low-cost undersea payload delivery vehicles called Copperhead that align with that. And we have Seabed Sentry, which is basically a Sentry watchtower that you drop on the sea floor that can communicate with other Seabed Sentries and detect things that are happening in the water around it. So it's a big part of our strategy on the maritime side.

CB: What kind of sensing does it use? 

TS: It's a sensor fusion platform. Just like our Sentry towers on land, you can't do the mission with electro-optical sensors alone, or with IR alone, or with metamaterial radar alone. You want to tie all these things together, and then pull the signal from the noise by fusing those data streams together.

CB: Anduril has been developing the CCA-Fury platform, which began production earlier this year. Meanwhile the Pentagon has been talking a lot about the F-47 and sixth-gen. One of the priorities of sixth-gen is having a platform that can network with autonomous systems. Where do these intersect?

TS: The motivations behind the F-47 feel somewhat unclear. Maybe we do need a sixth-generation fighter plane, but we're also 32 years into a fifth-generation fighter plane program that has cost the US taxpayer over a trillion dollars. Norm Augustine, the former CEO of Lockheed Martin, has this famous line: “in 2050, the entire US defense budget will buy exactly one airplane.” That’s an exaggeration, obviously, but that's basically what NGAD — Next Generation Air Dominance — is in my mind. It's like: okay, sure, let's build a $500 million airplane. I'm not really sure what dangerous mission you're going to send a human bag of flesh in a $500 million airplane to do — but sure, why not? I think it's not entirely coincidental that that development design contract — the $20 billion contract — was granted immediately in the wake of Boeing saying that it was about to wash out of its defense business. It’s really important that Boeing has enough contracts to keep them afloat, so is this a real contract, or is it TARP funding for Boeing?

CB: Last I checked, they still hadn't delivered the new Air Force One.

TS: Look into the Orca program — their XLUUV — or the Air Force One project. I mean, Boeing's defense portfolio is rough. I think you have to really look into the motivations behind these programs. $20 billion is not a lot of money for a prime to build a new generation of fighter plane.

On the CCA — Collaborative Combat Aircraft — front, the real question is: fighter planes don't really dogfight. That’s not a thing. I love Top Gun, but even in Top Gun, they had to create the most improbable scenario for how you would get into a dogfight with a fifth-generation fighter plane. It's just completely improbable and far-fetched. The whole game is about sensors and shooters. Can you see them before they can see you, and can you shoot them before they can see you or shoot you? That's it. And so, when you're flying around in a $200, $300 million airplane, you want to have attritable systems that you can tell: “go out and find me the thing and shoot at it before I put myself or this very expensive airplane in range for being targeted by an enemy plane.” That’s really what CCA is.

The pilot is no longer dogfighting like Tom Cruise — the pilot is a command and control operator, and so they're giving commands, saying, "Hey, go out and do a mission.” And then that aircraft has enough autonomy to do the mission without being remotely controlled by the pilot. So the pilot can give two, five, 10, 20 different CCAs different missions and have them just operating in the environment, feeding back whatever information is relevant for the pilot to make decisions. It’s really more of a C2 network. 

CB: Given the state of drone warfare, and once you really scale up production, how do you realistically defend against a massive, attritable drone arsenal, where the enemy could just send endless waves?

TS: Layered air defense. You can't rely too much on a single countermeasure. You need to have electronic warfare in play, really low-cost interceptors in play. You need low-cost cruise missiles with more capability. You probably need to have PAC-3-level stuff to take down the more complicated threats. You cannot be shooting sophisticated, expensive things at non-sophisticated targets. And we just have to be able to produce as a scale. Next year, the Pentagon already announced they're doing an LCCM – low-cost cruise missile – buy from Anduril; this is our Barracuda-500 class. We will produce 7,000 barracudas in calendar year 2027. And, keep in mind, deliveries on Tomahawks and Patriots are both less than 1,000 a year — so we will make ten times more Barracuda-500s than Patriots or Tomahawks. Getting that production cadence right is incredibly important.

CB: Palmer was speaking not too long ago about subterranean warfare. What does he have in mind?

TS: Who knows? There's nothing actively in the works.

CB: On the human enhancement front — whether that's through headsets, exoskeletons, or humanoid robotics — could they play a battlefield role in the next few years?

TS: I don't think we're close at all on humanoids, and I don't think it’s the next few years at all.

CB: Next few decades?

TS: Maybe the next few decades. There's a lot of issues: transportability, power, autonomy. Some of these are getting reasonably good when they're remotely piloted or controlled, but then you have comms issues with maintaining connection. I don't think that we're anywhere near that version of the future. But as far as — can you do a Call of Duty HUD for a soldier? Yeah, we own that contract at Anduril. It’s called EagleEye; it used to be called IVAS. We're doing all sorts of stuff with giving soldiers superhuman awareness of their environment using sensor integration, command and control from the operator, and really user-friendly interfaces.

CB: Why did Microsoft transfer the IVAS contract to Anduril?

TS: It wasn't working. They had made some platform decisions that were exactly the reason why Palmer had no interest in us bidding on this when we first started. Most people assumed that's what we would be doing, that the company was going to be this military XR company. But he felt strongly that the physics were not at the place where you could do it credibly, so he stayed out of it. Then things kind of got to the point where Palmer thought: if they have made different platform decisions, there are actually some things that you could do that would be really interesting. So we talked to Microsoft. They wanted the Albatross off their back, and I think we were just in a really good position to come in and do that.

This interview has been edited for grammar and clarity.

A virus is a pure opportunist, indifferent to geography. To cross an ocean, it requires only the steady incubator of a living host. That biological imperative has not changed. But where these microbes once hitched rides within the warmth of a child’s arm, the world’s most valuable medicines now require their own synthetic ecosystems — fragile glass enclosures meticulously maintained between two and eight degrees Celsius, preserving their microscopic cargo across 10,000 miles of tarmac, ocean, and atmosphere.

The most valuable pharmaceuticals in the world are biologics. These include monoclonal antibodies, mRNA vaccines, and peptides like GLP-1 agonists — all of which are manufactured in bioreactors, industrial vessels in which living cells are cultured to produce complex proteins. A biologic is an intricate lattice of proteins held together by weak chemical bonds. Its three-dimensional shape is its function. Disrupt the shape and you destroy the drug. Leave a vial of insulin on a sunny windowsill, and within hours the proteins unfold. The liquid still looks clear, but the proteins have unfolded into useless tangles. The drug is gone. By contrast, a chemical like aspirin is a small, sturdy molecule. You can leave it on a hot loading dock and it will be fine.

Because the Goldilocks zone for the climate stability of many medicines is so narrow, the industry regulates not just the packaging but the route itself. The European Union’s Good Distribution Practices require that every shipping lane be thermally validated before use. Prior to approval, a logistics manager must map the complete temperature profile of a specific journey — Dublin to Singapore, say — accounting for summer heat, port congestion, vessel swaps, and the brief intervals when a refrigerated container is unplugged. Pharmaceutical companies must prove that the packaging for their drugs can withstand the worst case. Change the ship, and the testing starts over. Change the port of entry, and the validation is void.

Why rely on the ocean at all to transport such delicate cargo? Until recently, much of the industry did not. Cheap, sturdy drugs — like generics, over-the-counter medicines, and raw ingredients — have always traveled by sea. But biologics and vaccines, which are light and enormously profitable, flew. In 2024, US trade data showed that by dollar value, 80 percent of pharmaceutical exports traveled by air. But by weight, the ratio inverted: more than 80 percent moved by sea. For decades, that division served as a kind of “insurance.” Because the highest-value drugs were also the most temperature-sensitive, the industry’s willingness to pay for air freight effectively shielded them from the risks of ocean transit. The drugs most vulnerable to heat were the drugs most likely to be loaded onto a plane.

In recent history, that insurance has been fraying. Over the past decade, the cost of air freight nearly tripled, rising from $700 to over $2,000 per ton of drugs. As a result, these economics have forced the fragile cargo onto the water. AstraZeneca, the Anglo-Swedish pharmaceutical giant, shipped just five percent of its products by sea in 2012; but by 2024, that figure had risen to 64 percent, with a stated target of 70 — part of a deliberate cost-reduction and decarbonization strategy. Eli Lilly and Merck, two of the largest US drug manufacturers, have reported similar shifts. As of 2025, maritime analysts estimate that 3.5 million tons of pharmaceuticals move by sea annually, against half a million by air — and an increasing share of that seaborne cargo is the kind most vulnerable to the journey.

A cargo ship takes five weeks to travel from Shanghai to Rotterdam. During that voyage, a refrigerated container might lose power half a dozen times — waiting on the dock, moving between trucks, or sitting in customs. And every disconnection is a window in which the temperature inside the container drifts away from the two-to-eight degree safe zone that most biologics require and toward whatever is outside it.

The routes themselves pass through some of the most geopolitically volatile water on earth -— straits and canals that sit between rival states, under the flight paths of military aircraft, within range of shore-launched missiles. For years, those vulnerabilities were entirely theoretical. Freight rates were predictable, the major chokepoints were open, and the few disruptions that occurred could be absorbed by buffer stock and rerouting. But in the past five years alone, the pharmaceutical cold chain has been seriously threatened three times.

The first crisis was the coronavirus pandemic. While the virus did not physically block the shipping lanes, it paralyzed the labor force that loads, unloads, and routes the containers that keep those lanes moving. Beginning in early 2020, ports closed, dockworkers fell ill, and shipping containers piled up in the wrong terminals. The pharmaceutical cold chain, now tasked with distributing billions of vaccine doses in addition to its normal cargo, had to compete for cargo space with every other industry on earth trying to restock at once. Freight rates quadrupled: the most critical medicines got priority, and everything else had to wait on the tarmac. In a survey by the Association for Accessible Medicines, a US trade group representing generic manufacturers, companies reported an average 224 percent increase in shipping costs during this period — compounded by the fact that the passenger flights that normally carry 70 percent of airborne medical cargo were grounded. The pandemic has not been a one-off shock. Instead, it revealed that the cold chain was built for a world of predictable schedules and open routes -— a world that no longer exists.

Just as the pandemic ebbed, the Red Sea — the corridor through which roughly 12 percent of global trade passes via the Suez Canal — suddenly became a hazard. In late 2023, Houthi militants in Yemen, an Iranian-backed armed group fighting a civil war, began attacking commercial vessels near the Bab el-Mandeb Strait, the narrow southern entrance to the Red Sea. By early 2024, Maersk, Hapag-Lloyd, and other major container shipping lines had abandoned the route entirely. Their ships diverted south around the Cape of Good Hope, adding seven thousand miles and two weeks of transit time under the equatorial sun to the route. Under normal regulatory rules, every one of those new routes should have been thermally revalidated. Instead, companies filed risk assessments and kept shipping, paying a $450 surcharge per container.

The industry had barely absorbed the Red Sea rerouting when a far larger disruption arrived. On February 28, 2026, the United States and Israel launched air strikes against Iran. Iran retaliated with missile and drone attacks across the region — and closed the Strait of Hormuz. The Strait of Hormuz is not the Red Sea. It is a separate, more consequential chokepoint roughly 1,500 miles to the east, a 21-mile-wide corridor between Iran and Oman that connects the Persian Gulf to the open ocean. A fifth of the world’s crude oil and natural gas passes through it. It is also the primary artery of pharmaceutical trade between Asia, the Middle East, and Europe — and unlike the Red Sea, there is no detour. A ship rerouted around the Cape of Good Hope can still reach its destination. A ship trapped inside the Persian Gulf by a closed Strait of Hormuz simply cannot leave.

Within days of the strait’s closure, commercial traffic through it dropped by more than 90 percent. By mid-March 2026, roughly 170 vessels were trapped in the Persian Gulf, including ships carrying pharmaceutical cargo. Insurance providers pulled war-risk coverage for the region entirely, which forced Maersk to suspend all refrigerated container bookings through the Gulf — reserving the small amount of reefer capacity on ships already en route, before the closure, for essential medical supplies only. Air freight, the obvious alternative, was no longer available either: the strikes had shut down Dubai, Abu Dhabi, and Doha airports, the three cargo hubs that link Europe with Asia and Africa. Global air-cargo capacity dropped 79 percent in the Gulf region within the first week of the war, driving a 22 percent reduction worldwide. Air freight prices quadrupled.

Pharmaceutical companies began rerouting shipments through Jeddah, Riyadh, Istanbul, and Oman, finishing the last leg by truck. Temperature-sensitive cancer treatments and monoclonal antibodies bound for the Middle East were among the cargo stranded. German multinational BASF, one of the world’s largest suppliers of active pharmaceutical ingredients and excipients, announced a global price increase of up to 20 percent, effective March 30, citing an energy cost spike from the Hormuz closure.

Of course, the price cascade did not stop at shipping. Oil above $120 a barrel raised the cost of the petrochemicals that are themselves raw materials in drug manufacturing — the solvents, the polymers used in packaging, the lipid excipients in mRNA vaccines. QatarEnergy declared force majeure on exports of polyethylene and polypropylene, plastics foundational to pharmaceutical packaging and delivery systems. Nearly half of all generic prescriptions sold in the United States come from India, which depends on the Strait of Hormuz for roughly 40 percent of its crude oil imports. Indian API manufacturers began reporting tightening supplies of solvents and key starting materials.

And the damage has extended beyond marketed drugs. A report by clinical data analytics firm Phesi found that 6.7 percent of all active clinical trials worldwide — 4,361 studies across nearly 8,000 investigator sites — have been disrupted by the conflict. Oncology trials are being hit hardest, particularly for non-small cell lung cancer and breast cancer. Most of the affected sites are in Turkey, Israel, and Egypt — and all ten of the world’s largest pharmaceutical companies have clinical trial operations in the region. For the pharmaceutical sponsors of these clinical trials, their delay does not merely reflect a financial loss, but also a delay in knowing whether a drug works which will, on the other end of the trials, mean a delay in patients receiving it.

For wealthy countries, these disruptions are expensive but survivable. The United States has reported no drug shortages directly attributable to the conflict. Large pharmaceutical companies can absorb surcharges, maintain buffer inventories, and reroute through alternative corridors. The head of NHS England said publicly in late March that he was “really worried” about medicine supplies — but the UK, too, has the purchasing power and logistics infrastructure to manage a temporary squeeze. For poor countries, the same disruption is lethal. They generally have no buffer stock to draw down, little leverage to bid for scarce air-freight capacity, and no credit to absorb a sudden spike in the price of generics. Generic drugs operate on single-digit margins: a 55-to-70-percent increase in freight rates, which is what manufacturers reported in early March, cannot be absorbed — it is passed on, first to patients at the pharmacy counter and then to the health ministries, whose procurement budgets have no slack to absorb it. And when a country’s entire public-sector spending on medicines is below two dollars per person, there is nowhere for the cost to go — except to empty shelves.

In some African countries, fewer than 10 percent of health facilities stock a complete basket of essential medicines for noncommunicable diseases. A recent USAID study of Zambia found that in normal periods, stockouts — periods during which a health facility has run out of a medicine entirely — occur 12 percent of the time across 10 essential products, lasting an average of 14.5 days. That was the baseline in Zambia, before any of the current maritime disruption. In Dubai, $600,000 worth of essential medicines destined for Sudan now sit, trapped, on the docks. 90 humanitarian organizations, which together serve around 400,000 patients, said in a joint humanitarian appeal that they will face empty shelves within two weeks.

Better routing cannot solve a supply chain that is already failing at rest. So efforts have turned to the drugs themselves: engineering molecules that can survive what the supply chain no longer can.

Engineering medicines to survive the journey, rather than engineering the journey to protect the medicines, has already been done once, at scale. In 2010, a meningitis A conjugate vaccine called MenAfriVac — a biologic of the kind that would ordinarily require unbroken refrigeration from factory to clinic — was deployed across sub-Saharan Africa’s “meningitis belt” — twenty-six countries stretching from Senegal to Ethiopia where cyclical meningitis epidemics had killed or disabled hundreds of thousands of people, most of them children. Funded by Western philanthropy and manufactured in India, the vaccine was designed to be lyophilized — freeze-dried into a powder — so it could be reconstituted with water. In 2012, regulators approved the new vaccine to be kept at 40 degrees Celsius for four days. MenAfriVac was the first vaccine to break the cold chain, and removing the need for constant refrigeration cut the delivery cost in half. By 2024, MenAfriVac had already reached 400 million people. Today, meningitis A has essentially vanished from the region.

The current goal is to engineer that same stability into the rest of the world’s medicines. A British company called Stablepharma is running clinical trials in Southampton, attempting to make existing vaccines heat-resistant. Their first human trial started in April 2025: for a fridge-free tetanus and diphtheria vaccine called SPVX02. In preclinical work, the vaccines remained fully potent after two years at 30 degrees Celsius and survived three cycles of temperature swings from negative 20 degrees Celsius to 40 degrees Celsius. Stablepharma claims to have 60 more vaccines in development, similarly engineered to sit on a shelf for four years. If their promise holds, a vial of a vaccine like SPVX02 could sit on a shelf in Juba or Aden for the better part of a presidential term and still work.

Other stable vaccine approaches want to abandon the vial entirely. Micron Biomedical, an Atlanta-based company backed by the Gates Foundation, has developed a small adhesive patch studded with hundreds of dissolving microneedles — each needle made of dried vaccine embedded in a solid matrix. Press the patch against the skin for a few minutes, and the needles dissolve, releasing the antigen directly into the dermis — no syringe needed. This design, in addition to being shelf-stable, removes the need for a trained professional, or for the safe disposal of used needles. The thermostable design of the patch is derived from the same principle as lyophilization: the vaccine is locked in a solid state, so there is no liquid medium for the proteins to unfold in. In a Phase 1/2 trial published in The Lancet in 2024, a measles-rubella version of the patch was tested on 285 participants in The Gambia, across age cohorts ranging from adults down to infants as young as nine months. Over ninety percent of infants in the trial achieved seroprotection -- comparable to a conventional injection. If those results remain in later trials, the logistical implications are dramatic: a nurse in rural South Sudan could carry a sheet of these patches in a bag.

None of these technologies have yet replaced the cold chain. MenAfriVac still required reconstitution and a four-day window; Stablepharma is only in Phase 1 trials; and the microneedle patch has not yet reached Phase 3. But the direction of this research — toward medicines that hold their structure outside a fridge — is clear, and it is the same direction Balmis intuited on the deck of the Maria Pita: if the journey is too perilous for the cargo, then perhaps it would be wise to make the cargo itself more stable. For Balmis, “making the cargo more stable” meant incubating the smallpox vaccine inside a living body. Today, a growing number of researchers believe they can design a molecule that can survive, on its own, the worst the journey will throw at it — one whose every shock has already been absorbed before the vial ever leaves the factory.

That is no longer the case. The sea has once again become a visible, contested geopolitical space. But what is different in the 21st century is that such contests for maritime supremacy are now transforming the ocean from an open space of circulation into an increasingly enclosed and territorialized system.

That challenge to the global maritime order emerges, first, from the relative erosion of the naval capabilities of Western nations, including the United States, which ostensibly remains the dominant maritime power. Second, from technological developments — in the military domain, such as long-range weapons that can easily disrupt maritime supply lines, and in the civilian domain, technologies that allow for the exploitation of undersea resources — which have fundamentally altered the terms of that competition. Third, from the rise of China, an industrial, commercial, and increasingly naval power, one that holds a fundamentally different view of the seas than the West does.

To answer that challenge, the United States and the nations of the West cannot rely on naval military superiority alone. Born as an outpost of a merchant republic, America’s coastal cities were the core of the nation from its birth. But soon after the revolution, the conquest of the West shifted most of the country inland. Since then, the nation has oscillated between periods of deep infatuation with the call of the seas and inland introspection.

Today the US Navy is the most powerful in the world, but the rest of the traditional American maritime economy is in terminal decline. That could be an advantage. Freed from the models of the past, the US can approach the maritime domain with fresh eyes. But in the new geopolitical context, even that may not be enough to sustain American maritime preeminence. The US must now do something it has never fully committed to: embrace its identity as a sea power in the fullest sense — commercial, industrial, technological, financial, and political. Open seas are the key to American global supremacy, and only a true sea-power identity can rebuff emerging rivals and the continentalization of the maritime domain.

What Is Sea Power?

The great popularizer of the concept of “sea power” was the American admiral Alfred Thayer Mahan, who published The Influence of Sea Power upon History in 1890. The book became an international bestseller and was read as a manual by key political leaders at home and abroad. Mahan advised Theodore Roosevelt on his campaign to turn the long-neglected US Navy into a force capable of challenging European rivals. Unintentionally, his ideas also influenced future enemies of the United States, most notably Kaiser Wilhelm II and Emperor Hirohito of Japan.

Mahan understood that war at sea has always been total war: “The control of the seas, and especially along the great lines drawn by national interest or national commerce, is the chief among the merely material elements in the power and prosperity of nations. It is so because the sea is the world’s great medium of circulation.” He saw sea power as the industrial-era fusion of naval strength, commercial shipping, overseas markets, colonies and coaling stations, and national productive capacity. For Mahan, command of the sea meant securing the maritime circulation of goods, capital, and strategic resources that sustained an industrial economy capable of winning wars.

A great power needed fleets, but also shipyards, a merchant marine, a civilian commercial fleet that could carry trade in peacetime and support national logistics in war, ports, and a political culture supportive of sea power.

A contemporary of Mahan, the turn-of-the-century British strategist Julian Corbett placed even more weight on a holistic vision of sea power: “Command of the sea… means nothing but the control of maritime communications.” Mahan’s understanding of sea power was broad and incorporated multiple dimensions, but his logic still pointed toward one central purpose: enabling a powerful navy to win decisive battles at sea. Where Mahan argued that “the enemy’s ships and fleets are the true objects to be assailed,” Corbett countered that “naval warfare does not begin and end with the destruction of the enemy’s battle-fleet.”

For Corbett, sea power became the architecture of a national grand strategy: a system in which trade, finance, industry, logistics, and naval force worked together to shape the global balance of power through control of the sea. Corbett’s current intellectual heir, the historian Andrew Lambert, argues that sea power cannot be reduced to material capabilities alone — it requires the political decision of a nation to make the sea the center of its imagination and economy. Rome, Habsburg Spain, and even Peter the Great’s Russia all had large navies, but their center of gravity did not pulse toward the seas. By contrast, the nations Lambert sees as the true embodiment of sea power — Athens, Venice, and Britain — were societies fundamentally shaped by and dependent on the seas.

The strategic logic of sea power can be seen in the classic example of Britain during the Napoleonic Wars. A “nation of shopkeepers” defeated the greatest general of the era by using the sea to strangle French power from afar — blockading French and allied ports, controlling trade routes, and subsidizing allied armies to fight Napoleon on land while keeping British boots off the continent.

The Royal Navy was the instrument of this strategy: one of the most meritocratic institutions of its age, manned by superbly trained sailors and led by commanders of genius such as Horatio Nelson. But the glory of Nelson’s martyrdom at Trafalgar would have meant little without Britain’s capacity to sustain those blockades for years. The British turned the sea into a force multiplier against Napoleon through total national mobilization, in which the navy was inseparable from the factory, the banker, and the accountants who built the sophisticated logistics that kept ships at sea for months at a time. It was “shopkeeping” as much as the discipline of British sailors that defeated Napoleon.

How China Is Expanding Its Maritime Power

The American position against China today is not equivalent to Britain’s position against Napoleon. America is not materially weaker in the same way, though it has deindustrialized. But the strategic logic is similar: the United States must use command of the seas, finance, alliances, technology, and a network of bases to offset China’s advantages in industrial scale and manpower.

The particular challenge that China poses, compared to Napoleonic France, is that it is much more than a continental empire to be contained from the sea. China is also a shipbuilding and manufacturing superpower, actively trying to contest the maritime order that once allowed sea powers to defeat larger continental rivals. Although China lacks a navy capable of projecting power overseas, it might be growing powerful enough to deny the US Navy control over Chinese waters. The naval historian Lincoln P. Paine observes that after the Second World War, “the US fleet exists to project power and safeguard trade, not to fight fleets of comparable capabilities, because there are none.” That won’t be true for much longer.

In recent years, China has become a more advanced student of Mahan’s and Corbett’s gospel of sea power than the United States itself. China has also learned from the failures of previous challengers to British and American naval supremacy, and has gone well beyond simply building more warships.

Shipbuilding sits at the center of an ecosystem of highly strategic supply chains. In the East Asian development experience of Korea and Japan, shipbuilding became a linchpin of industrial development because it allowed those countries to build up capacity in sectors like steel, transportation, and energy, producing large conglomerates that spanned multiple areas of the economy — Honda, Mitsubishi, Samsung.

China followed a similar path, but — as in many other sectors — has used its immense capacity to scale toward a wider ecosystem in ways its competitors couldn’t: shipyards, a merchant fleet, advanced port infrastructure, coast guard forces, fishing fleets, logistics networks, and even gray private security actors that can be deployed to protect Chinese interests in Africa and Central Asia while denying any direct military intervention. Quite the achievement for a nation that spent centuries banning its people from venturing out to seas.

By 2024, China accounted for 53 percent of the global commercial shipbuilding market, up from just five percent in 2000. China’s largest state-owned shipbuilder, CSSC, delivered more commercial tonnage in 2024 than the entire US shipbuilding industry has produced since the end of World War II. Many of these Chinese shipyards are dual-use, both building the merchant vessels that carry global trade and modernizing the People’s Liberation Army Navy (PLAN). Chinese cargo vessels are also capable of resupplying PLAN ships and of carrying container-sized missile launchers deployable from the decks of ordinary Chinese merchant vessels.

As per Corbett, Beijing is preparing to command the sea in wartime while embedding itself into the everyday architecture through which global maritime commerce moves. This is what makes the Chinese challenge different from a purely military naval arms race.

Chinese firms, above all COSCO Shipping and other state-linked companies, have acquired stakes in critical ports and maritime infrastructure across Europe, Africa, Latin America, and Asia. Although Beijing presents these acquisitions as commercial, the reality is geopolitical. In March 2025, under pressure from the Trump administration, CK Hutchison — a Hong Kong-based firm that had managed terminals at both ends of the Panama Canal — announced an agreement to sell most of its global ports business to a BlackRock-led consortium. The agreement included a 90 percent stake in Panama Ports Company. Beijing denounced the sale as contrary to China’s national interest and immediately ramped up inspections of Panama-flagged ships in its waters, threatening one of Panama’s main streams of revenue.

China’s maritime power also operates in the gray zone between state and private, civilian and military. Its maritime militia and fishing fleets already perform coercive functions in disputed waters. Chinese private security companies have expanded abroad alongside Belt and Road projects, providing armed escort against pirates, intelligence services, and protection for state-owned fisheries — often in a legal gray zone where the line between private firm and state interest is deliberately blurred. Counting only destroyers and aircraft carriers leaves too much out of the picture to understand China’s real strength at sea.

The Continentalization of the Sea

The question of sovereignty over the seas is not new. The classic debate set the idea of open seas, upheld by the Dutch philosopher Hugo Grotius’s Mare Liberum (1609), which argued that no state could own the ocean and that maritime trade should remain free, against English jurist John Selden’s Mare Clausum (1635), which defended the idea that seas could be claimed, enclosed, and governed by states. Both positions, however, shared a vision of the maritime domain as an endless space of circulation, meant to be charted. The debate about sovereignty over the seas was ultimately about who could navigate through them.

In the 21st century, the proliferation of new technologies capable of exploiting subsea natural resources has given new urgency and a different meaning to this centuries-old question. Disputes over Exclusive Economic Zones (EEZs), as well as seabed resources, fisheries, and maritime chokepoints, have multiplied. (An EEZ is the 200 nautical miles from a state’s coast, where the state does not have full sovereignty but has special rights over resources.) The past decade has seen a sharp rise in conflicts — such as that in the Eastern Mediterranean among Turkey, Greece, Cyprus, Egypt, Israel, and Lebanon over gas fields, pipelines, and drilling rights. In 2020, Turkish seismic surveys in disputed waters produced direct naval tensions with Greece, including a collision between Greek and Turkish frigates. Athens called for support from its European partners, but with meager success — both Turkey and Greece are NATO members.

In the South China Sea, Beijing has used a more sophisticated approach: creating artificial islands — which expand China’s land surface in the area and with it its “legal” claim to territorial waters and EEZs across the disputed region — and deploying maritime militia and fishing fleets to back territorial claims through fait accompli.

What is new is that this logic now reaches downward, into the seabed. Polymetallic nodules, cobalt-rich crusts, and deep-sea sulfides are becoming part of the critical minerals race. In its quest to become independent of China’s rare earth chokeholds, the Trump administration issued an executive order in 2025 that explicitly framed offshore and deep-seabed minerals as a national security and economic priority. Meanwhile, China already holds multiple exploration contracts with the International Seabed Authority, including in the Clarion-Clipperton Zone, a vast deep-sea region of the Pacific containing some of the world’s most abundant polymetallic nodule deposits, the minerals needed for batteries, power grids, data centers, robots, and the AI economy as a whole.

The result is that parts of the high seas are no longer simply open spaces of navigation, but increasingly pre-allocated as future zones of resource extraction. Even fisheries are part of this appropriation, as recent tensions between Argentina and Chinese fishing fleets in Argentine waters illustrate.

The traditional image of the ocean as a vast global commons — a highway over which great powers project naval and commercial power — still matters. But it is increasingly incomplete. States are no longer fighting only over command of sea lanes, but for the resources beneath them. They are fighting over the sea as land. The result is a new politics of sea appropriation in tension with the logic of circulation.

The sea was historically an open space — fluid, mobile, and more resistant to enclosure than land, under fewer political constraints. But in the industrial era, as Carl Schmitt wrote in his 1942 book Land and Sea, “The sea is today no longer an element as it was in the time of the whale hunters and corsairs. Today’s transportation and communications technology has made the sea into a space in the contemporary sense of the word. Today, in times of peace, every shipowner can know daily and hourly at which point in the ocean his ship on the high sea is to be found.” The dragons, in other words, are gone.

The Houthis show how the continentalization of the sea can be practiced by actors without a conventional navy. Since November 2023, they have used drones, cruise missiles, and anti-ship ballistic missiles to threaten commercial shipping and naval vessels in the Red Sea, turning a maritime corridor into a contested zone from land. The current Iran war extends this logic further: Iran has reportedly fired long-range missiles at Diego Garcia, a US-UK base in the Indian Ocean roughly 4,000 kilometers away. The lesson is that islands, ports, and distant bases are no longer sanctuaries. Long-range precision systems allow continental powers and their proxies to contest maritime space from shore without directly controlling the sea.

In this sense, the central challenge the West faces in the maritime realm is not only naval competition with China, but the continentalization of maritime space — the conversion of open sea into contested territory. We have already seen this logic unfold with the Houthis in the Red Sea and Iran in the Strait of Hormuz. China is an actor that can surpass those cases by several orders of magnitude; its current goal may not be a decisive naval battle, but rather to make access to its surrounding waters prohibitively costly. In this environment, sea power becomes a question of endurance. To project power across oceans for long periods, the United States must combine naval force with the “shopkeeping” foundations needed to command the seas.

Re-opening the Seas

The United States still retains the most traditional instruments of sea power: a powerful blue-water navy, aircraft carriers, and a global network of bases that give America the ability to project military power across oceans. But across nearly every other metric, the US has lost ground. It lacks a merchant navy, its shipbuilding output is negligible, and the state of its shipyards — with a declining workforce — is so poor that it has had to send US Navy ships to Japan for repairs, not as a matter of cost efficiency but for lack of capacity at home. No great naval power has endured without also being a great commercial maritime power, capable of sustaining trade, moving goods, and turning maritime wealth into strategic depth. Nor has any great naval power remained dominant without an industrial base capable of building and replacing its own warships. As they say, ships may win battles, but shipyards win wars.

In the last ten years, China built 6,765 commercial ships; Japan, 3,130; South Korea, 2,405. The United States built 37. China alone now accounts for 53 percent of the global commercial shipbuilding market; the United States, 0.1 percent.

America has compensated for its weakness in the commercial maritime domain with an informal bargain among its allies. Washington provided the military muscle while allied countries and their companies provided the maritime logistical depth. Japan and Korea provided the shipyards, together accounting for roughly 40 percent of global commercial vessel production, while European companies provided the merchant fleet of the West. Four European companies alone account for over half the world’s container fleet — MSC (Swiss-Italian, 19.9 percent), Maersk (Danish, 14.6 percent), CMA CGM (French, 12.7 percent), and Hapag-Lloyd (German, 7 perent). The US Navy provided the security umbrella under which the system flourished, protecting global supply lines by ensuring freedom of navigation for the ships of all nations.

That bargain worked reasonably well in an era when the sea was treated as a stable background condition of globalization. The return of maritime competition has exposed the fragility of the arrangement. Recent maritime disruptions in the Red Sea and the Strait of Hormuz revealed that the question was not only whether the US Navy could escort vessels, but whose vessels were being protected, under what authority, and with what political commitments from the states that benefited from American protection.

The United States therefore faces a hard choice. It can try to rebuild its own maritime and shipbuilding base at speed — but that will take years, and may never be sufficient on its own. Or it can formalize the maritime bargain with its allies, creating arrangements with allied governments and key shipping companies that would guarantee access to merchant vessels and their crews in the event of war or major crisis. This would not require nationalizing allied merchant fleets, but it would require a clearer framework of strategic availability.

This is where allied sea power becomes central. The goal should not be to make every ally dependent on American naval protection while their commercial fleets and industrial capacities remain strategically detached. American maritime strategy may already be reshaping along those lines. Washington has begun looking to Japan and South Korea as industrial partners in the reconstruction of American sea power. South Korea’s Hanwha acquired Philly Shipyard in 2024, giving one of Asia’s leading shipbuilders a direct foothold in the US industrial base, and has since expanded its role in US naval and commercial shipbuilding. Japan, meanwhile, is becoming central to American efforts to expand ship repair, maintenance, and potentially shipbuilding cooperation in the Western Pacific.

The same logic applies beyond shipyards. Greece is the world’s largest ship-owning country by deadweight tonnage; the United Kingdom is home to some of the world’s largest shipping companies and maritime insurers, alongside Denmark and the Netherlands, which add port and logistical capacities. These are the commercial and industrial foundations of sea power. A serious American maritime strategy would therefore not merely ask allies to spend more on defense. It would integrate their shipyards, fleets, ports, capital, and expertise into a common system capable of preserving Western hegemony at sea.

A New American Sea Power

The United States has always been tempted to turn away from the sea. Despite its two long coastlines, the sheer scale of its continental interior pulls the national imagination inwards. But America, thanks to its unique culture as a frontier society, offers an embrace of what the sea could truly mean for humanity that no present rival or past model could match.

At its essence, sea power is the use of an expanding space to maximize one’s capabilities against stronger or more populous enemies. Properly understood, it means orienting the energies of a nation toward communion with the maritime space, building a society open to risk, commerce, and techno-political imagination around the ocean. It implies the ability to open new worlds through the sea. Today we are approaching a new understanding of the sea, not only as a space for circulation or short-term resource exploitation, but as a new space to settle. That’s the new foundation of sea power that America can offer.

One advantage the US holds over China is that China lacks the capacity to protect its overseas interests. Even as it acquires assets overseas, its expansion of territorial claims at sea — via artificial islands — is best understood as an extension of Chinese continental needs. China is less able to settle the maritime frontier than the US is. That may decide the future of the seas.

The United States, then, can preserve the openness of the maritime system by preventing any continental power from enclosing key regions while enabling allied maritime capacity to grow. The paradox is that the maritime sovereignty of individual Western nations may depend on deeper collective integration, something poorly understood on both sides of the Atlantic. If the oceans become a patchwork of fortified regional spaces, the West loses not only military freedom of movement but the material foundation of its political order and the next frontier to conquer. Open seas are the operating system of Western strength. This is why the defense of innocent passage and open sea lanes cannot be framed merely as an American military project. It must be a common Western project: keeping the sea open enough to move through, stable enough to build upon, and protected enough that no continental power can threaten its freedom.

Ferdinand de Lesseps arrived in Egypt in 1854 to realize a dream millennia in the making. By 1869, he had accomplished through human ingenuity what had nearly been a natural accident: connecting the Mediterranean Sea and the Red Sea via the Suez Canal. Following this success, de Lesseps went to South America to lead a feat of engineering even more ambitious, connecting the Atlantic and Pacific through a canal in Panama, cutting the Americas in half. And though de Lesseps was unsuccessful in his time, the attempt set a model that inspired future engineers. The Panama Canal project was subsequently taken up and completed by the United States in 1914. Such undertakings of national advancement were celebrated without apology. God gave man wisdom and free will; it was expected that man would ambitiously use those capabilities to reshape the earth according to humanity’s own needs while using that same wisdom to prudently steward the natural world.

6,000 years ago, humans began reshaping the land they inhabit through surface irrigation. The Dutch made it a national project. Using rudimentary walls to keep out water from the North Sea, the Dutch undertook a multi-century project of claiming new land — called polders — beginning in the 13th century. This process continued for hundreds of years, taking on a new scale and level of sophistication after the Second World War. By 1996, the Dutch Delta Works had reclaimed 620 square miles of polders and created many more miles of transitional wetlands. Today, about 26 percent of the country’s land sits below sea level; this same area contains 60 percent of the nation’s population. Land reclamation enabled territorial growth without conquest and created the Netherlands as we know it today.

More recently, the South Korean government undertook the world’s largest single land reclamation project. Completed in 2006, the Saemangeum Seawall on the southwest coast of the Korean peninsula enclosed 150 square miles for about 3.2 billion dollars. Where the Delta Works largely reclaimed low-lying flooded land, the Saemangeum Seawall goes deeper, down to the seabed averaging 36 meters deep. Advances in reclamation technology enabled South Korean builders to raise more land from further depths.

Or take Singapore. The government wants the country’s landmass to grow from 244 square miles to about 300 by 2030. Singapore is more than halfway there, creating new land for airports, housing, and military infrastructure. Most of the reclaimed land previously sat within 20 meters of the surface. Deeper reclamation, down to 40 and 50 meters, would be an engineering and financial challenge. But for a country like Singapore where small parcels of land lease for millions of dollars, every acre matters.

The most high-profile and potentially concerning example of land reclamation from the American perspective is China’s decade-long effort to build islands in the South China Sea. In the early to mid 2010s, China built more than 3,000 acres of artificial islands across various reefs and shallows in order to assert territorial claims over the waterway. More recently, China has begun dredging work on a disputed shoal, likely to make it fit for docking large naval vessels.

The potential of land-building is enormous, and right now, seemingly constrained more by political will than any international law. Given an acknowledged right for a state to exercise full sovereignty in constructing an artificial island within its Exclusive Economic Zone as per Part V of the United Nations Convention on the Law of the Sea, there is no limit on the size of the island within the 200 nautical mile zone from a nation’s coast. Ambitious modern states are raising islands from the sea and reclaiming acreage from their shorelines to accommodate the growing needs of thriving cities. But not America.

That hasn’t always been the case. American heritage is one of exploring and settling new land. And when there’s no more land to settle, we have a history of building it. There is no better example than San Francisco, both the literal and metaphorical end of the Western frontier. So, when the builders of San Francisco needed new land, they created it.

Before the Gold Rush, Yerba Buena Cove extended inland to what is now Montgomery Street — the Transamerica Pyramid stands on ground that was, in the 1840s, a shallow tidal inlet. When a rush of forty-niners arrived, some five hundred ships clogged the harbor, many of them abandoned by crews who had already left for the gold fields. Enterprising San Franciscans sank them deliberately, claimed the land beneath, and filled the gaps with sand, rubble, and earth. Many of those same ships remain buried under the city’s streets today. By the early 1870s, a seawall enclosed the former cove along what is now the Embarcadero. The city had, within a single generation, manufactured an entire commercial district of 10 acres in what became the densest part of the city.

After the 1906 earthquake, the city used the rubble to fill a marshy cove on the northern waterfront, creating the Marina district in time to host the 1915 Panama-Pacific International Exposition. When the fair closed, developers bought the reclaimed land and built the residential neighborhood that stands there now. On the other side of the city, Mission Bay, once an actual bay, was filled over decades, first with hydraulic mining debris washing down from the Sierra foothills, then with rubble from the earthquake, until by 1910 it had been reduced to a narrow channel. South of Market, at Islais Creek, landowners formed California’s first non-agricultural reclamation district in the 1920s to fill 280 acres of tidal mudflats, quarrying nearly half a million tons of rock from Potrero Hill and Bernal Heights for the seawall alone. At Hunters Point, the Navy filled roughly a square mile of underwater property between 1940 and 1944 to build a wartime shipyard.

Then, in 1936, the Army Corps of Engineers began its most ambitious project of land creation: Treasure Island. A 400-acre artificial island built on the shoals north of Yerba Buena Island, it was dredged from bay mud at a cost of about $4.5 million at the time, completed in less than two years, and opened in time to host the 1939 Golden Gate International Exposition. When the fair ended, the Navy seized the island and processed 12,000 men a day through it for Pacific assignments.

None of this was controversial. Land reclamation was how the Bay Area grew, accommodating a population that settled in to work in the region’s shipyards and military installations. By the late 1950s, the Army Corps of Engineers estimated that 243 square miles of reclaimable land around the Bay had already been filled. Far from debating whether or not such policies were desirable, city, state, and national officials were consumed with the question of how ambitious and far-reaching land reclamation efforts should be. The idea of regularized terrestrial growth at the time was infectious, inspiring more than just engineers and municipal planners. One of the most ambitious proposals, one that received national attention and federal support, came from a theatrical producer and schoolteacher named John Reber.

Reber had arrived in the Bay in 1907 when he was 20 years old, the year after the earthquake, and spent the next two decades managing musical productions while sketching plans to reshape the entire shoreline. His proposal, first drawn in 1929 and refined over a quarter century, was the San Francisco Bay Project: two earth-and-rock dams, one spanning from Marin County to Richmond and another from San Francisco to Oakland. The dams would convert the Bay into two freshwater lakes storing 2.4 million acre-feet of water annually. Between them, 20,000 acres of filled land would carry a freshwater ship channel, airports, naval installations, and locks comparable in scale to those of the Panama Canal. Rail and highway corridors would run along the tops of the dams conducting wide lanes of automobile traffic.

But for all of Reber’s showmanship and enthusiasm, the project was hampered by his lack of technical expertise. In 1933, a meeting brought him face to face with former President Herbert Hoover, a Stanford-trained geologist who had returned to his alma mater after losing to Roosevelt. Hoover was a mining engineer of international reputation before he was a politician. According to Reber’s later accounts, Hoover called the Reber Plan “the most complete proposal for the Bay.” That endorsement gave the project credibility and Reber knew how to build on it. Soon after, he began to promote his plan full-time — debating with opponents, lobbying policymakers in California and Washington, corresponding with supporters — alone at home on his typewriter.

In 1940, Reber displayed a model of his plan at the World’s Fair on Treasure Island. Two years later, San Francisco’s Board of Supervisors and mayor had formally endorsed the plan. So did the San Francisco Chronicle. California’s farming interests backed it, because the freshwater lakes promised relief from saltwater intrusion into agricultural land. The Navy was interested — Reber, with a showman’s instinct for his audience, had incorporated military installations into successive versions of the plan throughout the war years. When Senator Sheridan Downey of California held hearings for the Senate Subcommittee on Public Works in San Francisco in 1949, advocates outnumbered critics five to one.

The opposition that did emerge was parochial, not principled. The Oakland City Council and the Alameda County Board of Supervisors opposed the plan, fearing that Oakland’s port would be trapped behind a dam, accessible only through locks on filled land off the Berkeley waterfront. But this was a fight over regional advantage, not over the legitimacy of the enterprise. Nobody argued that 20,000 acres of new land were the wrong kind of ambition.

In 1950, Congress funded a full Army Corps feasibility study through the Rivers and Harbors Act. The Corps built a hydraulic model of the entire Bay system in a warehouse in Sausalito — 1.5 acres, reproducing tidal flows from the Pacific to Sacramento at a 1:1,000 horizontal scale and 1:100 vertical. It was among the largest physical models ever constructed. Through testing in the 1950s, it demonstrated that Reber’s barriers would cause catastrophic tidal disruption, salinity collapse, and flooding throughout the southern Bay. By the early 1960s, the plan was dead. The hydraulic model still sits in that Sausalito warehouse, open to visitors as an educational exhibit. It has not been used for research since 2000.

What followed the failure of the Reber plan was not an improved and more sophisticated endeavor to expand San Francisco. It was, in effect, the end of any further land reclamation efforts in the Bay Area. Public pressure in response to land reclamation efforts caused the California legislature to pass the McAteer-Petris Act, creating the Bay Conservation and Development Commission (BCDC) with permit authority over all filling of San Francisco Bay. The BCDC became permanent in 1969. Under its jurisdiction, any placement of fill requires consistency with its Bay Plan document, which prohibits any further land reclamation except for certain transport and public use cases.

Reber’s plan was a flawed work of ambitious engineering. His cost estimates were fantasy and his hydrology was certainly wrong. He was an actor, not an engineer, and the project needed one in order to be rightly designed and implemented. But a new generation of municipal planners took the wrong lesson from his failure — deciding that ambition of that scale, and land reclamation of any scale, should not be allowed any longer.

More so now than ever, San Francisco is a city with too many people competing for too little land in a regulatory environment that heavily restricts new construction, whether transportation, infrastructure, or housing. Even bolder proposals of the modern day — like turning the Presidio into a dense ‘Freedom City’ —suffer from the lack of imagination encouraged by the stifling attitudes towards hardware and terraforming.

Growing San Francisco is entirely technologically feasible. The question is of political will. The fact that an endeavor to create new land in San Francisco would be highly profitable should be an incentive to both local officials and developers.

San Francisco’s 2025 property assessment puts a total land value, stripped of buildings and improvements, at $172 billion across roughly 30,000 acres. That lands at roughly $5.7 million per acre. At this valuation, every square mile of reclaimed land in the city would be worth about 3.6 billion dollars of land value alone. Since the assessor data also includes valuations of the improvements on top of San Francisco’s land, you can factor in improvements comparable to the city average and that figure rises to $11.3 million per acre — on land value alone, before buildings go up. For comparison, Singaporean land reclamation efforts in the 2010s cost between $860,000 to $2.7 million per acre. Building new land in San Francisco would cost a fraction of what that land would be worth once built. Rolling back the BCDC is unlikely to begin with the San Francisco city government: the same authorities who have spent decades tightening their grip on permitting are not likely to loosen it voluntarily. But that’s not to say that it couldn’t happen in San Francisco. Already, housing is being built on Treasure Island, one of the few cases of recent housing construction in the city. But when the city’s own government acknowledges a housing shortage, it should be willing to do much more.

Johan Hendrik van Mastenbroek, “The Day Before the Closing of the Zuiderzee,” May 1932. A sketch of the Afsluitdijk, the dam that closed the Zuiderzee off from the open sea, in its final days of construction.

American cities cannot reject the methods and ambitions that were prerequisites to their construction. A city that owes its existence to a willingness to create new land has only a political reason to halt such enlargement. An intergenerational compact is violated when one generation puts away the tools for building, maintaining, and growing that the next generation will need. Americans sense this, even where they can’t name it — in the anxiety about affordability, the eroding faith that taxes are being well spent, the feeling that the country is consuming itself rather than building towards something material. The unwillingness to do ambitious land reclamation projects is a symptom of those problems.

A country confident in itself would find land reclamation an easy and profitable win. Filling in San Francisco’s shallow shoals would be the least of the projects of civilizational transformation that we could pursue once more. And San Francisco is only the beginning. Such ambition embraces categorical change: if we can refill the Great Salt Lake, restoring water to arid land, there’s no reason that we can’t create new frontiers of land where there once was water. Imagine an America whose territory grows yearly — not through conquest, but by the power of our engineering talent and our innate desire to grow and expand.

Special jurisdictions have proliferated since the 1980s, from a few hundred globally to over 5,000 by the late 2010s. These areas, often acting as taxation and regulatory havens, have ballooned at a time when sovereign states have persistently failed to deliver economic growth, regulatory relief, and demographic stability. Special jurisdictions — like Hong Kong, the Dubai International Financial Center, or Shenzhen — often serve as engines for economic activity, drawing in corporations and high-net-worth individuals alike.

I sat down with Patri to discuss the pressures and challenges of declining state capacity, the rapid advancement of information technology, the legal and territorial dynamics of the ocean, and logistical questions relating to the concept of societal exit. What follows is a transcript of our conversation.

Carson Becker: Where are we in terms of governance and technology?

Patri Friedman: We’re at a place where a lot of people have a lot of ideas about how to make systems that will work better, and it’s really hard for them to put those ideas into practice. And we’re just starting to have the first few examples, like Prospera, of people successfully getting legal autonomy to create new types of societies. There’s Prospera; I’m working on a bunch more of these around the world. More and more governments and founders and investors and other stakeholders are interested. But there hasn’t been a success story yet. There’s a lot of interest. That’s what I dedicate myself to. But it’s still very early. The prototype is built and operating as a technology, but it hasn’t scaled.

CB: Tell me a bit more about this prototype.

PF: My personal North Star is physical territory with legal autonomy. That’s what I’m all about. And Prospera is the place where that’s closest to being realized. Under the Honduran ZEDE program, which is in the Honduran constitution, they have control over most of their local commercial law, beneath the country’s constitution and criminal law and treaties. It’s an operating independent jurisdiction. It’s a proof of concept.

CB: How do you address the state’s monopoly on violence in the context of charter cities? Providing security through military power is a core function of the state. How does one secure the territories that you’re describing?

PF: Law is the answer. Laws and courts and treaties and arbitration and bonds. A lot of people don’t know how disputes in the nation-state system are handled, or disputes between businesses and countries, but there is this global system with multilateral investor treaties that enable people to enforce agreements on sovereigns. We’re not trying to create or invent anything new. What’s most important to highlight is that sovereigns want to be able to bind themselves. Say you’re a country, you’ve got some resource, you want some company to come and extract the resource. Say you’re the strong man dictator — you want them to give you some of the money, and then they’re going to keep some of the money. Otherwise, they wouldn’t be there. So, how do you make that agreement? It is of massive value to the world, for businesses and countries to be able to make binding agreements. That’s what powers all investment in the Global South.

And we have a way of doing this through things like ICSID, investor-state dispute resolution networks and treaties. The way it works is: a company and a country make a deal, the country breaks the deal, the company sues them under these investor-state dispute resolution networks in arbitration. If they win — if they show they said they would do this and then they did that — the arbitrator rules in their favor. And they can take that award to almost any country in the world. It’s not like a top-tier treaty, but it’s in the next tier, just below the New York Convention. And every other country in the world will enforce that ruling by seizing whatever assets the losing country has domiciled elsewhere. You can’t seize military assets, you can’t seize embassies, and you can’t seize art. But everything else: bank accounts, property, anything that’s not military, embassy or cultural, can be seized. Countries enforce it all on each other.

CB: Do you think we live in a unipolar world system, or do you think we live in a more fractured system that will tolerate the level of competition among states that you’re envisioning?

PF: It’s hard to speak simply about the state of the world, but it’s definitely multipolar. We’ve got great powers in different regions, different continents, and they have a ton of influence. And there’s a question of how these new jurisdictions play with the great powers. Are they in their interests or out of their interests?

CB: Well, are they?

PF: It depends. And they can be. I don’t know whether I can give any specific examples for confidentiality reasons.

CB: Would you consider Britain’s network of offshore financial jurisdictions to be something that’s in the interest of the great powers? It’s convenient for many different parties.

PF: Yes.

CB: How would you describe that particular arrangement? You’ve got the City of London, and then, for example, Bermuda. You’ve got places like Guernsey. You have these little nodes with a quasi-sovereign function.

PF: It’s super relevant to the charter model because this Empire model is based on English common law. We have data showing it’s our best legal operating system, our best source code, or programming language, in the law-as-software metaphor. The UK mostly handles the international treaties and the National Defense and the military. And then they have these sub jurisdictions that have a ton of local autonomy that they can use to serve global customers under that umbrella.

I think it’s actually great for diversification of jurisdiction. It’s almost like an incubator. They’re doing the layer that’s really hard for all the startups to do. They’re doing this really hard, really deep, really technical, really expensive layer of military and international weight and treaties, but then choosing to delegate to these jurisdictions really meaningful autonomy, such that they’re different from the UK. I think that structure is very supportive.

CB: Hong Kong is also an interesting example. China absorbed Hong Kong, and they maintain their separate status because it’s convenient. China didn’t fully absorb it, as some anticipated.

PF: It’s a fascinating case study. There’s a lot of emotion around it. The land was China’s — and the lease expired. It’s their jurisdiction. They have that legal right. And of course, they’re kind of anti-democracy, it’s China. They restricted political freedom in a more China-like way. But on the economic freedom side, they didn’t want to kill the goose that laid the golden egg. They saw the special jurisdiction and the institutions and rules that applied there, and preserved it.

I think it’s a fascinating counterpoint to the majority perspective because people are often like, “but what if the government just comes in and seizes it” and there is this international arbitration thing that I mentioned earlier. You do have recourse. But a city is not a resource. A city is an economy. It’s an organism and it’s functioning and producing value because of the rules and institutions that are there. My mission is to rezone more of the world with better rules and institutions. But it means the government can’t take it over and change it and keep it. If you change it, you’ll ruin it. They can either take it over and not change it and try to capture some of the value, or they can ruin it — and then they’re gonna lose it. It self-destructs if you fuck with it.

CB: One of the things that you mentioned in your famous Cato article from 2009 — you said there are a few ways to bring about a change in governance. You can have a war, a revolution, or an election. But you don’t often change a form of governance without violence. How do you perpetuate these arrangements — which are fundamentally a change in governance, and sometimes a challenge to different forms of governance — without the corresponding violence and destruction?

PF: This international arbitration system that I mentioned, there is violence in a sense. A country can seize assets that are domiciled there, and they have the power of state-backed violence to do that with, and it’s in everybody’s interest to keep the system running. It’s what enables people and companies and countries to invest in places that they don’t trust, which is great for everyone. It’s all about at this stage, even though I want to change the existing system, at this early stage it’s about fitting in and finding allies. We’re not going to out compete their violence — we’re too small and too few, that would be ludicrous.

CB: Comparing islands versus ships — mobile jurisdictions — what are the respective advantages of both?

PF: This one’s pretty clear under the laws of the sea, which is one of the largest multilateral treaties in the world, and it’s basically the downside to being a mobile vessel. The ocean is a difficult, expensive operating environment. Everything is way more expensive. A boat is a hole in the water you pour money into, that’s the nature of it. This is seasteading’s motivation, and what you get in return for that is a remarkable legal environment.

The world came up with a jurisdiction system for a world of mobile little towns — the flag system, where a vessel has to register with a country and fly their flag, and then that vessel is under the law of that country, and they’re out past 12 miles. There are exceptions for taking resources, like fishing or oil, or any of those things are still owned by the coastal state out 200 miles. The ships passing over don’t get the right to take the fixed resources. But as far as like, what is the legal environment on the ship? It’s the flag.

It’s a bit like a floating embassy, but run by a third party. You’re franchising the sovereignty of the country whose flag you fly, and that’s a business that a bunch of countries engage in. And as a result, you can renew your flag registration annually, and there doesn’t need to be any association between the ship and the flag state, except that they’ve registered and so you have this virtual, consensual association between these mobile assets, ships, and the nation state, such that that forms a really competitive market. Unlike land, where “this country owns this, we’re claiming these fixed resources”— it’s not a competitive market. You can’t enter it. The cost of switching in terms of land is really high. But because any country can serve any ship, and ships can easily switch between different flagging providers, it’s actually this incredible competitive market for franchise and sovereignty, which, if we had that for land, I’d be done. I would do something with AI or whatever.

Being mobile on the ocean gets you something almost anarcho-capitalistic — not quite, but in the sense that you choose your service provider, you choose whose sovereignty you’re using and whose laws you’re under. You have to pick someone, and they have to be a country, and they have to say yes, but there’s a competitive market. The downside is the cost and the upside is this incredible legal environment. I work on charter cities because that cost downside is just too big, and we can get our protection in other ways, and that seems like a much better bet to me, but that’s the trade-off of going mobile.

CB: There’s another side of it which doesn’t really involve human beings at all — just parking assets in this mobile jurisdiction. A company, for instance. What would be the arguments against that?

PF: It’s just more expensive to do stuff on the ocean, so there has to be a reason.

CB: When you talk about the sea tax, at what point does it become cost prohibitive?

PF: The example I generally use is: there’s one condo cruise ship called The World, which lost a lot of money for its investors. And the cheapest apartments on it were maybe at one point around $50k a year. So the OPEX was roughly 1/6 of the capex, which is insane, compared to a house. And that illustrates the operational costs of having a moving home. If you do the math, depending on the discount rate — say 5% — that high of an OPEX ends up actually being more expensive. The net present value is more expensive than the capex.

CB: In recent years there’s been a proliferation of special economic zones in special jurisdictions.

PF: There are not very many deep ones, though.

CB: What makes them either deep or shallow?

PF: I think of this from a software perspective. What percentage of the codebase do I own? What percentage of all laws and regulations do I get to write that apply to this area? Is the operator different from the rest of the country? In most SEZs, the difference is really small. They’re just patching a few parts of the code base to change taxes, tariffs, labor laws, capital repatriation. They tweak some things that are in the way of getting business done, so we’re gonna make an exception. For whatever reason, we can’t change the whole country, but we can make an exception for this area.

Anything that changes key constants, something like Honduras, Prospera, or another example I would use would be Dubai International Financial Center, or you could say Guernsey. In any of these jurisdictions, the percentage of things that are different is radically higher than an SEZ. In the case of Prospera, they’re under the Honduran constitution, treaties, and criminal law, and they basically get to write all the rest of the law — what’s called commercial law, which is about everything from medical, finances, zoning, labor, and everything else that’s not a crime. I call it 80% autonomy. It’s a made up number, but that’s my feel for it. The Dubai International Financial Centre is a corporate jurisdiction with completely different corporate and commercial law. Businesses incorporated there virtually get to be under this different legal system for all their corporate law and everything that flows from that. So that was a much deeper reform, that or Hong Kong. It had different rules and regulations than the UK. The percentage of things that were different was very high.

CB: There were supposed to be some emancipatory qualities to the information age — that was something widely forecast. It was widely anticipated that there would be a liberating quality to the Information Age. I’m not really seeing it, though. The business model is basically surveillance in order to sell advertising, but it could easily be converted, if political circumstances were to shift, into something oppressive. What do you make of this? Is it an authoritarian or a libertarian age?

PF: I had a conversation with a founder friend last week about AI and robotics and this question. I think that people who are into decentralization or personal freedom tend to look at technologies and see the ways in which the technology would help that — and completely ignore the ways in which the technology would contribute to the opposite. Take 3D printers: people said oh, it’s gonna decentralize manufacturing, but there’s economies of scale. Like, mass production is always cheaper than custom production. That hasn’t changed. Just customer production has gotten cheaper.

With the internet, there are ways it makes it much easier for people to communicate and transact privately and form cabals and all of these things and coordinate and form insurgent networks. And there’s all these ways in which it gives the state the power to surveil and control. The core error is that people don’t think through both sides and understand: Why is there centralization? And does this actually change? Does this actually shift the balance?

With the internet, you can’t generalize, right? The world is complicated. There’s ways in which the ability of people to form secure, unbreakable communication networks — to coordinate without being monitored, with sovereign-grade security — has never been higher. That capacity is very, very strong right now. It’s not perfect. Humans aren’t great at OpSec, but that’s real. We have crypto and there’s ways that people can transact and save money that’s difficult for nation states to stop. The internet has enabled an increased ability for individuals to anonymously contract without government oversight. And it’s enabled mass monitoring at scale, right? Giant databases — that’s real too.

What’s the net effect? I tend to think that the net effect is towards more freedom, because more and more stuff is moving into cyberspace, and the rules of cyberspace are different. It is a much more voluntary competitive environment: low barriers to entry, low cost of switching. There’s way more competition, way more giving people what they want. And more and more important things are happening there, and it does give the surveillance state tools. But I definitely think that on net, the movement of power and of things happening that matter into cyberspace is draining power from the nation state.

CB: There’s this famous Peter Thiel essay — I think it appeared in the same issue as your essay, for Cato. Thiel talks about, on the one hand, the fundamentalist and totalitarian state threat, and at the other end of the spectrum, the unthinking demos of mass democracy. There’s this concept of exiting. Is it possible even to really exit? Is it ethical to think in these terms?

PF: Can you fully exit? No. But can you exit in a meaningful way? Definitely, yeah. You know somebody who’s got three passports and lives in Dubai — not a great example this year — is still interconnected. You’re still gonna get fucked sometimes with geopolitics. But you can meaningfully exit -— not completely — but individuals have to evaluate for themselves whether their partial exits that are available are worth it, and understand that utopia is not an option. There’s no Gulch. You’re gonna be part of the nation state system, but it doesn’t mean that there’s no way to work that system and improve it.

On the ethics, 99 out of 100 times I see a legal or regulatory change, I feel like it’s shit that’s for special interests and doesn’t fit my ethics. But the US exit tax change that happened in the Patriot Act — I don’t know how the fuck this happened, but it’s proper. Before the Patriot Act, the IRS could tax you for 10 years. You leave, they keep taxing you for 10 years, and then you’re free. The Patriot Act said, look, it’s very simple. If you want to exit, all of your unrealized gains get taxed that you’re leaving with because you owed those taxes to us. They just haven’t hit the books yet, they haven’t been realized, and then you’re free. And that’s the perfect rule, right? You grew this wealth inside the system. And this is the clear and explicit deal. You’re here, we do these things. We collect the taxes. You made the money. We’re just going to collect our uncollected taxes, and then you’re good to go. I think that’s so perfect and so moral and ethical.

I think there’s a codependent instinct that says, ‘No, I’m gonna stay and try to change and fix things.’ That can be beautiful and noble. It’s a great instinct to have. And the question is: Are there actual routes to change that, which you have the power to enact, or that the system permits that will work? And there’s nothing wrong with that strategy, but there’s also nothing wrong with saying it’s okay to leave. It’s okay to break up with someone, it’s even okay to divorce someone. It’s okay to say, I’m going to live differently, I’m going to change my provider. I don’t think there’s anything immoral about leaving society. You didn’t promise that you were going to stay until you were dead and help America when you were born here. You were just born here. And we have a system of taxes and Social Security that’s trying to keep track to some degree. You might feel grateful, you might want to give back, but you’re not morally inclined to stay in a failing system. That would be fucking retarded.

CB: Do you think the Patriot Act represents a shift in American governance as profound as, say, the New Deal?

PF: It seems smaller, but it’s important. Obviously, it was also a shift to a surveillance state, which I don’t like, but it seems smaller. It’s big on the scale of the laws of the past 50 years.

CB: Since you began your seasteading journey, what have you discovered, in terms of material and in legal obstacles to this endeavor?

PF: In the seasteading days, the main obstacle was the cost and difficulty — the economics. You’re in this expensive environment, and you’re far away from infrastructure and trade routes and people. One thing that the broader movement has realized is what we call the cold start problem. A new city, or a new society, is sort of a contradiction because cities depend on network effects and economies of scale, and new things don’t have that. The question of, how do you get the first people to move? How do you get your first users onto a network-effect business? That’s front and center. The ocean makes it a lot harder. It’s hard enough to get people to move to Roatan and live in Prospera, on a beautiful Caribbean island that has water and electricity and internet. Getting them out of the ocean is even harder. The cold start problem is a big challenge, even in doing this stuff on land.

I don’t think there are really legal obstacles. A sovereign is allowed to delegate their rulemaking authority. We have the New York convention for enforcing deals on sovereigns. The legal stuff is there. The economic model is big. One bottleneck is that startups are hard — trying to start a new society as a startup is another level of hard. It’s already a very limited population that can successfully found a company that changes the world, and now we’re doing a super hard-mode company. It’s a very limited set of people who can handle that, and there’s very few of them. Those people are often billionaires or centimillionaires who are incredibly successful and can do anything they want. We need those incredible people — we need the Elons of the world — but the space is too new and small to attract them. I think as we prove it out, that’ll change, because starting a new country is ultra-ambitious. That has appeal.

The other bottleneck is the economic engine. A lot of these projects are dead in the water because we have all these fancy ideas about how better legal systems and better laws lead to faster GDP compounded growth rate over time. That’s all true, but it kicks in over decades. And when you’re starting one of these, you need an actual small-scale economic engine to make it economically viable in the near term. What will the jobs be? What are your competitive advantages versus the rest of the world? What are your resources, who are your anchor tenants? Making those economic engines is really challenging. We haven’t really seen a success in the space yet. It’s fundamentally necessary. I believe it’s possible.

CB: What are some incentives that could make this work from the economic standpoint?

PF: There’s not a general formula. It takes a great entrepreneur to look at a country — their resources and industries and legal system — and figure out what they can bring to the table. One of the pitches for Prospera is FDI: if you make a stable, familiar regulatory environment, then people will be more willing to invest there than they would in the host country. But it’s really product design, it’s business development. You’re creating this real estate-plus-government business. The features are a better regulatory system, whatever infrastructure I can fund and build, and the companies you can bring in. Then you ask: who wants this product?

A lot of my focus these days is on Africa with our company, Alpha City. It’s different per country. There are some things like agriculture, a huge part of the economy everywhere in Africa. And bringing in better tech and more modern methods and better management can improve quality of life and economic returns across the continent. Then there are country-specific opportunities. Countries have specific resources. Parts of East Africa have insane amounts of untapped geothermal energy, massive amounts of cheap geothermal energy there that’s perfect for data centers, AI training. This is how we monetize. It used to be that crypto mining was how you monetize cheap electricity. Now it means AI training.

Going back to what I said about regulatory differences playing out over decades — my original idea was to identify things in a country’s legal system that prevent certain businesses, that they won’t change at the national level, but they are willing to grant an exemption to us. Find the ones with the highest potential and build around them. But I think the more important work is the entrepreneurial legwork: the business development and entrepreneurial market research and customer analysis and finding partners and putting together deals, bringing together the capital, the jobs, the workers, that work that an entrepreneur needs to do is far more what it’s about. It’s your specific business development for your product.

CB: What happens as you get further and further away from the coastline of a sovereign state?

PF: At about 20 miles, if you’re not on a flagged vessel, you’re considered a pirate. Any nation’s warships can seize you and board you. That’s the category where you’re at the mercy of anyone under international law.

If you’re a proper flagged vessel, the first 12 nautical miles are territorial waters — . exactly like being in the country. Then from 12 to 24 nautical miles is the contiguous zone, where there are limited rights; if you’re chasing someone out of your territorial waters, you’re allowed to keep chasing them, but mostly the contiguous zone doesn’t really mean very much. Then there’s the exclusive economic zone, which is out to between 200 and 350 nautical miles depending on the depth of the continental shelf. A ship there is in international waters but the coastal state owns all the resources. They own the oil, the seabed resources, the fish, even solar panels to collect the sun, that’s theirs. All the fixed resources are owned by the coastal state out to that limit.

Beyond that are the true high seas. But the only reason that states haven’t claimed those resources is that nobody knows how to exploit them — it’s too expensive and too difficult. There are no resources there, and so nobody’s bothered to make a claim. When you’re out there, nobody owns the resources, but you mostly can’t access them anyway. If you want to fish in international waters, you’re allowed to do that, but nobody’s gonna make money fishing 200 miles offshore. Your ship is still flying this flag, and you’re under various international laws and treaties about safety and pollution — and subject to the flag state’s law, however much of it they choose to apply.

CB: What are the most permissive flagging states?

PF: Liberia, Panama. Flags of convenience are what they’re called. If you don’t make a big stink, they don’t care.

CB: Are you able to switch your flag?

PF: I am pretty sure it’s illegal to have multiple flags and switch between them, but you can change your registration annually. It’s illegal to fly a false flag.

CB: What legacy systems of governance do you think are set to expire in the near future?

PF: Look at the whole welfare state and the demographic pyramid. All the welfare state countries are going broke. Their systems were not well-funded enough. They were not flexible enough. And the people on the dole are never going to vote to decrease it — they’re single-issue voters. How the fuck you beat a large block of coordinated single issue voters? That time bomb that some people projected, it’s going off. I think that’s a huge one.

Control of currency is obviously being degraded by crypto. And back to the cyberspace thing — nation states can and do regulate cyberspace with some degree of effectiveness, but it’s just harder. Mathematically, technologically, it’s a freer place where the rules of physics are different, and more and more of our world is moving there. Personally, I focus on meetspace — network states and all that, sure, sounds good, do whatever you want in the cloud. But things already work pretty well there. My focus is, we’ve got to change the world by affecting the real world.

But it remains the case that the more stuff that moves online where nation states have less power, the less power they have. The largest companies are now bigger than small nation states. There are a lot of ways in which the legacy systems are failing as the world changes and they don’t.

CB: What becomes of “the unthinking demos” in this new world?

PF: I see a full spectrum of possibilities — from crazy abundance and UBI for people, because the robots are making so much stuff, to AI killing us all. It’s going to empower those with agency. But the unthinking demos depend on the systems around them to work and function, and those systems are failing. This is why the graphs of trust in our institutions, or the number of people here in Texas who are saying, “Fuck all of this, I want to grow my own food and be self-sufficient, the system is broken,” is growing. The number of people in the unthinking demos who are realizing they can’t just go along unthinkingly, because stuff is falling apart around them — that number of people who are forced to take the red pill by their circumstances is growing and growing. I think the unthinking demos is fucked, and they’re waking up to it.

CB: And do you foresee the completion of the Sovereign Individual thesis? You think there will be a caste of individuals that master these new technologies and become sovereign unto themselves?

PF: Nothing’s black or white. We’re already seeing people become much more sovereign. With technology, some minority of people will become more so, and that’s only going to increase as the leverage of tech increases, but it doesn’t mean the nation states are going to go away.

CB: Who’s becoming more sovereign and in what ways?

PF: The ability for people to make passive businesses, or optimize their health with AI, and peptides rather than depending on the medical system, educating themselves with online learning instead of depending on an educational system. There are more and more tools where people can self-serve parallel institutions. The people who take advantage of those, given that the mainstream institutions are failing, whether that’s media, health, education, business, dating — are getting much better results. But it doesn’t make them fully sovereign, right? It’s clawing back degrees of it. What does it even mean? We depend on each other. We depend on each other for production, and unless you’re the Chairman of the Joint Chiefs of Staff, there are always people with more guns than you. That’s the world that you live in. Sovereignty is a direction. It’s not a destination.

CB: On the microstate concept: for example, the Order of Malta is not a country, it’s not a state, but it’s a Sovereign Order within the Catholic Church.

PF: I consider it the only sovereign entity without a clear geographic center, without territory. They’re the only decentralized sovereign. It’s super cool. But they don’t seem interested in doing my type of cool stuff with it.

CB: How would you replicate that? How would you convince a country to accept a passport for a nonexistent state like they do?

PF: I’m a really big fan of this. What we’re doing with charter cities is not trying to replace the state, but find partners and plug into the nation state system. You’re licensing their sovereignty, right? So I think it’s all about finding a country willing to issue passports that maybe don’t confer citizenship and voting rights, but come with full background checks. If a country just chose to divorce the travel document more from citizenship, and partnered with a company that said, we’re going to sell as many $25,000 passports as we can and make sure holders have been cleared to the same or better standards as other countries — I think that’s already an accepted model. Use what’s already there. And over time, people will come to understand that a travel document authority is really different from a nation state — you want a much narrower set of things from it. It doesn’t need to be the same. We’ll transition at some point.

CB: The sovereign response to the Covid-19 pandemic was a remarkable thing. For you, what are the lessons from what happened?

PF: It was such a huge tailwind for my movement. It showed people how bankrupt the legacy systems were — how fucked up they were — in a way which became vivid and real to a lot of people. It led them to be like, “wait, there are a bunch of other things that aren’t working. Let me look into this. Oh my god, everything’s actually falling apart.” So it’s a huge tailwind to the parallel institution movement. You can’t just expect the future to be like the past. Sometimes their systems are failing, but you haven’t realized it yet, because it hasn’t been made clear to you, or they haven’t been subject to the right stressors, and you should be more careful about what you trust. Today’s governments are not effective decision making machines. Let’s do something different.

For the authoritarians, there’s a public good problem here where if you misuse authority, everybody uses that as ammunition against authority. But you are getting the benefits of exercising your authority, and states are not unitary or rational actors at all. That’s the problem. If they were going to learn a lesson, they should learn lessons about what rights people care about and what impact it has on the trust of your citizens to violate those, and how badly some of our institutions have been allowed to decay. We needed them, and they weren’t ready. Keep the departments running smoothly. Be prepared for uncertainty, because there’s just going to be more of it.

The world is changing faster and faster. We are in the AI soft takeoff now. It’s 2026. The singularity as an abstract future idea that people argued about — that ended in the last year, I would say. It is happening. I’ve been using agentic AI with OpenClawfor six weeks and it’s changing my life. I’m seeing more and more people pick up these AI tools and accelerate their growth and improve their health. It’s not a hard take off where we’re suddenly fighting Skynet, but this is what the soft takeoff looks like. We’re in it. That means more uncertainty, more need for resilience, and more need for systems that are responsive and functional and adaptive, because the future is coming faster and faster, and it’s starting to come really fast now. That’s not going to let up for a while.

This interview has been lightly edited for clarity and length

That idea - of turning life into an epic - carries through to Ulysses, a young company in San Francisco developing autonomous underwater robots. "Ulysses is the Latin for Odyssey," Will O'Brien tells me, "which is also a maritime quest." Homer's Odyssey follows the king of Ithaca, Odysseus, at the end of the Trojan War, on a ten year return voyage across the Mediterranean Sea to his home kingdom, during which he and his crew of roughly 600 men, face the sea and all its perils - natural disasters, monsters, and enchantresses. The name stuck for a few other serendipitous reasons: the first Ulysses office in Dublin was situated between James Joyce's childhood home and the Museum of Literature Ireland - home to an original copy of Ulysses - with a statue of Joyce in the park across the street. "So, north, south, east, and west was Ulysses."

Their current headquarters in the South Beach neighborhood of San Francisco is one of the most colorful offices I've ever visited, teeming with life. The color blue runs through everything, and so do sharks: there are sharks on the walls, as stuffed animals on the floor and railings; small ones, medium sized ones, and one big enough to cover the ceiling of the ground floor cafe. There is a bookshelf in the shape of a wave stocked with reading material like Zero to One, textbooks on rocket propulsion and climate policy, and, of course, Ulysses.

On the ground floor, a fully equipped engineering lab flanks an open space. A map of San Francisco is plastered on the ceiling, ocean motifs adorn the walls and doors, and packed industrial shelves are arranged in rows. To the right opens a door to their 14,000 square foot factory, where you can find an industrial grade inflatable pool set up in front of a massive flag of the US, manufacturing machines with engineers at work, multiple desks, and modules of their hero product in varying stages of development. A Mako is being tested in the pool, with cable attachments to a rugged laptop and a Playstation controller. "You'd be surprised by how much of the world's GDP depends on Playstation controllers," Will tells me.

As I sit down with CEO Akhil Voorakkara, he brings me water - in a Guinness glass. "Jamie is the only non-Irish co-founder, he's from Scotland," he explains. The four founders - Akhil , President Will O'Brien, CTO Jamie Wedderburn and COO Colm O'Brien (co-founders O'Brien and O'Brien are not related) - founded Ulysses Maritime Technologies in Dublin in the summer of 2023.

"It started with me and Jamie," Akhil tells me. "We both worked at the same company straight out of college, a drone delivery startup in Ireland called Manna. I was an electronics engineer there and he was a mechanical engineer there. And we had a lot of fun working together until that job wasn't fun. It got really political and messy." Akhil ended up in a consulting role at McKinsey and Jamie went back to Scotland to work as an engineer at a space data analytics company called Spire.

"18 months later, we were both pretty bored where we were. And Jamie hits me up out of the blue: 'I'm thinking about doing a startup.'" The two first entertained the possibility of doing another drone startup, considering their experience and know-how. The idea they settled upon eventually sprang from the complaints of a friend of Jamie's, a marine biologist, about the cost and manual labor of doing field work in the ocean.

"It's just pretty hard to do things in the ocean effectively, even on a small scale - let alone on a large scale," Akhil explains. The friend's woes were exacerbated by "miserable Scottish weather, having to wade into this intertidal area. Very cold, very wet." Jamie was perplexed: why were there no tools to do this? Why were a group of 50 human beings going out into the water to perform what he called a "relatively simple task" of seagrass restoration?

Seagrasses are the only flowering marine plants - forming dense fields of long, narrow leaves in shallow coastal waters, resembling terrestrial grasslands - and are crucial to the ocean ecosystem. Seagrass meadows sequester about 10% of the ocean's carbon and provide a thriving habitat for wildlife, with just one hectare able to sustain around 80,000 fish and 100 million tiny invertebrates. A typical day of fieldwork in seagrass restoration lasts eight to ten hours - marking plots in shallow coastal sites, measuring environmental conditions, manually planting seagrass shoots or scattering seeds, securing them in the sediment while working against tides and murky water, all while documenting the process with photos and water quality samples.

Jamie's friend's team were transplanting shoots they had harvested elsewhere to restore their site's seagrass. "It was really expensive. You're talking hundreds and thousands of dollars for a hectare, which is a 100th of a square kilometer - which is tiny. And then hundreds of man hours of planning and labor. The economics of it were just insane." Akhil compared the lack of development in sea stewardship to the efficiency and ease of modern day agricultural farming, which humanity has refined over more than 10,000 years. "We could do drones, but underwater robots are so much cooler." The other realization was that "there's no one actually building tools to operate on a massive scale in the ocean," despite the technology to do so already existing.

The ultimate factor that bolstered Akhil's faith in the sea robots idea was how each of his future co-founders lit up at the mention of the idea "I'm not crazy, really smart people, people that I respect were also getting excited about it."

"It was just sufficiently weird," says Will. Colm, who was working as an Aerodynamics Engineer at Red Bull Racing and was about to join Formula One - "literally a dream job" - was also in.

Akhil was the one who knew each of the boys well before Ulysses came together. He had been friends with Colm since they were 12; Colm's father ran a coding club for children in Dublin, where both Colm and Akhil later taught. While at Dublin City University, Akhil founded the Irish Student Hyperloop team, and Colm took over as leader in the second year of the project. It was there that they met Will, who reached out asking, "How can I help?". Jamie and Akhil met later, working together at Manna after graduation. Coincidentally, Will was working in the same office building at the same time at a startup called at Zipp, the first bike sharing app in the UK and Ireland. Akhil then assembled the group. He recalls thinking: "me, Colm, Jamie, Will. We can literally build the ocean company. I'm making the group chat now."

Jamie and Akhil started 3D printing "contraptions of a robot that couldn't swim but could operate under water and deposit seagrass seeds," with Colm working on designs. "It was just pure excitement," Jamie recalls.

I ask where they sourced the 3D printers from, "I just had them", answers Akhil. "I've had a 3D printer of some form or the other in my bedroom since I was 15." He ran a small online business while in high school - clients would meet him at local shopping centers with cash in exchange for the trinkets he 3D printed for them. Another project was a small-scale self-driving car experiment, "I 3D printed a bunch of them [small cars], and then I put radio communications in them, so they would talk to each other and drive around the course and not crash into each other. They crashed into each other, but it was really fun."

As Jamie flew in from Scotland, the boys convened at Will's apartment in Dublin the next weekend, bringing bags of 3D printed parts to assemble. "There were no tools in the house. Not even a ruler." So they spent the weekend on their laptops, ironing out non-hardware details. "Will fleshed out the business plan, did customer research, and then Colm and Jamie were scheming on whiteboards. One thing Will did have was a whiteboard." The next morning they all headed to Akhil's apartment. "We spent the next day on the living room floor of my house and were just gluing these things together."

In just a weekend, the robot they had assembled could do the basic job of planting seeds into the ground - they tested and filmed a demo of the rudimentary machine in Dublin Bay's murky waters. The Ulysses team had gone from an idea to a robot that could pass as autonomous, and they had also proven they worked well together as a team.

Will and Akhil spent the next couple of weeks making a deck, a data room, speaking to customers, and reaching out to investors - including European ones. "The risk appetite in Europe is just not the same as here. The appetite for ambition is nothing compared to here," Akhil told me, gesturing at San Francisco.

Their dream investor was Lowercarbon Capital. Will met someone at an event who gave them an email address and promptly sent over their pitch deck. Akhil recalls that he received an email with some questions while walking to dinner with his then fiancée, Julia. "I had to stop and was like, 'I'm gonna go to the car for 20 minutes, and type a response to the email.'" In the first week of September, 2023, they registered Ulysses as a Delaware C-Corp, quit their jobs, and flew to San Francisco. "We landed here on the 15th of September, went back on the 20th of October, and the round was closed."

Their $2 million pre-seed round was led by Lowercarbon, with participation from Superorganism, Regen Ventures, and angel investors. In their month-long fundraising sprint, Akhil recalled, "we had this prototype and we had this tub filled with water and we were driving around San Francisco. And every time we went over a bump - the road quality is not good here - the water would splash around in the back of this rental car."

"Our first underwater vehicle, we called it the Mako. What we're selling today is still the Mako," Akhil jokes. The name for their first product has stayed the same, but the Mako has evolved into something far more versatile than the Ulysses founders had first anticipated.

Named after a species of shortfin shark known for its agility - small and fast - the Mako is a family of underwater autonomous vehicles built on a modular architecture: each section serves a specific function that can be swapped or extended depending on the mission. The nose cone houses acoustic and visual sensors; the "brain" module contains the compute and software that guides the vehicle. Power comes from a 2kWh battery that can be stacked, and a universal payload connector allows integration of any sensor or robotic tool. Made mainly of aluminum and carbon fiber, the Mako runs without external cables - though users can attach a cable for live viewing if desired. The six-foot-eight vehicle can work for up to 72 hours without surfacing or recharging, dive to 5,000 feet, and carry a 200-pound payload. It can precision plant seagrass, map the seafloor, and monitor and scan underwater habitats. Aided by Ulysses' technology, the Mako is able to run a data-center class GPU on board. A launch, recovery, and recharge system named Kraken and an autonomous surface mothership called Leviathan are currently under development. The Mako, Akhil says, "punches way above its weight. And it's also a cool name."

Ulysses did a million dollars in revenue planting seagrass in their first year, and realized along the way that the Mako could do a lot more; inspect and survey maritime infrastructure - pipelines, telecoms, cables, LNG terminals, ports, and bridges. And, being a robot, it can fix, maintain, and even install said infrastructure.

"We knew how big the opportunity was, but we didn't know how easy it would be for a product that we were building for one market to serve another market." Only through doing the field work did the team realize "how similar the attachments are, how similar the robots need to be, and how similar the payloads that needed to be carried. Even the software that runs the robots" - Ulysses' proprietary software - "that you'd use to go and survey seagrass is the software that can scan the Strait of Hormuz for mines." Akhil stresses that expanding the scope of the Mako was less about updating the product than discovering the full extent of their market. He explained: "The more time we spent in the water, the further we could see."

I ask how Ulysses would deploy at the Strait of Hormuz today. (We spoke at the height of the 2026 US-Iran conflict, with the Strait dominating most defense and tech conversations.) Akhil describes today's systems as fragmented and labor-intensive: "Two people have to put them [machines] in the water, then go out and do a scan, then they review the data, and then send out other vehicles to neutralize the mines." The method limits coverage to small patches at a time and depends on expensive, aging equipment that "cost millions of dollars apiece." The Mako, by contrast, can detect and neutralize threats in a single mission at a far lower cost - enabling far greater numbers in the water. Its lightweight design and aluminum hull give it a low acoustic and magnetic signature, allowing it to approach and disable mines while remaining undetected. Autonomous mothership and launch-and-recovery systems mean that the vehicle can remain on station for extended periods, continuously monitoring and responding. The result is a system designed to "see, decide and act" in real time - covering "the whole Strait" rather than of isolated sections patrolled intermittently by high-value assets.

Such a shift in cost and autonomy is what enables a fundamentally different security model for the ocean at large. Even outside of geopolitical flashpoints, "the ocean is like the wild West," Will explains - piracy, murder on the high seas, illegal fishing. Ulysses' solution is a dense, distributed presence: the goal is to become "the world's largest private navy," with "millions of robots in the world's oceans keeping them safe, healthy, and prosperous."

The company now serves commercial and defense customers well beyond its origins in conservation. "Immediately after deciding to branch out from seagrass, we closed a $2 million contract with the California Energy Commission for offshore wind monitoring and maintenance - which is huge," Akhil recalls. They soon after signed their first contract with the US Navy. Jamie believes Ulysses was well-positioned to meet an increase in the Navy's demand for high volume surface and subsea platforms - a demand driven largely by the war in Ukraine.

Ulysses designs, manufactures, and writes the software for everything in house. "All the manufacturing is right here in this office - 3D printers, CNC machines, PCB assembly machines. Everything is in house," Akhil says.

Making their own motors was a recent breakthrough. They were using "semi-custom ones from China," but ran into problems: "Lead time is terrible, and they're not actually very good motors," Jamie explains, with suppliers failing to grasp key requirements like depth sealing. Off-the-shelf alternatives for subsea applications weren't viable either - they cost "a bajillion dollars" - leaving Ulysses caught between low quality and extremely high cost. Taking the initiative, Ulysses' 13 engineers whipped up novel prototypes of motors within a few months.

I ask if it's a lack of bureaucracy that enables this "move fast, create things" culture at Ulysses. "A hundred percent," Jamie replies. "We exclusively hire engineers that are comfortable with a lot of autonomy and a lot of ownership," Colm adds. "It's hard to get motors that are made in America. And it's even harder to get subsea motors made in America." They have ordered some from China for inventory - but Ulysses has built their own in the time it is taking for them to arrive.

I ask them why Ulysses was able to produce their own motors, when others haven't. "It's not that others can't, it's that others won't," quips Akhil. "A lot of people are afraid to vertically integrate. Whereas we're maybe a bit very retarded sometimes: let's just put it in the water and see what happens."

"There's just no other option," Jamie adds. Will recounts that while fundraising, "a software investor in our seed round started asking questions - 'Why can't you just buy it off the shelf? We have this other company that does it like this.' And I don't think I've ever seen these two guys [pointing to Jamie & Colm] more outraged in their lives."

"If you're just resorting to buying everything off the shelf," Colm says, "skill issue."

Ulysses is also building their inertial navigation system (INS) in-house, engineering sensors that cost "50x less" without compromising on positional accuracy. This bespoke navigation system gives Mako a "sense of position" when GPS isn't reliable - as is typical underwater.

"We have a very clear picture of what we want to build now. We started off with this vision of the low cost modular underwater vehicle, the deployment system" - the Mako - "and that was just based on common sense reasoning," Jamie says. "We invented the tech and now we've found that what we're building is right for the industry. The only thing that keeps me up is that people want to buy it right now. And so we need to build it as quickly as possible. But we need to do it well."

Today, Ulysses is at their sixth iteration of the Mako, and preparing for first unit sales to defense customers.

After raising their pre-seed in San Francisco, the team headed back to Dublin to develop and build the company while waiting for their US visas. Jamie commuted from Glasgow to Dublin and lived in the office. "We initially thought it would take us six months to get the visa, but it ended up being a year. So I was stuck in this mouldy office for a whole year. It was also tough only getting to see Tilly" - his fiancée - "on the weekends."

"April 2024," when the team was still in Dublin, "we got our first employee." Ulysses brought on Dylan Cannyghin, another childhood friend of Akhil's, who became the company's founding engineer.

"And then, one by one, we got our visas sorted, and then it was time to move." In September 2024, the team made the transition to San Francisco. "Julia and I were planning on getting married, and we had to accelerate the timeline," Akhil says - Julia eventually had to quit her job at Intel to make the move. Jamie's fiancée, Tilly, transferred her job at Novotech to the San Francisco office. "Big commitments from everyone.".

A major customer deployment was awaiting the young company in Australia the following November. The move had been "super disruptive" and when the Ulysses team arrived Down Under, "a lot of stuff just didn't work." It was the company's first stress test - and they were failing it.

"That was deeply traumatic," Jamie exhales. "The setup in these operations is there's you and the customer on the boat with your robot. So, there's no covering up that your robot's broken. So you're there scrambling to fix it. You can look in their [the customer's] eyes and they look disappointed."

The team was supposed to plant Posidonia, a type of slow-growing seagrass found in Australia and the Mediterranean, during a narrow seasonal window when temperature, growth cycle, and seed viability align. "It was just hubris," Jamie explains, "We collectively had this delusion; oh, we're great engineers, and all you're doing is putting a seed in a tube and pitching into the ground. Must be easy." What they found was that while they had been perfecting the core mechanics of the Mako, they had neglected system level integration like control, communication, and robustness. Once in the field, failures stacked up: "it just started dissolving, our tether broke, connectors had problems - it was just a shocker."

The mission failed. Colm explains the mechanics of it: "it had an electrical short and it was shorting against the salt water, essentially started dissolving its own anodization. So its face was melting away. Because of the electricity, we called it Benji - Benjamin Franklin." The client was angry. And this wasn't a problem they could iterate around in a week. The next Posidonia seeding window was exactly a year away. "You only really get one shot per year," explains Colm. It took until April 2025 - months - for the team to recover mentally and spiritually. "We basically overpromised and underdelivered. It was our first reality check," Akhil explains. But, the entire team agrees, it forced a reckoning: it was only afterwards that the co-founders defined their individual roles and niches within the company. "It was a big reset. As founders, we had a lot of stuff to work through." Most importantly, they did eventually return to the Australian coast after a year - and succeed, "that was pure redemption," Jamie says. "Everything worked."

In May 2025, Ulysses closed an eight million dollar seed round - nearly a year and a half after their pre-seed - moved into a bigger headquarters, and grew from five people to sixteen. A few weeks after these interviews were conducted, Ulysses announced a Series A fundraise of $38 million, led by a16z.

What made the fundraise work, and what has kept Ulysses moving fast, is the balance of the founding team. "We are such a stacked team," Akhil explains, "It wasn't one co-founder or two. It was four." I ask him what each founder excels at. "Will, he could sell sand in a desert. He's just an incredible communicator. He loves deals, a natural born salesperson." Will put Ulysses on the map in San Francisco. The company's December 2025 launch party - the Ulysses Rodeo - made "top holiday events in the Bay Area" lists and offered guests a chance to ride a mechanical shark. In his spare time, he serves on the leadership for the Hamilton Society, a San Francisco debate society that is attended by local parishioners (the debates are hosted at Star of the Sea Church) and tech billionaires alike. Will informed me that he was even invited to the White House for Saint Patrick's Day.

Jamie and Colm, both mechanical engineers, form Ulysses' engineering core, "right from the birth of it, the company had so much engineering bandwidth," Akhil boasts.

I ask the two if it's challenging to share a skill set on one leadership team. "I think we argue the least," Jamie says. Colm elaborates: "We get in front of a whiteboard and just scheme something out. And to any observer, it would look like harebrained lines on a chalkboard, completely meaningless. But we understand what's going on." Jamie views their engineering work as a form of art - not illustrating an emotion, but making beautiful, perfect systems work. Colm's definition of perfect is straightforward: "If the thing you're making doesn't function in the real world, and can't function for a fucking long time, and isn't satisfying to use, then it's not perfect."

That philosophy directly shaped Ulysses' approach to the incumbent underwater vehicle industry, where Jamie believes companies don't actually build systems - they buy components from different suppliers and assemble them. "All they do is put it in a tube. And have a rat's nest of cables in it." Their competitors, he says, put massive markups on aggregated components: "it would cost us a little over $300K - and they retail them for $15 million. That's just a complete joke."

Ulysses' applications are straightforward - seafarming, underwater mine detection, surveillance - answers to problems that have existed for decades. The robots are novel, but conceptually simple and elegant. And yet the work Ulysses has cut out for itself in building and deploying their robots at scale is immense.

From building critical systems in-house to uncovering viable use cases, Ulysses has had to confront a domain that remains largely uncharted.

According to Jamie, the lack of innovation in subsea technology is rooted as much in history and culture as in engineering. During the Cold War, while the Space Race captured public imagination, equally advanced work in the ocean remained hidden: "there was a huge amount of money going into subsea, but it was top secret." Rockets and fighter jets were visible, celebrated, and aspirational; underwater systems, by contrast, operated in silence, shaping a talent pipeline where "all the best engineers, they graduate college and they want to work at SpaceX," with very few even considering nuclear subs.

"People used to talk about the undersea domain in the same breath as space," Will tells me. "In fact, JFK, in the same year that he increased funding for the Apollo programs, he also funded the first undersea habitats - SEALAB I, II, III. There were about 60 of these habitats all around the world. We were sending aquanauts there to live underwater." Then, months before Armstrong landed on the moon, aquanaut Berry L. Cannon died in a tragic accident while deployed on the SEALAB III. The Vietnam War followed, and funding for ocean exploration faced considerable cuts, leaving the sea a neglected frontier.

The structure of the subsea market reinforced stagnation. Unlike aviation or space, where large commercial demand drives competition and cost reduction, the ocean economy has been dominated by oil, gas, and defense - sectors where, Jamie believes, "they don't care about cost, all they care about is functionality... and cash." Incumbents operated with little pressure to innovate, "taking everyone for a ride." The result is an industry that never developed the same culture of iteration or scale. Now, in Will's view, that is beginning to shift: advances in autonomy, EV batteries, and satellite connectivity are creating conditions for a new wave of ocean innovation - doing for the sea what SpaceX did for space.

"All the AI stuff," Will says -"we need more energy, we need more cables, we need more critical minerals. All of these supply chains rest on the ocean, in a very serious way." His argument is a simple chain: the ocean is going to be an increasingly critical resource this century; robots will do most of the essential work of the future; and companies will build and operate these robots. Put those three together, and the opportunity becomes clear: unlike land-based robotics, which already has many competitors, nobody is building a company to operate robots in the ocean, despite the fact that it covers two-thirds of the planet and underpins much of human civilization, from protein sources to commerce.

"There's no company like Ulysses. There's nobody that's doing commercial, nature and defense. There's a $100 billion market waiting for us - and we want it all." Ulysses, then, is a frontier exploration company, one that has chosen the ocean as its primary domain.

That idea - of turning life into an epic - carries through to Ulysses, a young company in San Francisco developing autonomous underwater robots. "Ulysses is the Latin for Odyssey," Will O'Brien tells me, "which is also a maritime quest." Homer's Odyssey follows the king of Ithaca, Odysseus, at the end of the Trojan War, on a ten year return voyage across the Mediterranean Sea to his home kingdom, during which he and his crew of roughly 600 men, face the sea and all its perils - natural disasters, monsters, and enchantresses. The name stuck for a few other serendipitous reasons: the first Ulysses office in Dublin was situated between James Joyce's childhood home and the Museum of Literature Ireland - home to an original copy of Ulysses - with a statue of Joyce in the park across the street. "So, north, south, east, and west was Ulysses."

Their current headquarters in the South Beach neighborhood of San Francisco is one of the most colorful offices I've ever visited, teeming with life. The color blue runs through everything, and so do sharks: there are sharks on the walls, as stuffed animals on the floor and railings; small ones, medium sized ones, and one big enough to cover the ceiling of the ground floor cafe. There is a bookshelf in the shape of a wave stocked with reading material like Zero to One, textbooks on rocket propulsion and climate policy, and, of course, Ulysses.

On the ground floor, a fully equipped engineering lab flanks an open space. A map of San Francisco is plastered on the ceiling, ocean motifs adorn the walls and doors, and packed industrial shelves are arranged in rows. To the right opens a door to their 14,000 square foot factory, where you can find an industrial grade inflatable pool set up in front of a massive flag of the US, manufacturing machines with engineers at work, multiple desks, and modules of their hero product in varying stages of development. A Mako is being tested in the pool, with cable attachments to a rugged laptop and a Playstation controller. "You'd be surprised by how much of the world's GDP depends on Playstation controllers," Will tells me.

As I sit down with CEO Akhil Voorakkara, he brings me water - in a Guinness glass. "Jamie is the only non-Irish co-founder, he's from Scotland," he explains. The four founders - Akhil , President Will O'Brien, CTO Jamie Wedderburn and COO Colm O'Brien (co-founders O'Brien and O'Brien are not related) - founded Ulysses Maritime Technologies in Dublin in the summer of 2023.

"It started with me and Jamie," Akhil tells me. "We both worked at the same company straight out of college, a drone delivery startup in Ireland called Manna. I was an electronics engineer there and he was a mechanical engineer there. And we had a lot of fun working together until that job wasn't fun. It got really political and messy." Akhil ended up in a consulting role at McKinsey and Jamie went back to Scotland to work as an engineer at a space data analytics company called Spire.

"18 months later, we were both pretty bored where we were. And Jamie hits me up out of the blue: 'I'm thinking about doing a startup.'" The two first entertained the possibility of doing another drone startup, considering their experience and know-how. The idea they settled upon eventually sprang from the complaints of a friend of Jamie's, a marine biologist, about the cost and manual labor of doing field work in the ocean.

"It's just pretty hard to do things in the ocean effectively, even on a small scale - let alone on a large scale," Akhil explains. The friend's woes were exacerbated by "miserable Scottish weather, having to wade into this intertidal area. Very cold, very wet." Jamie was perplexed: why were there no tools to do this? Why were a group of 50 human beings going out into the water to perform what he called a "relatively simple task" of seagrass restoration?

Seagrasses are the only flowering marine plants - forming dense fields of long, narrow leaves in shallow coastal waters, resembling terrestrial grasslands - and are crucial to the ocean ecosystem. Seagrass meadows sequester about 10% of the ocean's carbon and provide a thriving habitat for wildlife, with just one hectare able to sustain around 80,000 fish and 100 million tiny invertebrates. A typical day of fieldwork in seagrass restoration lasts eight to ten hours - marking plots in shallow coastal sites, measuring environmental conditions, manually planting seagrass shoots or scattering seeds, securing them in the sediment while working against tides and murky water, all while documenting the process with photos and water quality samples.

Jamie's friend's team were transplanting shoots they had harvested elsewhere to restore their site's seagrass. "It was really expensive. You're talking hundreds and thousands of dollars for a hectare, which is a 100th of a square kilometer - which is tiny. And then hundreds of man hours of planning and labor. The economics of it were just insane." Akhil compared the lack of development in sea stewardship to the efficiency and ease of modern day agricultural farming, which humanity has refined over more than 10,000 years. "We could do drones, but underwater robots are so much cooler." The other realization was that "there's no one actually building tools to operate on a massive scale in the ocean," despite the technology to do so already existing.

The ultimate factor that bolstered Akhil's faith in the sea robots idea was how each of his future co-founders lit up at the mention of the idea "I'm not crazy, really smart people, people that I respect were also getting excited about it."

"It was just sufficiently weird," says Will. Colm, who was working as an Aerodynamics Engineer at Red Bull Racing and was about to join Formula One - "literally a dream job" - was also in.

Akhil was the one who knew each of the boys well before Ulysses came together. He had been friends with Colm since they were 12; Colm's father ran a coding club for children in Dublin, where both Colm and Akhil later taught. While at Dublin City University, Akhil founded the Irish Student Hyperloop team, and Colm took over as leader in the second year of the project. It was there that they met Will, who reached out asking, "How can I help?". Jamie and Akhil met later, working together at Manna after graduation. Coincidentally, Will was working in the same office building at the same time at a startup called at Zipp, the first bike sharing app in the UK and Ireland. Akhil then assembled the group. He recalls thinking: "me, Colm, Jamie, Will. We can literally build the ocean company. I'm making the group chat now."

Jamie and Akhil started 3D printing "contraptions of a robot that couldn't swim but could operate under water and deposit seagrass seeds," with Colm working on designs. "It was just pure excitement," Jamie recalls.

I ask where they sourced the 3D printers from, "I just had them", answers Akhil. "I've had a 3D printer of some form or the other in my bedroom since I was 15." He ran a small online business while in high school - clients would meet him at local shopping centers with cash in exchange for the trinkets he 3D printed for them. Another project was a small-scale self-driving car experiment, "I 3D printed a bunch of them [small cars], and then I put radio communications in them, so they would talk to each other and drive around the course and not crash into each other. They crashed into each other, but it was really fun."

As Jamie flew in from Scotland, the boys convened at Will's apartment in Dublin the next weekend, bringing bags of 3D printed parts to assemble. "There were no tools in the house. Not even a ruler." So they spent the weekend on their laptops, ironing out non-hardware details. "Will fleshed out the business plan, did customer research, and then Colm and Jamie were scheming on whiteboards. One thing Will did have was a whiteboard." The next morning they all headed to Akhil's apartment. "We spent the next day on the living room floor of my house and were just gluing these things together."

In just a weekend, the robot they had assembled could do the basic job of planting seeds into the ground - they tested and filmed a demo of the rudimentary machine in Dublin Bay's murky waters. The Ulysses team had gone from an idea to a robot that could pass as autonomous, and they had also proven they worked well together as a team.

Will and Akhil spent the next couple of weeks making a deck, a data room, speaking to customers, and reaching out to investors - including European ones. "The risk appetite in Europe is just not the same as here. The appetite for ambition is nothing compared to here," Akhil told me, gesturing at San Francisco.

Their dream investor was Lowercarbon Capital. Will met someone at an event who gave them an email address and promptly sent over their pitch deck. Akhil recalls that he received an email with some questions while walking to dinner with his then fiancée, Julia. "I had to stop and was like, 'I'm gonna go to the car for 20 minutes, and type a response to the email.'" In the first week of September, 2023, they registered Ulysses as a Delaware C-Corp, quit their jobs, and flew to San Francisco. "We landed here on the 15th of September, went back on the 20th of October, and the round was closed."

Their $2 million pre-seed round was led by Lowercarbon, with participation from Superorganism, Regen Ventures, and angel investors. In their month-long fundraising sprint, Akhil recalled, "we had this prototype and we had this tub filled with water and we were driving around San Francisco. And every time we went over a bump - the road quality is not good here - the water would splash around in the back of this rental car."

"Our first underwater vehicle, we called it the Mako. What we're selling today is still the Mako," Akhil jokes. The name for their first product has stayed the same, but the Mako has evolved into something far more versatile than the Ulysses founders had first anticipated.

Named after a species of shortfin shark known for its agility - small and fast - the Mako is a family of underwater autonomous vehicles built on a modular architecture: each section serves a specific function that can be swapped or extended depending on the mission. The nose cone houses acoustic and visual sensors; the "brain" module contains the compute and software that guides the vehicle. Power comes from a 2kWh battery that can be stacked, and a universal payload connector allows integration of any sensor or robotic tool. Made mainly of aluminum and carbon fiber, the Mako runs without external cables - though users can attach a cable for live viewing if desired. The six-foot-eight vehicle can work for up to 72 hours without surfacing or recharging, dive to 5,000 feet, and carry a 200-pound payload. It can precision plant seagrass, map the seafloor, and monitor and scan underwater habitats. Aided by Ulysses' technology, the Mako is able to run a data-center class GPU on board. A launch, recovery, and recharge system named Kraken and an autonomous surface mothership called Leviathan are currently under development. The Mako, Akhil says, "punches way above its weight. And it's also a cool name."

Ulysses did a million dollars in revenue planting seagrass in their first year, and realized along the way that the Mako could do a lot more; inspect and survey maritime infrastructure - pipelines, telecoms, cables, LNG terminals, ports, and bridges. And, being a robot, it can fix, maintain, and even install said infrastructure.

"We knew how big the opportunity was, but we didn't know how easy it would be for a product that we were building for one market to serve another market." Only through doing the field work did the team realize "how similar the attachments are, how similar the robots need to be, and how similar the payloads that needed to be carried. Even the software that runs the robots" - Ulysses' proprietary software - "that you'd use to go and survey seagrass is the software that can scan the Strait of Hormuz for mines." Akhil stresses that expanding the scope of the Mako was less about updating the product than discovering the full extent of their market. He explained: "The more time we spent in the water, the further we could see."

I ask how Ulysses would deploy at the Strait of Hormuz today. (We spoke at the height of the 2026 US-Iran conflict, with the Strait dominating most defense and tech conversations.) Akhil describes today's systems as fragmented and labor-intensive: "Two people have to put them [machines] in the water, then go out and do a scan, then they review the data, and then send out other vehicles to neutralize the mines." The method limits coverage to small patches at a time and depends on expensive, aging equipment that "cost millions of dollars apiece." The Mako, by contrast, can detect and neutralize threats in a single mission at a far lower cost - enabling far greater numbers in the water. Its lightweight design and aluminum hull give it a low acoustic and magnetic signature, allowing it to approach and disable mines while remaining undetected. Autonomous mothership and launch-and-recovery systems mean that the vehicle can remain on station for extended periods, continuously monitoring and responding. The result is a system designed to "see, decide and act" in real time - covering "the whole Strait" rather than of isolated sections patrolled intermittently by high-value assets.

Such a shift in cost and autonomy is what enables a fundamentally different security model for the ocean at large. Even outside of geopolitical flashpoints, "the ocean is like the wild West," Will explains - piracy, murder on the high seas, illegal fishing. Ulysses' solution is a dense, distributed presence: the goal is to become "the world's largest private navy," with "millions of robots in the world's oceans keeping them safe, healthy, and prosperous."

The company now serves commercial and defense customers well beyond its origins in conservation. "Immediately after deciding to branch out from seagrass, we closed a $2 million contract with the California Energy Commission for offshore wind monitoring and maintenance - which is huge," Akhil recalls. They soon after signed their first contract with the US Navy. Jamie believes Ulysses was well-positioned to meet an increase in the Navy's demand for high volume surface and subsea platforms - a demand driven largely by the war in Ukraine.

Ulysses designs, manufactures, and writes the software for everything in house. "All the manufacturing is right here in this office - 3D printers, CNC machines, PCB assembly machines. Everything is in house," Akhil says.

Making their own motors was a recent breakthrough. They were using "semi-custom ones from China," but ran into problems: "Lead time is terrible, and they're not actually very good motors," Jamie explains, with suppliers failing to grasp key requirements like depth sealing. Off-the-shelf alternatives for subsea applications weren't viable either - they cost "a bajillion dollars" - leaving Ulysses caught between low quality and extremely high cost. Taking the initiative, Ulysses' 13 engineers whipped up novel prototypes of motors within a few months.

I ask if it's a lack of bureaucracy that enables this "move fast, create things" culture at Ulysses. "A hundred percent," Jamie replies. "We exclusively hire engineers that are comfortable with a lot of autonomy and a lot of ownership," Colm adds. "It's hard to get motors that are made in America. And it's even harder to get subsea motors made in America." They have ordered some from China for inventory - but Ulysses has built their own in the time it is taking for them to arrive.

I ask them why Ulysses was able to produce their own motors, when others haven't. "It's not that others can't, it's that others won't," quips Akhil. "A lot of people are afraid to vertically integrate. Whereas we're maybe a bit very retarded sometimes: let's just put it in the water and see what happens."

"There's just no other option," Jamie adds. Will recounts that while fundraising, "a software investor in our seed round started asking questions - 'Why can't you just buy it off the shelf? We have this other company that does it like this.' And I don't think I've ever seen these two guys [pointing to Jamie & Colm] more outraged in their lives."

"If you're just resorting to buying everything off the shelf," Colm says, "skill issue."

Ulysses is also building their inertial navigation system (INS) in-house, engineering sensors that cost "50x less" without compromising on positional accuracy. This bespoke navigation system gives Mako a "sense of position" when GPS isn't reliable - as is typical underwater.

"We have a very clear picture of what we want to build now. We started off with this vision of the low cost modular underwater vehicle, the deployment system" - the Mako - "and that was just based on common sense reasoning," Jamie says. "We invented the tech and now we've found that what we're building is right for the industry. The only thing that keeps me up is that people want to buy it right now. And so we need to build it as quickly as possible. But we need to do it well."

Today, Ulysses is at their sixth iteration of the Mako, and preparing for first unit sales to defense customers.

After raising their pre-seed in San Francisco, the team headed back to Dublin to develop and build the company while waiting for their US visas. Jamie commuted from Glasgow to Dublin and lived in the office. "We initially thought it would take us six months to get the visa, but it ended up being a year. So I was stuck in this mouldy office for a whole year. It was also tough only getting to see Tilly" - his fiancée - "on the weekends."

"April 2024," when the team was still in Dublin, "we got our first employee." Ulysses brought on Dylan Cannyghin, another childhood friend of Akhil's, who became the company's founding engineer.

"And then, one by one, we got our visas sorted, and then it was time to move." In September 2024, the team made the transition to San Francisco. "Julia and I were planning on getting married, and we had to accelerate the timeline," Akhil says - Julia eventually had to quit her job at Intel to make the move. Jamie's fiancée, Tilly, transferred her job at Novotech to the San Francisco office. "Big commitments from everyone.".

A major customer deployment was awaiting the young company in Australia the following November. The move had been "super disruptive" and when the Ulysses team arrived Down Under, "a lot of stuff just didn't work." It was the company's first stress test - and they were failing it.

"That was deeply traumatic," Jamie exhales. "The setup in these operations is there's you and the customer on the boat with your robot. So, there's no covering up that your robot's broken. So you're there scrambling to fix it. You can look in their [the customer's] eyes and they look disappointed."

The team was supposed to plant Posidonia, a type of slow-growing seagrass found in Australia and the Mediterranean, during a narrow seasonal window when temperature, growth cycle, and seed viability align. "It was just hubris," Jamie explains, "We collectively had this delusion; oh, we're great engineers, and all you're doing is putting a seed in a tube and pitching into the ground. Must be easy." What they found was that while they had been perfecting the core mechanics of the Mako, they had neglected system level integration like control, communication, and robustness. Once in the field, failures stacked up: "it just started dissolving, our tether broke, connectors had problems - it was just a shocker."

The mission failed. Colm explains the mechanics of it: "it had an electrical short and it was shorting against the salt water, essentially started dissolving its own anodization. So its face was melting away. Because of the electricity, we called it Benji - Benjamin Franklin." The client was angry. And this wasn't a problem they could iterate around in a week. The next Posidonia seeding window was exactly a year away. "You only really get one shot per year," explains Colm. It took until April 2025 - months - for the team to recover mentally and spiritually. "We basically overpromised and underdelivered. It was our first reality check," Akhil explains. But, the entire team agrees, it forced a reckoning: it was only afterwards that the co-founders defined their individual roles and niches within the company. "It was a big reset. As founders, we had a lot of stuff to work through." Most importantly, they did eventually return to the Australian coast after a year - and succeed, "that was pure redemption," Jamie says. "Everything worked."

In May 2025, Ulysses closed an eight million dollar seed round - nearly a year and a half after their pre-seed - moved into a bigger headquarters, and grew from five people to sixteen. A few weeks after these interviews were conducted, Ulysses announced a Series A fundraise of $38 million, led by a16z.

What made the fundraise work, and what has kept Ulysses moving fast, is the balance of the founding team. "We are such a stacked team," Akhil explains, "It wasn't one co-founder or two. It was four." I ask him what each founder excels at. "Will, he could sell sand in a desert. He's just an incredible communicator. He loves deals, a natural born salesperson." Will put Ulysses on the map in San Francisco. The company's December 2025 launch party - the Ulysses Rodeo - made "top holiday events in the Bay Area" lists and offered guests a chance to ride a mechanical shark. In his spare time, he serves on the leadership for the Hamilton Society, a San Francisco debate society that is attended by local parishioners (the debates are hosted at Star of the Sea Church) and tech billionaires alike. Will informed me that he was even invited to the White House for Saint Patrick's Day.

Jamie and Colm, both mechanical engineers, form Ulysses' engineering core, "right from the birth of it, the company had so much engineering bandwidth," Akhil boasts.

I ask the two if it's challenging to share a skill set on one leadership team. "I think we argue the least," Jamie says. Colm elaborates: "We get in front of a whiteboard and just scheme something out. And to any observer, it would look like harebrained lines on a chalkboard, completely meaningless. But we understand what's going on." Jamie views their engineering work as a form of art - not illustrating an emotion, but making beautiful, perfect systems work. Colm's definition of perfect is straightforward: "If the thing you're making doesn't function in the real world, and can't function for a fucking long time, and isn't satisfying to use, then it's not perfect."

That philosophy directly shaped Ulysses' approach to the incumbent underwater vehicle industry, where Jamie believes companies don't actually build systems - they buy components from different suppliers and assemble them. "All they do is put it in a tube. And have a rat's nest of cables in it." Their competitors, he says, put massive markups on aggregated components: "it would cost us a little over $300K - and they retail them for $15 million. That's just a complete joke."

Ulysses' applications are straightforward - seafarming, underwater mine detection, surveillance - answers to problems that have existed for decades. The robots are novel, but conceptually simple and elegant. And yet the work Ulysses has cut out for itself in building and deploying their robots at scale is immense.

From building critical systems in-house to uncovering viable use cases, Ulysses has had to confront a domain that remains largely uncharted.

According to Jamie, the lack of innovation in subsea technology is rooted as much in history and culture as in engineering. During the Cold War, while the Space Race captured public imagination, equally advanced work in the ocean remained hidden: "there was a huge amount of money going into subsea, but it was top secret." Rockets and fighter jets were visible, celebrated, and aspirational; underwater systems, by contrast, operated in silence, shaping a talent pipeline where "all the best engineers, they graduate college and they want to work at SpaceX," with very few even considering nuclear subs.

"People used to talk about the undersea domain in the same breath as space," Will tells me. "In fact, JFK, in the same year that he increased funding for the Apollo programs, he also funded the first undersea habitats - SEALAB I, II, III. There were about 60 of these habitats all around the world. We were sending aquanauts there to live underwater." Then, months before Armstrong landed on the moon, aquanaut Berry L. Cannon died in a tragic accident while deployed on the SEALAB III. The Vietnam War followed, and funding for ocean exploration faced considerable cuts, leaving the sea a neglected frontier.

The structure of the subsea market reinforced stagnation. Unlike aviation or space, where large commercial demand drives competition and cost reduction, the ocean economy has been dominated by oil, gas, and defense - sectors where, Jamie believes, "they don't care about cost, all they care about is functionality... and cash." Incumbents operated with little pressure to innovate, "taking everyone for a ride." The result is an industry that never developed the same culture of iteration or scale. Now, in Will's view, that is beginning to shift: advances in autonomy, EV batteries, and satellite connectivity are creating conditions for a new wave of ocean innovation - doing for the sea what SpaceX did for space.

"All the AI stuff," Will says -"we need more energy, we need more cables, we need more critical minerals. All of these supply chains rest on the ocean, in a very serious way." His argument is a simple chain: the ocean is going to be an increasingly critical resource this century; robots will do most of the essential work of the future; and companies will build and operate these robots. Put those three together, and the opportunity becomes clear: unlike land-based robotics, which already has many competitors, nobody is building a company to operate robots in the ocean, despite the fact that it covers two-thirds of the planet and underpins much of human civilization, from protein sources to commerce.

"There's no company like Ulysses. There's nobody that's doing commercial, nature and defense. There's a $100 billion market waiting for us - and we want it all." Ulysses, then, is a frontier exploration company, one that has chosen the ocean as its primary domain.

By the Second World War, New York Ship employed 47,000 across a 160-acre site along two miles of the Delaware River. What started as a factory had become an entire economy built to power the largest and most productive shipyard on Earth. New York Ship delivered 670 merchant and naval ships during its life, including three nuclear submarines and even the first nuclear-powered merchant ship.

By the end of WWII, America’s fleet represented 50,000,000 deadweight tons — a measure of total carrying capacity — two-thirds of global tonnage. But after the war ended, the burden of upkeep for such a large fleet felt unnecessary, leading the US to immediately start mothballing and scrapping ships rather than maintaining them. The Cold War sustained demand for specialized naval work, like nuclear submarines and carriers, but that work concentrated in a select group of dedicated yards. Meanwhile, commercial shipbuilding migrated to lower-cost foreign yards. Caught between a shrinking commercial market and increasingly specialized naval contracts, New York Ship didn’t have the business to stay afloat. By 1968, it had shut its doors. Camden, the city that powered New York Ship, tracks an eerie road map for what happens when industrialism falters. The city lost a third of its population, including 30,000 manufacturing jobs. It now faces a poverty rate of 28.5% and has become one of the most dangerous cities in the country.

The New York Shipbuilding Corporation is just one of dozens of examples of the cost of assuming that America no longer needed a strong industrial base. Particularly in the 1980s, the US came to the conclusion that manufacturing and shipbuilding were commodity processes better suited for lower-cost labor markets in Asia. We’ve seen the same dynamic play out in semiconductors, batteries, mining, and pharmaceuticals. But shipbuilding represents a distinctly gnarly systems problem tangled between defense funding, policy, and a fading American manufacturing core.

Why Build Ships?

Any national soul-searching about reindustrialization must start by disabusing ourselves of the notion that we don’t really need to build our own ships. We do. For most Americans, boats are a leisure category — a pleasure cruise, a fishing trip — while the working ocean has faded from the national imagination in a way that airports and interstates haven’t. While airplanes and highways may feel like the global connective tissue, the reality is that 80% of trade volume moves by sea. Without domestic shipbuilding capacity, the flow of global trade, including energy, raw materials, and consumer goods, is ultimately at the mercy of those who do control it.

The last few years have given us countless examples of that vulnerability. COVID pushed the median cost of transporting a shipping container from $2,000 to $20,000. When the Ever Given, a massive container ship, blocked the Suez Canal in March 2021 for six days, each day of disruption held up over $9 billion worth of goods. And when the recent US-Israeli strikes on Iran led Iran to effectively halt shipping through the Strait of Hormuz, it cut off 20 million barrels of oil per day, representing roughly 20% of global seaborne oil volumes.

Alfred Thayer Mahan, a Navy officer and historian described as “an apostle of sea power,” famously argued that “whoever rules the waves rules the world.” He saw the need for a navy as springing from the existence of peaceful commercial shipping. Increasingly, we live in a world where “peaceful commercial shipping” is more at risk than it has been in decades, owing to the volatility surrounding global chokepoints like Hormuz and the Taiwan Strait. The Council on Foreign Relations has put the odds of “a cross-strait crisis between China and Taiwan” at 50% in 2026. Such a crisis could dwarf a Hormuz closure in its ramifications for the global economy; nearly half of the world’s container shipping passes through the Taiwan Strait each year.

The seafaring domain is critical. The threat of great power conflict draws closer; a conflict between the US and China that may define the 21st century; a US-China war over Taiwan could deliver a roughly $10 trillion shock to the global economy in its first year, nearly 10% of global GDP — larger than COVID, the 2008 financial crisis, or the war in Ukraine. The US itself would absorb a 6 to 7% GDP hit, primarily through severed access to advanced semiconductors and the collapse of shipping through one of the world’s busiest sea lanes. Unfortunately, the majority of the ocean-going vessels are built by our would-be competitor.

In 2024 alone, one Chinese company built more commercial vessels by tonnage than the United States has produced since the end of World War II. Today, the US accounts for roughly 0.1% of the global commercial shipbuilding market while China accounts for 53%, up from just under 10% in 2000. China has become the dominant global shipbuilding superpower. It builds the vast majority of global shipping, including shipping critical to US naval power. Three of the ten oil tankers in the US fleet designed to carry fuel for military operations were constructed in Chinese yards. Seven of the twelve ships in the Maritime Security Fleet are made in China. The dependence is reckless, but the economic justification is straightforward: Regulators only require these ships to be US-flagged and crewed; otherwise, the ships would cost roughly $200 million, instead of their current sticker price of $50 million.

Meanwhile, domestically, US shipyards are failing to deliver on Navy shipbuilding goals. Some projects, like the Virginia-class submarine program, running at just 60% of its target production rate, with some ships delayed by as much as three years. Going forward, the US Naval fleet is already 20% short of the minimum capacity needed to carry out core missions. And that shortfall will continue to grow as maintenance backlogs pile up.

Meanwhile, the Chinese Navy, which surpassed the US Navy in total fleet size around 2020, keeps extending its lead. Huntington Ingalls Industries, the US defense prime most focused on seafaring, reported a backlog of $56.9 billion in 2025 — roughly five-years of work at current production rates. Even if Congress tripled shipbuilding appropriations tomorrow, Huntington Ingalls couldn’t build meaningfully faster. China’s advanced capacity, meanwhile, enabled it to add 30 ships to its fleet last year, while the US decommissioned 19 ships and built six, for a net loss of 13. In a protracted naval conflict, China can replace what it loses. We cannot.

To address this vulnerability, we must resurrect our national shipbuilding industry. But any attempt to do so must reckon with the chicken-and-egg problems inherent in bootstrapping a dying industry back into global competitiveness.

Shipbuilding’s Doom Loop

Every industrial process is a system; each input has process and timing dependencies, and every output demands execution. But when the process includes dozens of states and countries, multiple regulatory bodies across land and sea, and thousands of suppliers, each with their own regulatory regime and manufacturing idiosyncrasies, that system becomes almost hopelessly complex.

Regaining domestic shipbuilding capacity in the US means reckoning with a doom loop of interdependencies: labor, stable demand, competitive costs, scale, supply chain depth, and functional shipyards.

You can’t attract workers without stable demand: Shipbuilding apprenticeship completion rates sit below 35%. Even major shipyards, like Puget Sound, only begin with around 200 apprentices. The US Merchant Marine Academy, which offers a Marine Engineering and Shipyard Management major, was described by Transportation Secretary Sean Duffy as “dilapidated”. For those who make it into the shipbuilding industry, first-year attrition runs around 50-60% even at the most consistent shipyards. This attrition happens because there is no confidence in demand signals relative to more reliable industries, like oil and gas or construction. What’s more, around 27% of maritime workers are over the age of 55, meaning we’ll see a silver tsunami of retirements over the next decade.

You can’t generate stable demand without competitive costs: Asian shipyards produce at 5-6x lower prices and twice the delivery speed, driven in part by the massive regulatory burden American shipyards carry. Every input in the US, from labor to steel, is dramatically more expensive. No rational commercial buyer will pay that premium unless compelled by law or subsidy. Even artificial demand from the US government has wavered, with multi-billion-dollar shipbuilding contracts fluctuating between administrations.

You can’t achieve competitive costs without scale: China builds 1,000 commercial ships per year compared to about 10 in the US. Everything gets cheaper with scale. Workers move down learning curves faster with repetition, suppliers price lower with volume, and large workforces can specialize instead of generalize. As a result, the fixed costs of running a shipyard — the cranes, docks, management, financing — spread across more ships, making each one cheaper.

You can’t achieve scale without supply chain depth: Even a simple commercial ship requires thousands of components from hundreds of suppliers. Since 2000, the US has lost more than 25,000 of those suppliers. Some components, like marine diesel engines or marine-grade steel, have limited or no domestic production left. Of the six American companies manufacturing large marine diesel engines before 1980, only one — Fairbanks Morse Defense — still exists, surviving exclusively off of military contracts. Suppliers have done the math and found the US market wanting. No one will invest in marine diesel engine production for a market of ~10 ships per year.

You can’t achieve supply chain depth without functional shipyards: Of all the bottlenecks, shipyards themselves may be the hardest to resolve. You need deepwater coastal sites — exactly the kind of sites that attract the most onerous regulation — including 50-100+ contiguous acres of waterfront with deep draft access, heavy industrial zoning, and proximity to rail or highway. There are vanishingly few sites in the US that meet all those criteria and aren’t already developed.

Which brings us back to labor: Even if you could build a functional shipyard, with access to a deep supply chain, commanding sizable demand at competitive prices, you still need workers to staff it. Which brings us full circle.

No matter which lever you pull, nothing makes enough of a difference on its own. Legislation can artificially lift demand, but it cannot sustain a workforce. Foreign expertise can bridge some of the knowledge gap, but it can’t magically spin up a supplier base. What’s more, solving all of these issues simultaneously requires a level of sustained, coordinated, multi-decade industrial policy that the US hasn’t demonstrated since before the Cold War.

The solution here isn’t akin to the CHIPS Act for semiconductors or the IRA for batteries, solutions that were, primarily, acts of capital deployment by the federal government. The closest analogy for a solution is the interstate highway system. From 1956 to the 1980s, the National Interstate and Defense Highways Act enacted a 20+ year, multi-administration, federally funded infrastructure program that required simultaneous land acquisition, engineering, workforce development, materials supply, and political consensus across all 50 states.

USS Sturtevant (DD-240) under construction at New York Shipbuilding Corporation on December 1, 1918. Credit: Naval History and Heritage Command, Washington, DC

Some good news and some bad news there.

The good news is Washington seems committed to prioritizing shipbuilding, at least for now. Congress has consistently appropriated $2.5 billion more per year for shipbuilding than the Navy has requested since 2015. In 2025, Congress dedicated $34 billion to shipbuilding through the “Big Beautiful Bill,” making it the largest defense line item in a $150 billion Department of Defense budget. By placing shipbuilding at the top of the list of defense modernization priorities, Washington is signaling to private capital, allied industry, and the workforce that the sector has durable political backing.

The bad news is we don’t have 20 years to shore up America’s shipbuilding capacity, and it’s going to cost a lot more than $34 billion. What’s needed is an entirely new strategy. In World War II, we leveraged commercial industrial dominance to build the Arsenal of Democracy. But today, we have a dramatically limited ability to transition our domestic industry into wartime production because we don’t have a domestic industry to transition. We can’t run back the same playbook. Things have changed.

New Problems Need New Playbooks

In World War II, the US simply outbuilt our enemies. William S. Knudsen described the strategy as “[smothering] the enemy in an avalanche of production, the likes of which [they] had never seen, nor dreamed possible.” That won’t work this time because, today, China occupies the dominant manufacturing position the US held in 1942. Where the US produced around 40% of global manufacturing output in 1942, China represents 35% today. In fact, China has even more unilateral capability than WWII-era America did given its authoritarian political and economic system, including heavily subsidized financing and “military-civil fusion,” which funnels commercial capacity directly into naval output.

Instead, we must rely on innovation rather than sheer output. Trying to beat China at its own game of sheer production volume would be misguided. The opportunity lies in finding the highest leverage innovation. There is actually a precedent for this from World War II: The Liberty Ship program popularized arc welding, which is running an electric current to fuse steel plates into a continuous seam, to replace riveting, which required skilled teams hammering heated pins through pre-drilled holes. That enabled rapid prefabricated shipbuilding with minimal training required. That one innovation took construction of each ship from 355 days down to 40 — within months. These wartime welding advances in shipbuilding contributed to broader US manufacturing efficiencies post-war.

The focus should be on what could allow us to leapfrog ahead, not just catch up. Modular construction techniques would allow components to be built in distributed locations and assembled more efficiently, prioritizing building many standardized, cost-effective vessels over a few overengineered ones. Integrated autonomous systems and AI-driven production software could multiply the impact of both.

Jumping ahead will also require meaningful acceleration of digital integration. Unfortunately, we’re not currently moving in the right direction on this front. The Navy has hollowed out its in-house technical capacity, having cut the number of naval architects and engineers from 1,200 to 300 since the 1990s; a post-Cold War cost-saving measure that transferred design work to private industry just as private industry itself began to atrophy. As a result, we’re seeing costly design flaws and mid-construction rework across major programs, driving up costs by at least 25%.

Closing that gap doesn’t require inventing new technology; just deploying what already exists. Advanced block assembly, robotic welding, and AI-driven logistics can halve construction timelines. The Navy’s new Shipbuilding Operating System (Ship OS) — a $448 million Palantir-built platform launched in December 2025 — integrates planning software, legacy records, and operational sensor inputs to flag bottlenecks, optimize workflows, and refine risk models. Similar tools at General Dynamics Electric Boat cut planning tasks from over 160 hours to under 10 minutes. At Portsmouth Naval Shipyard, Ship OS AI systems have cut material review times from weeks to hours. Those are the kinds of leverage points that could be replicated across the industry to enable the US to work smarter, rather than harder.

The goal would be to shift the economic curve of shipbuilding in our favor. A comparable analog is visible in the space industry. We were able to shift from a government-dependent monopoly with high, administered prices to a thriving, competitive ecosystem after SpaceX demonstrated reliable reusable rockets that dramatically reduced launch costs. In shipbuilding, modular construction and software-driven production automation can play a similar catalytic role in which high-integrity blocks, standardized interfaces, and tightly sequenced assembly can reduce labor hours per ton, shorten cycle times, and lowering barriers for new entrants in the space.

While we’ve made progress on some of the most compelling innovations, we’re far from the cutting edge. In fact, some of the best testing grounds are in allied nations like South Korea, with companies like HD Hyundai leading the charge in smart manufacturing, robotic welding, and digital twin systems. If innovation is where we’re going to pull ahead, we aren’t going to be able to do it alone.

Walking (With Allies) Before We Can Run

We can’t lose sight of the long-term vision of rebuilding domestic shipbuilding capacity in the US. Focusing on innovation over raw output is critical. But in the short-term, we’ll have to alleviate some of our dependencies by leaning even more heavily on another: international partnerships.

It’s worth noting that many of the most important processes in modern shipbuilding were developed in the US during the 1950s and then exported to Japan for execution. Advances like statistical process control, a method for using real-time measurement to catch manufacturing defects as they emerge, rather than inspecting finished products, were invented by Walter Shewhart at Bell Labs, then brought to Japan by W. Edwards Deming, who over the course of two decades, trained almost 15,000 Japanese engineers in them. Over time, those capabilities spread to other allies, like South Korea — and eventually to China.

Competing with Chinese capacity can feel insurmountable when you compare its 53% of global capacity to our 0.1%. But the playing field feels dramatically more even when compared to the combined share of Japan and South Korea at around 43%. That capacity has enabled those allied nations to retain the scale, skilled labor, and industrial depth the United States has lost. We’re already seeing a precedent for this kind of allied collaboration in semiconductors, with TSMC’s $65 billion investment into its Arizona facilities, the largest single foreign direct investment in US history.

And a shipbuilding partnership is a win-win for a country like South Korea. Despite a half-century of continuous large-scale shipbuilding, it still faces labor shortages and relies on foreign nationals for roughly a quarter of its shipyard workforce. Sooner or later, every nation will require rapid innovation and, given the US’s industrial deficiencies, a collaborative approach would be more effective than going it alone. Allied countries like Japan and South Korea can provide the stop-gap of workforce expertise and capital required, but in the long run everyone stands to benefit from an independent American shipbuilding innovation machine.

The critical distinction in relying on allies today, compared to the offshoring of the late 20th century, is about capability flows. In the 1980s, America surrendered the technological and workforce capabilities around shipbuilding, allowing them to flow towards other countries. Today’s proposed partnerships invert that flow. American workers travel to South Korea to absorb labor capabilities that then return home; Korean capital and equipment flow into American shipyards to rebuild industrial infrastructure on US soil. The long-term question is whether we can treat this as a generational capability build rather than a cost-minimization play.

Precious Few Sites

Beyond labor and supply chains, the US faces a hard physical constraint: space. The US has roughly 150 private shipyards, but only seven can build major oceangoing vessels, and those seven are dwarfed by Jiangnan Shipyard alone, a single Chinese yard that has more capacity than all US shipyards combined. China also manufactures roughly 80% of ship-to-shore cranes and 96% of shipping containers used in the US. Since the 1950s, the number of American shipyards capable of constructing large oceangoing vessels has declined by more than 80%. The Brooklyn Navy Yard shut in 1966. The Philadelphia Naval Shipyard shut in 1996. The Mare Island yard outside San Francisco shut in 1993. Each of these was, at one point, among the most productive shipbuilding facilities in the world. Their workforces have long since retired. Even if the appetite for additional shipyards grows, there is simply a limited number of viable sites for large-scale shipyard construction.

Modular construction could alleviate that barrier. By splitting a vessel into discrete block sections that inland or distributed facilities can build in parallel, the US can leverage interior regions previously considered irrelevant to the industry. Completed modular sections would then be shipped to coastal yards for final assembly. The Navy has already expressed support for this approach. Modular construction could double the current annual production rate of existing US shipyards and grow employment in the sector from roughly 400,000 to nearly one million.

Our Future Is Just 30 Miles Downriver From Our Past

Just 30 miles down the Delaware River from the New York Shipbuilding Corporation’s former yards lies a symbol of the future of American shipbuilding. The Philly Shipyard, recently renamed Hanwha Philly Shipyard, is where South Korea is making a $5 billion investment to execute exactly the playbook we’re describing. An all-out ally-first innovation build-out.

Starting with labor, Hanwha is already sending American workers to Korea for training, while simultaneously embedding Korean production systems into its operations in Philadelphia. It has implemented a joint-build model with South Korea for an LNG carrier ordered at their Philly Shipyard — a project that combines Korean engineers, production systems, and supply-chain processes with the US workforce, allowing American employees to gain hands-on exposure to advanced construction techniques, regaining the process knowledge the shipbuilding industry spent decades losing.

The $34 billion American investment in shipbuilding is the down payment. The Hanwha partnership is a bridge loan. Modular construction and software-driven production are the long-term bets. But none of it works without the one thing the US has failed to sustain for the last 40 years: the political will to treat shipbuilding as a generational commitment, not a line item. The Delaware River, running from the ghost of New York Ship’s past to Philly’s rebirth, is where we find out whether the country still has the resolve to build.

“What think you of an American Fleet,” Adams asked James Warren, paymaster of the American forces at Watertown. Adams knew the colonies would balk at bankrolling a navy, so he looked to the private sector. “A public Fleet as well as privateers,” he wrote, “might make prey enough of the Trade of our Enemies to make it worthwhile.” Adams argued that privateers — private ships manned by private citizens on behalf of the government — could sail at their own expense and take their pay from enemy prizes. He grasped the power of harnessing the private sector in service of the national interest.

Just a few months later, before the signing of the Declaration of Independence, the Continental Congress began issuing commissions for private ships of war and letters of marque and reprisal, authorizing private ship operators known as privateers to destroy enemy vessels and property. John Hancock later signed blank authorizations and dispatched them to the colonies. They came with exacting instructions: which enemy vessels could be seized, what cargo could be taken, which ports had jurisdiction to adjudicate prizes. Shipowners seeking commissions had to register their vessels and post bonds guaranteeing lawful conduct. Over the course of the Revolutionary War, Congress granted roughly 1,700 of these commissions. Nearly 800 privateer vessels went to sea, capturing or destroying some 2,300 British ships. Adjusted for inflation, the losses to British commerce were north of $300 million.

This motley fleet of naval vessels and private merchant ships was among the most novel powers the colonies had at their disposal. Private citizens, operating their own ships at their own expense, imposed enormous costs on the world’s preeminent naval empire. Their pay came from a share of whatever a prize court judged a seized vessel and its cargo to be worth. The Framers of the Constitution codified this authority in Article I, Section 8, placing the power to “grant Letters of Marque and Reprisal” alongside the powers of Congress to declare war and regulate captures on land and water.

Congress hasn’t invoked the marque and reprisal clause since the War of 1812, when several hundred privateers put to sea with letters of marque in hand. The United States later signaled its intent to abstain from privateering, but never bound itself to that position by joining the 1856 Declaration of Paris. Any modern revival of the framework would no doubt draw resistance, especially from those who see treaty commitments like the UN Convention on the High Seas and evolving norms of international law more broadly as constraints even on constitutionally enumerated congressional powers. That opposition should be engaged seriously, but not accepted uncritically. As the nation marks its 250th anniversary, the time has come to revive the authority that helped bring the republic into being. It’s time to bring back the American privateer.

Sanctions, as currently practiced, are a financial tool. They prohibit the provision of services to designated individuals or entities and, in effect, push banks to freeze those parties’ accounts — US banks by legal mandate, and increasingly foreign banks by fear of secondary sanctions that would cut them off from the dollar-clearing system. And still, sanctioned oil makes its way to market. Tankers registered to shell companies in offshore secrecy jurisdictions carry billions of dollars’ worth of Russian and Iranian crude across open water to willing buyers — often in full view of commercial satellite coverage. These illicit oil exports, overwhelmingly supported by China, India, and Turkey, fund Russia’s war against Ukraine and Iran’s attacks on US military bases and Gulf energy infrastructure. The gap between sanctions on paper and the vessels carrying millions of barrels of sanctioned oil at sea is one of the most consequential vulnerabilities undermining American economic statecraft today.

The Navy does interdict some of these shipments. But the Navy’s mandate spans the world’s oceans and runs from great-power deterrence to humanitarian assistance. Dedicating scarce ships and crews to chasing a fleet of tankers is neither sustainable nor strategically sound. Shadow fleet operators — the loose network of brokers, shell-company owners, and registries that move sanctioned oil — know it. They exploit the vastness of the ocean and the limited bandwidth of Western navies to move sanctioned cargo with near impunity.

And the problem is getting more dangerous. Sanctioned tankers departing Russia’s Baltic ports appear to have recently sailed with armed two-man teams with ties to the Kremlin-linked Wagner Group, Russian military intelligence, or Russian special forces. Meanwhile, Moscow is now going so far as to provide armed Russian navy escorts for sanctioned ships transiting the English Channel.

Consider the scope of the problem. More than a thousand shadow fleet vessels operated by Russia and Iran routinely move sanctioned crude and liquified natural gas around the world, generating billions in revenue for regimes hostile to the United States and its allies. That revenue pays for the thousands of Russian drones attacking Ukrainian soldiers and Iran’s years-long development of a rogue nuclear weapons program. Every barrel sold undermines the effectiveness of American sanctions. We must understand these vulnerabilities now — especially in preparation for a potential conflict with China, which operates the largest navy and commercial fleet in the world.

The scale of America’s maritime decline compounds the problem. The US Navy fleet has shrunk to 287 deployable ships, down from a 1987 peak nearly twice that size — and well below China’s 370-plus warships. The picture is even worse in terms of ships that can be drawn on from the commercial fleet. Fewer than 200 oceangoing ships fly the American flag today. In 1975, that number was 857. Meanwhile, a single Chinese state-owned shipbuilder produced more commercial tonnage in 2024 than the entire US industry has built since the end of World War II. According to the Office of Naval Intelligence, Chinese shipbuilding capacity exceeds America’s by a factor of 232. And China’s massive merchant fleet is capable of acting as an arm of the Chinese Communist Party at a moment’s notice.

America’s shipbuilding gap is not insurmountable. For privateers in particular, all that’s needed are ships fast enough to close on sanctioned tankers, large enough to carry a boarding team, and cheap enough to stand up a fleet with private capital alone.

Fortunately, that hull already exists in ready supply. The global fleet includes more than 5,000 offshore supply vessels that can be retrofitted with helicopter decks, dynamic surveillance systems, ample working deck space for fast boats, and accommodations for up to several dozen personnel. Designed to operate in harsh seas for weeks at a time, these vessels have been used to support offshore oil and gas platforms around the world. A used vessel of this class can trade for orders of magnitude below the cost of a Navy combatant, including the cost of conversion to accommodate necessary communications, intelligence, surveillance, and reconnaissance tools, as well as a light helicopter detachment. The operational know-how for applying these tools against maritime threats already exists. What is missing is the legal authority for private actors already engaged in anti-piracy and vessel protection services to enforce American sanctions at sea.

A modernized letter of marque framework would do just that by authorizing vetted firms to identify and board vessels carrying sanctioned cargo. Congress would pass legislation specifying which categories of activity trigger the authority, declaring specific categories of activity to be within the bounds of authorized conduct, establishing verification protocols, and licensing the firms permitted to board and inspect suspect ships.

The United Kingdom Maritime Trade Operations (UKMTO) offers a model. A Royal Navy organization, UKMTO maintains a presence in Dubai to provide situational awareness of commercial traffic and coordinate threat reporting for merchant vessels in contested waters. The US and its allies could build something similar to support privateers operating under letters of marque. A US-led Maritime Sanctions Intelligence Center could fuse satellite tracking, transponder and radio-frequency data, shipping registries, oil-trade records, and financial intelligence to identify sanctions evaders and push vetted intelligence to the firms authorized to act on it.

To make this framework viable, the United States will need to reckon with its geography. Most shadow fleet activity happens far from American waters — in the Strait of Malacca, the Persian Gulf, and the eastern Mediterranean. Privateers would need authority to operate from allied ports. That, in turn, would require the United States to negotiate basing and access agreements so that licensed vessels could refuel, resupply, and stage operations from forward positions. Without forward basing, enforcement would be tethered to the Western Hemisphere while sanctions evasion flourishes half a world away.

The United States could also modernize existing tools — starting with the Proliferation Security Initiative (PSI). Created after the September 11 attacks to counter the proliferation of weapons of mass destruction, PSI now includes 116 participating countries and facilitates bilateral ship-boarding agreements that permit participating naval forces to board foreign-flagged vessels suspected of transporting prohibited cargo. Repurposed for sanctions enforcement, it could provide a ready-made legal framework that privateers operating under letters of marque could strengthen.

Technology, meanwhile, can help close the gap between limited maritime reach and the need for persistent awareness of sanctions-evasion hotspots. Over the past few years, American defense technology firms have poured billions into autonomous maritime platforms. An entire industrial ecosystem is emerging — one that could provide the backbone for privatized sanctions enforcement without requiring a single additional Navy hull. Saronic Technologies, founded by a former SEAL Team Six operator, has raised close to a billion dollars, won a $392 million Navy production contract, and bought a shipyard in Louisiana where it built the hull of a 150-foot unmanned vessel in nine months — the fastest turnaround for a ship built in America since World War II. Saronic aims to build 200 autonomous vessels per year and is planning Port Alpha, a purpose-built shipyard for unmanned vessels at scale. Saronic has company. Blue Water Autonomy is building a 190-foot robot patrol craft designed for missions that last months without a crew aboard. HavocAI builds swarm-capable platforms controlled through software that lets a single operator coordinate dozens of assets across surveillance and interdiction tasks. These are real assets already being deployed and tested by the United States government.

Their capabilities map directly onto the requirements of a modern letter of marque operation. Picture a licensed firm deploying a fleet of autonomous surface vessels across designated Maritime Sanctions Enforcement Zones — chokepoints and known handoff areas where sanctioned oil changes hands. Think waters off Southeast Asia, Mediterranean transit corridors, anchorages near the UAE where Iran has long conducted ship-to-ship transfers. Equipped with high-resolution optical and infrared cameras, radar, LiDAR, and AIS receivers (the transponders ships use to broadcast their identity and position), these unmanned vessels could maintain watch over enormous stretches of ocean at a fraction of the cost of manned patrols. They would fill the gaps that satellites alone cannot cover — including cloud cover and the deliberate timing of transfers to slip between overhead passes.

When an autonomous vessel, supported by a full suite of surveillance tools and signals, flags a suspect ship, the data would route to the Maritime Sanctions Intelligence Center. The center could cross-reference the target against its priority list and, if the match is confirmed, a manned interception team would move in to board, inspect, and, if warranted, seize the ship and its cargo. To make these efforts economical, however, the government must solve a problem that already plagues existing efforts to seize sanctioned oil at sea: slow judicial review.

A letter of marque framework depends on financial incentives. These incentives collapse if seized assets sit in legal limbo for months or years while costs mount. Consider the MT Skipper. After US forces seized the tanker and its 1.8 million barrels of Venezuelan crude in December 2025, the federal government spent at least $47 million repairing a vessel worth just $10 million. Storage alone for the offloaded oil runs $450,000 a month. Meanwhile, the Justice Department did not file a forfeiture complaint until late February 2026 — more than two months after the seizure — and by then prosecutors conceded that ongoing expenses threatened to consume the cargo’s entire value.

Some of the Skipper’s complications may be Venezuela-specific, including competing claims over the vessel and its crude in particular. But where the government can move faster, it should — whether by executive order, statutory reform, or regulatory action. No private firm will invest capital in autonomous fleets, manned enforcement vessels, and training boarding teams if the supposed financial reward is bogged down by years of legal disputes.

Any legislation establishing a letter of marque framework must therefore build in procedural reforms fast enough to keep private enforcement viable: statutory deadlines for judicial review of interdiction claims, authority to sell contraband cargo before forfeiture is finalized, and time-limited forfeiture proceedings measured in weeks rather than months or years. Without them, the system will fail.

With these foundations in place, one more authority could extend the framework’s reach even further: a sanctions qui tam — a private right of action modeled on one of the oldest enforcement mechanisms in American law.

Qui tam has deep roots in American legal tradition. The False Claims Act, signed by Abraham Lincoln during the Civil War, lets private citizens bring fraud cases on behalf of the federal government and collect a share of whatever is recovered — typically between 15 and 30 percent. Since Congress strengthened the statute in 1986, qui tam plaintiffs have driven the recovery of tens of billions of dollars in fraud against the government. The mechanism works because it aligns private financial incentives with public enforcement goals, building a self-sustaining system that does not compete for appropriations.

Congress ought to create a maritime sanctions version. A licensed firm could petition an expedited admiralty tribunal, presenting intelligence and physical evidence corroborated by the Maritime Sanctions Intelligence Center that a specific vessel is transporting sanctioned cargo. Upon authorization by the tribunal, the firm could then conduct the interdiction and file for forfeiture under the compressed timelines established by broader legislation. The enforcing firm would then receive a defined share of proceeds when the cargo and vessel are sold. A separate tranche of the recovered funds could also flow to the United States Victims of State Sponsored Terrorism Fund, which supports the families of Americans killed by groups such as Hamas, Hezbollah, and the Houthis — organizations funded in part by Iran’s illicit oil economy.

America’s founders recognized that national power does not belong to the state alone. The Revolution was fought, in significant part, by private shipowners who armed their own vessels, risked their own fortunes, and sailed under commissions issued by a government that barely existed. The Constitution preserved that authority because the Framers understood its strategic value.

The challenge today is structurally similar to the one our founders faced. Western navies cannot police every sea lane. But American industry — with autonomous ships, AI-driven surveillance networks, and renewed shipbuilding capacity — is already assembling the toolkit to close the enforcement gap. Two hundred and fifty years after the Continental Congress first armed private citizens to defend American interests at sea, it is time to bring back the American privateer.

At the center of this history was Norman Augustine. One of the defense industry’s major figures of the late 20th century, he merits a lengthy introduction: Raised in Colorado during World War II and educated in the aeronautical engineering department at Princeton University, he began his career at the Douglas Aircraft Corporation in 1958. He entered government service in 1965, serving as assistant director of defense research and engineering under Secretary Robert McNamara. In this capacity, Augustine wrote the approval papers for several weapons systems that remain in regular use today, including the Patriot missile and the Airborne Warning and Control System (AWACS).

After leaving government, Augustine joined LTV in Texas, where he expected to remain. But he was approached about returning to the Pentagon as Assistant Secretary of the Air Force for Research and Development in 1973. After initially refusing, he met Defense Secretary James Schlesinger, who instead steered him toward the Army; Augustine was persuaded by Army Secretary Howard 'Bo' Callaway to accept the role of Assistant Secretary of the Army for Research and Development. There, he championed the Army's "Big Five" modernization programs — Patriot, Abrams, Bradley, Apache, and Black Hawk. He earned a reputation for pushing troubled programs through development rather than canceling them, leading Callaway to promote him to Under Secretary and ultimately positioned him to become acting Secretary of the Army in 1975.

Augustine joined Martin Marietta corporation in 1977, becoming CEO in 1987 and chairman in 1988. After the “Last Supper” at the Pentagon in 1993, he served as chief architect of the merger between Lockheed and Martin Marietta, forming the Lockheed Martin Corporation, America’s largest defense contractor, in 1995. Augustine retired as chairman and CEO in 1997. During his tenure, he initiated Lockheed’s bid to acquire Northrop Grumman, which was blocked by the DOJ on antitrust grounds in 1998. In 1999, shortly after retiring, Augustine was recruited by Director George Tenet to co-found In-Q-Tel, the venture arm of the Central Intelligence Agency, which has invested in hundreds of early-stage startups including Palantir and Databricks. He has also served on various corporate boards including Procter & Gamble and ConocoPhillips, in addition to Lockheed Martin. He is the author of several books, including Augustine’s Laws and The Defense Revolution. 

In order to better understand the state of the American defense primes and the litany of challenges that the industry faces, I sat down with Norman Augustine, now 90 years old, to discuss one of the most distinguished careers that the US aerospace and defense industry has ever seen. What follows is a transcript of our conversation.

CB: What impact did World War II have on you, growing up as you did in Colorado during the 1930s and 1940s?

NA: I was 10 years old when World War II ended, and so I was old enough to understand. We played soldiers all the time. We had a victory garden in the yard where we grew vegetables, and we saved newspapers and tin. You couldn't get gasoline, and the number one thing you couldn't get when you were a kid was bubble gum, for some reason. I grew up at the foothills, or actually on the high prairie next to the foothills, and the mountains were a big thing in my life. I spent a lot of time there as a youth. I was raised as an only child because my sister had died.

My memory of when World War II started: I would have been six years old. It was a Sunday morning. My father and mother and I were at a restaurant in a little town called Louisville, Colorado, and there were a bunch of people there having brunch. Somebody burst in the door of the restaurant and yelled. I still remember it like it just happened. “The Japs bombed Pearl Harbor,” and he slammed the door, went running down the street like Paul Revere.

At the end of the war, my parents took me into downtown Denver, because they thought this would be a historic event. Down there they were celebrating, the streets were full of people. Then, of course, my father worked for Remington Arms.

CB: Your father was also in the defense industry?

NA: No, he was working at the fruit and vegetable market, but during the war, Remington opened a huge plant outside of Denver, and it paid an awful lot better. A lot of people quit and went to work for Remington as long as the war lasted. Then they all left, but Dad was in wholesale fruit and vegetables on the old Wazee Market in Denver. He had fought in World War I.

CB: Do you know what weapons your father was building at the Remington plant?

NA: I think they were just bullets.

CB: Was he furloughed or let go after the war ended?

NA: He was. Everybody was. The whole plant disappeared, gone. America has its habits. We always think that when there's a war that there's going to be peace forever after, and then a surprise happens.

CB: How did you end up at Princeton University from Colorado?

 NA: It's like most everything else in my life — because it was a random event, an accident —  and the only credit I can take is that I took advantage of it. My family didn't have very much money, there was a question of whether I was going to be able to go to college or not.

 I had wonderful parents, loving parents — hardworking, decent people that taught me all the things that were important. My dad only went up to eighth grade, and he used to help me with my math in high school. How he did that, I can't imagine, but he did, and he just never had the same opportunities. He had to go to work when he was in eighth grade, and his father only got through fourth grade. My dad was born in 1894, my mother in 1893. Those were pretty tough times. Mom grew up in Denver, which was a pretty wild western town. And my father grew up in a little mountain town about 10,000 feet up on the mountain. It was 200 people, a mining town — really a wild town up there. His stories are something else.

 CB: There were some very interesting people orbiting Princeton in the 1950s when you studied there, especially around the Institute for Advanced Studies, you had Albert Einstein and Robert Oppenheimer, among others. Did you ever encounter them?

 NA: I passed by Einstein on the street, but never knew him, never met him, really. Some of my classmates were recently telling a story that they had some question they wanted to ask Einstein. They walked down to his house and rang the doorbell, and he invited them in to talk. My freshman physics teacher was a fellow by the name of John Wheeler. Wheeler was Einstein's main colleague — though I didn't know that at the time. He was my physics teacher and put on the greatest demonstrations of physical principles.

One thing that was really funny: my roommate and I were in the physics lab, freshman physics, and we were making what was called a rheostat, which measures electric charge. The way it measures it is that it has two little pieces of gold leaf that are suspended together at one point on a little wooden stand, and these two little pieces of gold leaf, depending on the amount of charge you put on them, would move further and further apart. So, depending on the angle between them, you could determine how much the charge was on the gold leaf, and it was a way to make a measurement. Our assignment was to make a rheostat. We were there for hours trying to make a rheostat. Gold leaf is only a couple of atoms thick, and it’s so thin that if you barely touch it, certainly with a pair of scissors, it just cuts apart, sort of floats away. It will eventually settle on the ground in a perfectly still room, but in a room with any air at all, it just floats around the room — little pieces of gold. 

We were having a terrible time, and there was a janitor in the room who was moving tables and chairs around in the room. And after a little while— we were struggling mightily — he came up to us and asked if we would help to move a couple tables. We were ready for a break anyway, so we said, "Sure.” We helped him move a couple tables, and at the end of that, he came up to us and he said, "You did me the favor, now I'll do you a favor.” He said, "I see you're having trouble making your rheostat. What you do is get your sheet of gold leaf  and make a sandwich out of it between two pieces of notebook paper, and then you can cut it with your scissors. It will tear up the gold, and then you can peel the notebook paper away very carefully, and it'll work well.”

 It worked like a charm, and so we went away thinking that even the janitors at Princeton are pretty smart. About a week later, there was a lecture in Alexander Hall. We went to the lecture, and when we came into the room, up on the stage, seated in one of two chairs up there was the janitor — and we both thought, "What’s he doing up there?” It was supposed to be a lecture by Oppenheimer. Well, it turned out that the janitor was Oppenheimer.

CB: You had told me when we last spoke that there was a sort of caste within the US aerospace industry which formed out of Princeton’s Engineering department. Could you tell me a little bit more about that?

NA: It was Princeton Aeronautical Engineering at the time. In my class, there were eight students, and it had a large faculty. It was one of the better aeronautics programs of the country, with a  faculty much larger than the student body. Over the years a pretty good collection of people that graduated from Princeton Aeronautical Engineering. If there was anything unique about us, I think it was that Princeton taught engineering in a liberal arts environment. They would ask us: can you do this — some technical problem — but they would also ask, should you do this, and why? We had to take liberal arts courses, quite a few of them. For most of us that was kind of a pain, but it had a very good effect — it broadened us a whole lot. And Princeton didn't have a law school, didn't have a medical school, and no business school. 

Years later, at the end of the Cold War, when the aerospace industry had to consolidate, that little group of people from Princeton had become, just by chance, a whole group of us who were running companies and knew each other. The Boeing CEO was from Princeton, Lockheed, Martin, Grumman. The Hughes CEO, just retired, was a Princeton guy; Westinghouse was a Princeton guy; McDonnell Douglas was led by Princeton guys for two generations of CEOs. We all knew each other well, and we sort of knew who you could trust, which was most everybody. 

CB: That's a very interesting network effect.

NA: I don't know anybody who's ever told that story. It's just kind of interesting.

 CB: You wrote your thesis on vectored slipstream aircraft. What is that?

 NA: That's correct. Vertical takeoff and landing was kind of a big deal in those days. Helicopters were great, but couldn't carry very much and couldn't go very fast. The idea was to build regular airplanes that could hover, some hybrid of the two. And today the V-22, the Air Force and Marine airplane, is an example, much more advanced than what we were thinking of. The idea was you would have a regular airplane, but with big propellers, and you would have flaps on the wings — very complicated flaps, far more complicated than today. 

The one I worked on was called a double-slotted flap, and when you put these flaps down, with a great big windmill, or a propeller blown in front of them, it would blow the air downward, and if you could make the air go downward, it would build a pressure to push the aircraft upward. If you got it just right, you could hover in the air; but if you wanted to act like an airplane, you raise the flap up, and then you could go forward — nothing spectacular, but at forward speed. That was all I worked on, my thesis. It involved a lot of wind tunnel testing, some theoretical work. My thesis advisor was Professor David Hazen, who was really a great guy, I thought he was old then. He was probably 35. 

CB: The year before you started your career was 1957, the year of Sputnik, and I wonder what kind of an impact the Sputnik moment had on your generation of engineers.

 NA: Huge, huge. It was late ‘57, I was studying engineering, and the United States was on top of the world in terms of technology. Russia was not much of anything in terms of technology, they had large forces in terms of numbers of tanks and things, but pretty primitive technology. The United States was king of the hill when it came to technology, particularly military technology. We were so confident, so sure of ourselves. Then, one day, Sputnik goes up. The way I heard about it was I was walking into the graduate school entryway, and somebody passed me and said, "The Russians just put up Sputnik.” 

I didn't know what Sputnik was, but I found out pretty quickly. It was a shock to the technical community, as well as the country, that these sleepy Russians put up a spacecraft before we did. And then they had this whole history: they put up a person before we did, and it was a huge wake-up call to the country writ large, but particularly to the aerospace industry, which during World War II had really risen to the occasion, more so in production than in technology.

The Germans probably had better technology, but we were vastly superior as an industry, building 100,000 airplanes in one year. The record for putting out destroyers was three days from going down the ways into the water, from the time they laid the keel, if you could imagine that. Sputnik came all of a sudden. It's a little like now, when China is suddenly waking up and our problems with Iran, where they're producing somewhat of a wake-up call. 

Sputnik was the trigger that caused a lot of good things to be done in this country. We created ARPA, we raised the defense budget and became more serious about higher education, particularly technical education. And thank goodness we brought in a lot of foreign individuals through our graduate schools in science and technology. Had we not done that, we probably still wouldn't have caught up — and that's not a view that's widely held, but it's true. That was kind of the story of what it was like when Sputnik went up: The space race was suddenly on, and anybody who could spell “space” had all kinds of opportunities. 

CB: You arrived at the Pentagon not too long after Robert McNamara did, if I'm not mistaken.

 NA: McNamara was a business-focused guy. He was not happy with the way programs were getting started and canceled over the years — there had been a lot of that — and he wanted a new way of starting programs. It just happened that Patriot was one of my assignments. I worked in the Office of Defense Research and Engineering, which was a fairly small group of fairly young guys with a remarkable amount of authority, scarily so. Some of the programs I was responsible for were Patriot, AWACS, and a couple of space programs.

The army was really pushing to build Patriot — it was called SAM-D at the time, for surface-to-air missile development, SAM-D. McNamara had started a little outfit called Systems Analysis, which I'm sure you heard all about. They were the “whiz kids.” We were the dumb guys — it was a really smart, young, totally inexperienced group of often fairly arrogant guys. I was one of the few at DoD who managed to make friends with a number of them. Still am to this day. They had a lot of power, and they were deeply opposed to Patriot. 

To settle the battle, McNamara told my boss, who told me — this was on a Friday — on Monday, McNamara wants a paper no more than 20 pages long, that goes through all the issues in a very balanced fashion of whether we should develop Patriot. At the end of it, there should be a place for the Secretary of the Army, the Deputy Secretary of Defense, and so on, to sign which option — there should be a group of options — that they wanted to support, then give it to McNamara. I spent the whole weekend, day and night, working on this paper. I'm pretty sure it's classified, because I remember having trouble with the classified printer that Saturday night. I was alone in the Pentagon, working on it much of the night. The paper went up to McNamara, who gave the go-ahead to build Patriot. This probably would have been 1966. That's how it got started. And if you think about it, that was 57 years ago from today. I just did the arithmetic. 

Recently, the President of Ukraine said the biggest thing the United States could do was to “give them us Patriots.” Well, the Patriot is 57 years old. If you think of the commercial consumer technology in 1966 compared to today, it's laughable. The Patriot was upgraded along the way, but I've often said you can't turn a Volkswagen into a Maserati. You could make it a little better, but there's a limit. So much of the equipment we're fighting with today was developed back in those days when I happened to be in the Pentagon.

 CB: I read that you were sent on a special mission to Vietnam, in uniform, on behalf of David Packard, the famous Hewlett-Packard co-founder. What was your Vietnam mission?

 NA: The Army had developed a fighting vehicle. They were very sensitive. I keep wanting to call it a mini tank. It weighed 17 tons, whereas the main battle tanks weighed 60 tons. The Army was very sensitive when somebody referred to these things as tanks. But anyway, it looked like a tank, it had a big cannon on it, a machine gun, and a crew of, I think, four.

 CB: Was this the one with aluminum armor, the Sheridan?

 NA: The Sheridan was the infantry fighting vehicle. This was a lightweight tank. They had just finished building it — and there were some already in Vietnam — and it could be carried in a C-5A, as I recall. It was having a bad time in Vietnam. The armor wasn't strong enough to counter anti-tank weapons, and its reputation wasn't very good, and Dave Packard was DepSecDef, had been one of the founders of Hewlett-Packard, was kind of a business guy like myself; I was junior to him in the pecking order, but David and I had worked together a fair amount, just because we were both interested in the business and engineering and manufacturing stuff. So he asked that we set up a group of experts from outside of the Pentagon who could go to Vietnam.

There were about eight or nine who were to go to Vietnam, tour around, and look at the equipment and see what the problem was, what should be done. The guy who we chose to lead it was the head of Caterpillar Tractors, and I was sort of the Pentagon representative in charge of this study. We all spent, I guess, two weeks in Vietnam. It was a couple months before the Tet offensive.

 The group of us wandered around to different places in the field to find these vehicles. We talked to people, found a bunch of problems with the vehicles, and learned a lot about why they're different from tanks. We flew around in helicopters, landing out in the middle of the jungle, and talked to soldiers. We were in Army uniforms, but no rank or unit signature. So we did our job, then we came back to Washington. 

I happen to remember that when we got back to Washington, we landed at Andrews. That was always a dreadful experience, because invariably, when you landed at Andrews, there was an aircraft coming back with soldiers to be taken to Walter Reed, or worse, who’d been shot up in Vietnam. On my way home, I decided to stop at the Pentagon, because I hadn't been there in two weeks. When I walked through the halls, there were troops from the 82nd Airborne sitting on the floor through many of the halls of the Pentagon, sleeping up against the walls. That was when the riots were going on in Washington. They had war riots, and very quietly, they brought in a lot of troops from 82nd and had them hidden in the Pentagon and the courtyard in the middle of the Pentagon, in case things really blew up in Washington. Ultimately they weren't needed or they weren't used — and anyway, I was shocked by that. I thought there were more troops here than there were in Vietnam. I went home, and that was the end of most of my Vietnam experience.

CB: You're a co-founder of In-Q-Tel, the CIA’s venture arm. I believe George Tenet personally reached out and requested you help set that up.

NA: I had just left Lockheed Martin; I took early retirement. George  asked that I come by and visit, and he said that their organization was facing a real problem: it was dependent upon the very leading edge technology, and that the leading edge of technology was no longer governmental — it was in Silicon Valley and Boston. The problem was that people in Boston and Silicon Valley wanted nothing to do with the government in terms of business, and even less to do with his organization. He said that it was a real problem — that these people won't even talk to us, and yet they control the throttle on the technology we need to do our job.

 He said we’ve got to figure out a way to get the most advanced technology without having to develop it all ourselves. He said he had a little group there thinking about it — would I go off and think about it, and let him know what I came up with? This other group would come up with a similar idea. The problem with why Silicon Valley didn't want to deal with the government was the toilet seat rolls of tape kind of problem — they just couldn’t waste their time on this, and so they wouldn't talk to the government. The question was how to arrange things so that they'll deal with you.

 The basic idea we came up with was for the government to own, but not control, an independent company that was not for sale to the private sector in any way. The purpose of this company was to promote technologies important to the Agency’s capabilities. We did set that up, and George asked me if I would be chairman or hire a board. We put together a fabulous board. George was a big help at doing this. Bill Perry was on it. Paul Kaminski, all names you’ve probably run across. Any Fortune 100 company would have loved to have that board. 

The first year and a half, we spent being investigated by everybody you could think of in Washington. They were asking what we were up to — creating a business that was going to get money from the government, but didn't have to produce a profit for the government or anything. The basic idea was that we would follow all the laws of the land, but none of the government's procurement regulations — we decided we would not build things ourselves, but we would go out and talk to people. We could find people with new ideas and give them contracts overnight, or we could give them a grant overnight, or we could take an equity position overnight — whereas the government probably couldn't do that at all, but if they did, it would take two years. So people liked to deal with us, and within a very short period of time we had 500 startups, coming to us with their ideas — one of which was Palantir. It worked beautifully. It's been a huge success, and I stayed for a few years.

CB: What does the name mean?

NA: “In” stood for information. “Q” was the name of the guy from James Bond, and “Tel” was for telecommunications.

CB: When you were chairman and CEO of Lockheed Martin, to what extent did the ghost of Skunk Works’ Kelly Johnson continue to haunt the company? What kind of impact did those men have on the company?

 NA: I happened to know him very well from my government years and my involvement with the SR-71. I knew him, Ben Rich, and Sherman Mullin, who ran it more recently and was actually a classmate of mine in college. All of those guys, particularly Kelly, had a huge impact. It was lasting. A lot of what he did was hard to carry over into the rest of the company, but there was great pride in all of his accomplishments. There was a lot of pride in the company. 

Kelly lived in a different world than we live today, but he was a  highly admired guy. He told me that he had worked on 32 different airplanes that flew, some within six months of the time they first drew the centerline on graph paper. He was doing things nobody else — certainly not at Lockheed — but nowhere else in the world could do at that time.

CB: Following the end of the Cold War, and with the resulting decline of the defense budget, why did the Department of Defense decide to force or encourage the consolidation of the defense primes? We went from dozens of major contractors to five.

 NA: The place to start here is a little before that, when we had the Reagan buildup. Defense spending had been increasing for a number of years, and an awful lot of people got used to it and thought that was going to go on forever. I guess I can say I was not one of those people. I'm kind of a history buff, and I knew it wasn't going to happen. 

The Cold War suddenly ended. None of us expected it, and within about four years the defense industry lost 40% of its employees and about 75% of its companies, all totally unexpected, and I'm not aware of any other industry that went through that sudden and unexpected change.

I say that because it was an existential and a very challenging and surprising time for the defense industry. The defense industry is known for its ups and downs, but few downturns had this impact, in terms of loss of employment, companies disappearing, the public environment, the congressional environment, and the media environment. The mood at the time, I might say, was “Russia's over.” Nobody had ever heard of China at that time. That's an exaggeration, but my first visit to China was 1978, and at that time there was just no threat from China, and by 2000 there still wasn't much of one. So it didn't look like there was going to be any real need for defense, and the United States has a long history of immediately getting out of a war, shutting down the military a few years later, then wishing it hadn't done that.

It goes all the way back to Tonkin Gulf. Suddenly, there was Tonkin Gulf, and all of a sudden we needed to build up for Vietnam. World War I: some guy gets shot on the streets in Europe, and suddenly there's a big war in Europe going on. Japan, suddenly out of the blue, attacked Pearl Harbor. So when the Cold War ended, the feeling was that there wouldn’t be much of a military for a while, for probably a long while, and so there was really no demand. We decided we were going to have what the nation called a “defense dividend,” and that dividend was going to come from cutting the size of our military forces and cutting the size of our industrial base. 

At that time, Bill Perry was Deputy Secretary of Defense, and John Deutch was an undersecretary — they were the ones who were really in charge of the Pentagon part of the world. They were all conscious of the history of suddenly rolling back the defense industry and the active-duty military. At that time there was also an understanding that nobody wanted to spend much money — we wanted a peace dividend from the Cold War. Congress didn't want to spend more money, the public didn't for sure.

So the guys running the Pentagon now had a choice. The choice was to try to keep all the companies going that were alive at that time —depending on how you define the defense industry, it might have been 15 companies, covering everything from tanks to ships to airplanes, you name it. That choice was to try to keep them all going in case they were needed in the future. The other was to recognize the fact that nobody wanted to spend any money on them. If you kept all 15 going, you were paying for a CEO of 15 companies, a CFO of 15 companies, factories for 15 companies that were only about 80% full. When you go through, and you add up all these costs, you were paying a lot for no output. If you argued for defense spending at that time, you were basically arguing for a very inefficient industry: overstaffed, empty factories, too many people, high overhead.

So, in my view, SecDef Aspin and DepSecDef Perry had no choice. Keep all the companies going and throw away a ton of money that people felt was needed for other purposes, or they could try to trim down the size of the industry to make it more efficient. They chose the latter. 

In my view, that was the only choice they had. I was out in the industry at the time and I didn't like that choice, obviously, but I tried to put myself in their shoes. I made the point in many speeches at that time that I'd much rather have 15 strong companies in the defense space than two, or three, or four, but I'd much rather have four strong companies than 15 weak companies. So I could get with the program. But it wasn't my first choice, because I'm a big history buff. I know that there's always a threat out there. You just don't realize how quickly it could appear. Ever since the first cavemen threw rocks at each other, humans haven't got along very well. 

I was fully prepared for whoever emerged next, and having traveled some in China, I thought that might be China — but I didn't know that yet. And anyway, Bill Perry decided that we would have to downsize the industry. 

So with all that as preface: one day, about 20 or so CEOs of major defense companies each got a letter inviting them to dinner at the Pentagon, in the Secretary's dining room. As far as I know, that had never happened before — there was a lot of curiosity about why we were going. The evening of the dinner comes, and they happened to seat me next to Secretary Les Aspin. I said, “These are really tough times in the defense industry, and we all like a free meal, but why are we here?” And he said to me, “If you wait a few minutes we're going to tell you — and I don't think you're going to like it.” That was my first inkling of where the Defense Department was going to step into this morass — which, to my knowledge, it generally had not done in the past, not in my lifetime. And so we had a nice dinner.

CB: What did you have for dinner? 

NA: I've been asked that a bunch of times; I can't remember. Anyway, we have dinner, and afterwards, we all go into the next room, which is a room between the SecDef’s conference room and his office. It's a small room, and there were metal folding chairs lined up for about 20 or 25 of us, whatever it was, no seating assignments. 

Everybody just sort of sat down, and the Secretary said that the Defense Department had some news they had to share with us, and that Deputy Secretary Bill Perry was going to do that. Bill gave the presentation, and the presentation involved a number of slides, which he projected on the screen. There were probably 10 of them, and Bill went through sort of what I just went through: that there's no fervor in the public at all to support defense spending, that the Defense Department was not going to pay for a bunch of empty factories, nor for every one of us in there to continue being a CEO. It’s a hard decision for all of us, but the defense industry was going to have to downsize — and the Defense Department was not going to do the downsizing. 

He said, “You guys are all well paid, it's up to you to figure out who survives, and we're not going to get involved, but it's going to have to happen, and so you better start figuring out what your strategy is.”

I had been worried about this kind of thing for a while. I’d been saying there were only three main options: consolidate, deconsolidate, or diversify. I made the comment at that time —  which I shouldn't have made — that the record of the defense industry diversifying in the past was a record unvarnished by success. We almost always failed, and this industry is just so different that it's hard to do something else. Without getting into what I was thinking at that time, the meeting ended with the comment: “You guys are going to have to do this — get on with it, and don't expect us to figure out who survives.” 

As we left the room, it was kind of funny. The reaction of the group was, “Wow, that's going to really be tough on you guys.” And nobody said, “It’s going to be tough on me.” I guess that's the way CEOs are. 

One of the charts had three columns and, as I recall, 16 rows — different kinds of defense equipment, fighter aircraft, submarines, tanks, you go down the list. The first column was the name of the product. The next column was the number of companies that at that time served the DoD in that category. The third column was the DoD's view of how many companies it could afford to sustain in that category. This is what really startled me. Tanks, it was maybe three in the first column, of how many they had, and how many they could sustain was one. There were four, five, six columns of the products where the number was one. That really stunned me. I believe in free enterprise and competition. I've worked in the government, I've studied other governments. Government — just like most things — is just not as efficient as industry, where you can get fired tomorrow, and where you have competitors who are very competent, and you have incentives, financial and otherwise. So, when I saw those ones there, I thought, "Oh my gosh.” And even then, it occurred to me: the next thing that will happen after one is that the government nationalizes that industry, and it becomes part of the government — and that's the worst of all possible worlds, because governments aren't very good at running factories, and there's no competition.

I was distressed by that. The next day I went back to the Pentagon and stopped by the controller's office — they had apparently prepared the charts — and I asked if I could have a copy of that chart, because I thought it was going to be a historical document. They said, "Well, it wasn't classified, so sure you could have a copy.” They printed one for me. Leaving the Pentagon, I ran into a reporter who I knew slightly, and he said, "What happened last night at the dinner?” I thought for a second, and I said, “Well, it was the Last Supper.” It just popped into my mind. It was going to be the last time that group would ever meet together, and we had supper together. So it was the Last Supper, literally. The name kind of caught on.

I went back and gathered the team and said, “What are we going to do?” The industry broke up into groups: one was going to do nothing and tough it out, one was going to buy companies, one was going to sell. And we had been worried about this problem for some time in our company — we had a head start on most folks. We had decided we wanted to buy.

The easy thing at the time was to sell your company. It was obvious the next five years were going to be miserable. And when you sell, you call a bank up in New York, get your big payout, then go to Florida. Most investors thought the defense industry was a terrible place to be, so there were few buyers, quite a few sellers. In our case, we said we're not going to participate in auctions on the advice of a friend, Warren Buffett. We also said we're going to buy quality companies. We're not going to bottom-fish for companies that are really cheap. We'll pay a fair price — not an exorbitant price, but a fair price — and we won't be at auctions, so we'll make you one offer. And nobody believed us about the one offer or the auctions.

But the companies started talking to each other — a lot of talk, a lot of steak dinners, not much happening, and I think the first deal out of the box was probably when we bought GE Aerospace; I was running Martin Marietta at that time. We put together GE and Martin Marietta, and the industry set out to do what the Defense Department had said. The sad part of the whole story was that the government didn't keep its word and backed out later, as usual. We ran into that early on, on the GE deal —  it wasn't Bill Perry and those guys, but the Justice Department and antitrust people. A couple of new people came in, one into Justice and one to DoD — a friend of mine by chance — who didn't like what was going on. The guys who were promoting what we had been told to do had all gone on to jobs outside the government.

Over the several years of major mergers, the country downsized a lot in terms of the defense industry. And the bottom line is there's a lot of criticism today of Perry and Les Aspin. Look at us here — we're facing China and Russia, and we're not prepared, we’re caught today with a terrible lack of equipment, a lack of companies to build it, and  a terrible budget problem. But we got there through good intentions. As you can gather, I'm not criticizing those guys. I didn't agree with them at the time — but I think they were right. There was no other choice.

CB: What do you know about the short-lived partnership on vertical takeoff and landing technology between Lockheed and Russia’s Yakovlev Design Bureau?

NA: I can't help you with that because it was before my time at Lockheed. There was a relationship between Lockheed Martin and Krunichev industries in Russia — the Russian company that built most of their ICBMs; Lockheed built some of ours and basically all the SLBMs. We knew about each other, but not a lot — and we viewed each other as enemies, obviously.

The Cold War came to an end, and the space launch vehicle business really cooled off, because foreign governments — particularly the French government, the European governments — had stepped in and begun to subsidize their industries. When that happened, it became very difficult for the US industry to compete in the launch vehicles sphere. Right after Lockheed and Martin Marietta merged — and I don't remember how the first move was made — it was suggested that Krunichev and Lockheed should get together as a launch vehicle business.

 We agreed to meet with people from Krunichev somewhere in the US to talk about forming a company that would launch space vehicles. The whole meeting was really weird — we were sitting across from people who were bitter enemies all these years, talking about putting a business together. Anyway, we did put together a little business for a while.

CB: Why did your attempt to acquire Northrop Grumman fall through? 

NA: That's the bitterest experience I have in my business career. Northrop was a logical candidate for Lockheed Martin to merge with, because they were in the airplane business when we were trying to horizontally integrate. They had a good reputation, but they hadn't been winning much new business, whereas we had been winning a lot. It looked like a good deal to both companies, and so we put together an arrangement, got our shareholders' approval. We sent our lawyers — 300 lawyers between us and Northrop — to talk to the government lawyers, who told us that there were some problems, but none were a major, deal-ending matter. 

The night before the meeting at the Pentagon, we had a meeting in the Northrop Grumman's Rosslyn office. There were about a dozen lawyers from all around the country that worked on this thing for a year. My question to each of them was, “What's the worst thing that could happen to us tomorrow?” They all said, “Well, we'll offsell the infrared business, and maybe this little business or that.” We went into the meeting the next day, chaired by John Hamre, who's a fine gentleman, an honorable man. The Under Secretary of Defense for R&D was brand new to the job, and there was a brand-new Deputy Assistant Attorney General for Antitrust at Justice. Anyway, there's a meeting, and this guy gets up — who's about three levels down in the Justice Department — and says the Justice Department has decided to stop the deal.

We almost died in the room. There was no shouting — it was silent. It was such a stunner. Kent Kresa from Northrop Grumman said, “Why?” And the man said: “Because we think Northrop Grumman has a great future, there's no reason for you to be absorbed.” And Kresa said, “Do you think I'm selling my company because I want to?” He said that the last fighter airplane of the kind Northrop Grumman had built was completed 30 years ago, and “we didn't even bid on the F-35 because we didn't think we were competitive, and the next version of the F-35 will be 30 years from now — and we can't just sit around for 60 years.”

In this case, the government had clearly lied to us. I said, “If you didn't want the merger, you should have told us so. And for the last year, you could have told us so, if not before.” And I said this all very politely. And the meeting ended.

CB: Do you think we've had our Sputnik moment with China?

NA: I think it's near at hand — and if China decides to grab Taiwan, it'll get very interesting. We're doing so many things wrong. I wrote an article two or three years ago trying to list some of the right things to do. We're doing just the opposite when it comes to defense capability. We're on the way down and just beginning to wake up — but this time it's a lot harder. It's a different problem this time. There were signs of this problem at the time of the Last Supper — that was one of the reasons I was worried, and why I think Deputy Secretary Perry did the right thing in downsizing somewhat. The big thing that we face, which we haven't faced before at this level, is the federal debt. 

We either have to gut the Social Security budget, or the public social budgets, or the defense budget, or we have to borrow a ton of money — which makes it worse — or we have to make the GDP grow by a huge factor that we've never come close to. We really are headed toward a train wreck. I hate to leave the story in those terms. We can always hope that China implodes like the Soviet Union did. Maybe it will, maybe it won't — but counting on your enemy to cave in is probably not a good strategy.

CB: What was the greatest aircraft that Lockheed Martin ever built?

NA: The SR-71. Not only because of the impact it had geopolitically, but because of the technology of the airplane. It was 15 years ahead of what anybody anywhere on this planet was trying to do, and they did it.

This interview has been lightly edited for length and clarity.

Since then, Dino Mavrookas tells me, “There have only been five major paradigm shifts in naval power since the beginning of human history. First, there’s the invention of the ship. Then, in the 1500s, you have sails and gunpowder and cannonballs. Then in the 1800s, you have steam and the pre-WWII battleship. On December 7, 1941, the day of Pearl Harbor, carrier-based aviation became the center of naval power. And then, between December 7, 1941 and October 29, 2022, not too much happened.” In those eighty-or-so years, he explained, “the ships became larger, more exquisite, more powerful, more capable, but nothing fundamentally changed. And in that process, they also became much more expensive, much harder to build.” What changed on October 29, 2022 was that Ukraine — a country without a navy and so forced to innovate — used an unmanned surface vessel (USV) to hit a Russian ship. “This was the first time an autonomous boat hit a naval ship — and you had the transition to maritime autonomy as that fifth paradigm shift in naval power.”

Dino Mavrookas is the co-founder and CEO of Saronic, an Austin-based startup building autonomous boats and ships for both defense and commercial use. In other words, he is ushering in the transition to maritime autonomy: the long-awaited fifth paradigm shift. Saronic incorporated six weeks before Ukraine’s navy-less military hit a Russian ship on October 29, 2022. “We were very much ahead of the market, and just believed that maritime autonomy was where the Navy had to go,” Mavrookas told me.

Vib Altekar, cofounder and CTO of Saronic, gives me the pitch for autonomous vessels, explaining that “Ukraine had no navy when this war started. Russia had the third largest navy in the world. And with no navy, Ukraine was able to detect, defeat, and mitigate a lot of the lethalities that the Russian Navy was putting on them.” This success of asymmetric warfare, where David prevails — or at least fends off Goliath — proved that affordable, autonomous boats and ships were a technological leap worth taking.

But the price asymmetry in autonomous naval warfare cuts both ways. If the Ukrainians, sans navy, could sink a third of the Russian fleet and neutralize the Russian naval threat, then a hostile power could easily deploy swarms of cheap autonomous vessels against American destroyers, large, multi-mission warships designed to protect smaller vessels and project force across the ocean that cost billions of dollars and years to build. The answer isn’t to keep building these destroyers. Saronic’s argument is that the defender can reclaim the advantage, but only by building cheaper, faster, and at scale. Altekar mentioned the large delta between the cost of a USV — which costs less than $1 million — and a destroyer, which costs $3 to $4 billion. “Nobody’s wealthy enough that they could cover a 1,000x, 2,000x difference.” He continued by articulating how Saronic had to foreground the economic reality of defense to be successful: “These are prohibitively expensive costs,” but these costs can be driven down by autonomous vessels. The solution is to build thousands of autonomous boats, and to do so faster and cheaper than anyone else can.

When Altekar was first introduced to Dino back in 2022, “the pitch was, we’re gonna build a ton of ships for the Navy.” The pitch didn’t take long. “And I was like, honestly, that sounds chill. Self driving cars are here. How hard can boats be? Turns out, they’re harder than I expected.”

Within the first 90 days of Saronic’s incorporation in September 2022, the company had a cooperative research and development agreement with the US Navy. At the time, Saronic had a thesis and a team — but no boats. “We partnered with the Navy from almost day one,” Mavrookas says. “We signed the contract within 90 days of starting the company. This was a completely unpaid contract. It was a cooperative research and development agreement, but it let us work alongside the Navy, understand the problem, understand the mission set, and start that development and iteration process.”

Saronic’s first model was a six-foot boat called Spyglass — ”not bigger than this table,” Mavrookas says. Spyglass — and all of Saronic’s vessels thereafter — were designed around the principle of scalability. “I don’t care what the design is, and I don’t care what the software does, unless you can build thousands of them. And then let’s work back: if we’re going to build thousands, what does the hardware need to look like? What does the software need to be capable of? So you just instill that into everything that we build. Now that six-foot product is actually sunset, but it built that fundamental DNA into our company.”

“Making existing shipyards 30 percent, 40 percent, 50 percent more efficient does not actually move the needle,” Mavrookas says. Building ships one-by-one is no way to catch up to China. “So you have to do something wildly different; that’s autonomous, that’s mass production.” Enter Saronic, building autonomous boats and ships at scale.

Today, Saronic’s smallest vessel is the Corsair, a 24-foot autonomous surface vessel capable of traveling over 1,000 nautical miles while carrying a 1,000-pound payload. Saronic iterated through the six-foot prototype, as well as a 14-foot version called Cutlass before landing on the 24-foot boat. “Range and payload capacity were what those smaller boats were missing.” The Corsair didn’t exist 19 months ago. Now Saronic is producing high volumes of these low-profile boats that carry cargo, sensors, weapons — you name it — in a payload bay. It’s fueled by a Volvo engine at the stern and a 270-gallon fuel tank.

As of April 2026, Saronic has the capability to “build thousands of Corsairs every single year,” Mavrookas tells me. Part of that construction occurs in Austin, Texas, where the company moved into a 420,000 square foot facility in October, 2024. Speed was of the essence. “We grew to over 25,000 square feet in our first nine to 12 months. Then we took over a 75,000 square foot building, which quadrupled our footprint. Three months after that, we were running out of space again, and so we started basically piecing together office buildings on that road. And so within 24 months of our company starting, we had committed to over 500,000 square feet of space.” In addition to the Austin facilities, in the past two years, Saronic “opened an office in DC, acquired the shipyard in Louisiana, opened 80,000 square feet in San Diego, an office in New Orleans, and opened Australia and UK offices” — the latter two have served as test sites for Saronic’s fleet. In east Austin, we walked through a few buildings in the seven-building main campus; and across East Ben White Boulevard, in a slightly older office, we saw the Corsair manufacturing lines in action (the original Saronic office was next to a Tesla repair shop, an Austin whiskey distillery, and a sizable pickleball complex).

While pointing at a Corsair in an open warehouse with a few dozen folding chairs and a Corsair where Saronic had their company Christmas party, Mavrookas anthropomorphized the boat. “Up here is kind of the brains of everything. This is where the compute lives.” I would tell you where he’s pointing at, but doing so would be a security risk. “This is where all the communications live. And then, you have the sensors and the cameras. So that’s the eyeballs. This is the brain. These are the legs, essentially.” Because the boat is built without humans in mind, Saronic is “redesigning what a ship actually is from the inside out.”

Saronic’s largest offering — one that passes the threshold from “boat” into “ship” territory — is the Marauder; at 180 feet, it can no longer be built inland. The distinction matters when building these vessels. Anything that you can build inland, build here in Austin, you can put it on the back of a truck, you could ship it anywhere in the world, that’s a boat,” Mavrookas explains — “anything that has to be built on the water, that’s a ship.” The Marauder can carry four ISO shipping containers and operate autonomously for extended periods at sea. In between is the 52-foot Mirage, which can travel over 2,000 nautical miles carrying a 2,000-pound payload (double that of the Corsair). A fleet is being assembled, designed in the landlocked city of Austin, Texas.

The Saronic Gulf is the body of water that separates Athens from the Peloponnese. The second Greco-Persian war was fought in the Saronic Gulf. “This was the battle of Salamis,” Mavrookas explains. “The Greeks split up the Persian fleet and were able to beat the Persian fleet, which was a much larger fleet. It was this really fascinating story, very reflective of what we’re doing here at Saronic.” Saronic’s titular nod to Greece is also personal: Mavrookas’ father immigrated from Greece to New Jersey, where he opened a Greek diner that his son worked at. After working in his father’s diner, he studied computer engineering at Rutgers. 9/11 was his junior year of college; soon after, he walked into an FBI career day event and said that he wanted to do tactical operations. “They said, ‘You need military experience.’ I said, ‘Okay, what does military experience look like?’ Long story short, I learned about the Navy SEAL teams from a Navy recruiter.” The day Mavrookas graduated, he enlisted in the Navy, where he served for 11 years; his last five were spent in SEAL Team Six.

In 2003, the year before Mavrookas enlisted, the US Navy fell below 300 battle force ships — the ships counted towards the Navy’s stated size — for the first time since the early 20th century. As of October 2025, the Navy had 293. China has more than 370, with projections of 435 by 2030.

After the Cold War and the collapse of the Soviet Union, the defense industrial base shrunk, thinking that the end of history would usher in a new age of globalization and commercially facilitated peace, making great power conflict a thing of the past. To make room for this miracle of capitalism, the defense budget as a whole shrunk by a quarter; “the Navy’s budget actually shrunk by 40%,” Dino explains. “You had decreased competition, decreased investment, you had closings of shipyards — all of these things were happening because of globalization, because you can go and build ships cheaper in China.” Throughout the 1990s and 2000s, China, rather than being viewed as our greatest geopolitical rival, was generally thought of as America’s “everything-factory,” a neutral ground where things were simply made. The dream of globalization made it unthinkable for the United States to do the dirty work of shipbuilding in house when it was easier — and cheaper — to have that same work done in China. At the end of history, suggesting that offshoring these critical defense functions would be a problem down the line was verboten as it implied the idyll of mutual economic dependence might end. (Similarly, the US imports a quarter of our enriched uranium from Russia because we have offshored the dirty work of uranium enrichment as well.) Now, China builds about 1,000 cargo ships a year; the United States builds about three. Throughout our conversation, Mavrookas kept repeating a figure from the Office of Naval Intelligence that, measured by tonnage, China’s shipbuilding capacity is 230 times greater than that of the US.

“China had ‘a backwater navy’ only half a century ago,” Mavrookas explained. “Right now they have aircraft carriers, nuclear submarines and hypersonic missiles. They’ve done all of this in a period of 30 years, because there’s an intense focus from the government on pouring capital into that space.” State capitalism has allowed China to pump money into national defense. In the US, by contrast, the drowning shipbuilding industry can be understood as a lack of financial incentives. As Altekar put it, “we spend a billion dollars a year on technology that is worse than an iPhone that’s going into the pocket of soldiers.”

The United States has the largest defense budget in the world. But compared to rapid improvements in aerial drones, intelligence software, and missile systems, naval defense stagnated after the Cold War, deemed wasteful without a great power to compete with — and the stagnation has fed on itself. Fewer dollars going into maritime research and development meant less innovation, which meant fewer investors willing to put their money into research and development, which meant less innovation. Altekar told me that “the economics never really worked out for people to innovate on the boat or the ship itself.” The commercial incentive to innovate was thin. Plus, the post-9/11 conflicts in the Middle East oriented the American defense establishment towards fending off ground-based insurgency groups, not building fleets to fight peer competitors at sea. In the past 10 years, it has become more and more obvious to Washington that the US is only as strong as its navy in a fight against China over the island of Taiwan. However, this realization alone has not created market incentives strong enough to change the status quo. Saronic is trying to reverse-engineer those incentives.

Get the Mag in Print.

Arena publishes four stunning print editions per year, full of stories just like this one on American technology, capital, and industry.

Despite the ocean’s indifference to human affairs, there’s unquestionably an allure to the open sea that makes us want to explore it. Altekar recalled: “Humans had canoes in 7000 BC, right? That predated the wheel. The wheel came out in 3500 BC.”

Since human beings have had fleets at their disposal, Altekar tells me, naval supremacy has been defined by having “large ships, lots of ships, and a very skilled set of people that can operate those ships.” Navies are expensive: “they’re large, they’re complex, they’re built for humans and the supply chain.” Mavrookas agrees, telling me that “autonomous ships that don’t have to support people allow you to strip out 80 to 90 percent of the complexity of the platform.” There don’t need to be hallways, bathrooms, nor kitchens. Since much of the expense of shipbuilding comes from making them fit for human crew, autonomous ships are actually an economical option once the tech is sorted out. “Our ship,” Mavrookas says, “we can break down to seven major subcomponents,” compared to “thousands” on a naval destroyer.

Removing humans from the equation unlocks the possibility of what a boat or ship can do. Unlike autonomous cars, autonomous boats get to change the physics of the vehicle at hand. Waymos, Altekar explains, “don’t have increased fuel efficiency or energy efficiency because the car is self-driving — maybe a little bit, but the same amount that you get from cruise control.” Autonomous boats and ships, by contrast, are not floating hotels like their old-school counterparts; they don’t need gyms, bathrooms, kitchens, and so on, which allows more space for engineers and designers to reimagine what a boat or ship ought to look like. “When you design the ship for there to be no humans at all... suddenly you can take a sharp 90-degree turn, take three and a half lateral G’s, like an F1 car. And it doesn’t matter. You don’t need Lewis Hamilton in there.”

Common parlance in start-up world is the “X for Y” formula: for example, “Uber for food” (DoorDash), or “Netflix for fitness” (Peloton). At first glance, one might call Saronic, a company that builds autonomous boats and ships, “Waymo for the ocean.” But even that characterization would be missing an important part of the point, and the ingenuity of Saronic’s tech: while autonomous cars are designed to be both more pleasant and safer for human passengers, autonomous boats cut out the idea of passengers — or, in this case, a “crew” — entirely.

What makes Saronic’s task far more difficult than, say, Waymo or Zoox — both marvels of engineering that have required decades of sustained software and hardware engineering, is that there is no ‘training data’ about maritime navigation and sea conditions for autonomous surface vehicles to begin with. Altekar explained how “Lyft would publish its own data, Waymo publishes data sets.” On the seas, no such foundation existed — before Saronic. He continued: “The economics don’t work for a large ship within the Jones Act” — the century-old law requiring that cargo shipped between American ports must travel on American-built, American-crewed vessels — “to go, ‘let’s put $10,000 of LIDARs on this thing.’” (The Jones Act protects a small domestic fleet of just 92 ships from foreign competition — but without competition, there’s no pressure to innovate.) In order to teach its vessels to navigate the open ocean, Saronic had to go collect this data itself; the “MOPs” (mission operations) team within Saronic now operates round-the-clock testing in Galveston, an East Coast test site, Sydney, and the UK. The ocean has no traffic laws, but it has difficult physics. An unexpected wave or a sharp turn can throw an entire boat to one side. And all of the electronics on a ship, down to the plastic fan inside a computer, must be able to withstand pulling five Gs at once. Saronic’s software engineers can push software updates to their vessels from anywhere; their growing team of software engineers push code from Austin.

Another reason why Saronic’s task is so much harder than Waymo’s is the water environment itself. Altekar — who grew up in the Bay Area and developed an interest in defense tech through his time working as an engineer at Anduril — has a sailor’s respect for, and deference to, the ocean. He puts it bluntly: “Putting hardware in the ocean means that your hardware will break. Salt water is cancerous for electronics. The ocean can, at any given point in time, do whatever it wants to you.”

One of the benefits of building in the ocean is that “the ocean is vast and sparse,” Altekar says, so there’s no need to take pedestrians or cyclists into consideration. But the same emptiness that makes things easier also means there are no street signs or lane markings to guide the way. The vast, open sea that drove many a sailor to madness is the same sea that Saronic’s vessels must navigate through. Worse yet, different classes of boats must turn, accelerate, and slow down extremely differently — so that data collected from one vessel is not easily applied to another. It’s much harder to transition data from a tanker to a destroyer to a Corsair than it is to transition comparable data from a sedan to a sports car to a minivan. Saronic’s solution to this problem has been to design its fleet from a “software-native perspective” from the beginning” — Saronic’s software engineers can see their code tested on the water within 24 hours.

What makes Saronic’s task far more difficult than, say, Waymo or Zoox — both marvels of engineering that have required decades of sustained software and hardware engineering, is that there is no ‘training data’ about maritime navigation and sea conditions for autonomous surface vehicles to begin with.

Saronic’s boats and ships are totally vertically integrated. The management systems for the factory lines are designed in house; so is the software running on all of the ships. Walking down the factory floor, you can see in real-time what percent of each vessel, built from scratch in Austin, Texas, is complete. Saronic has a machine shop in house, a composite shop in house, and a wiring shop — where individual wires and cable assemblies are made — in house. “As the manufacturing leader,” John Morgan tells me, “I want to be able to control my own destiny, right? And the only way you can do that is being able to control the machine shops, the wiring houses.”

Saronic’s manufacturing chief John Morgan got an education in the “school of Elon,” as he puts it. He joined SpaceX in 2013, where he ended up running the final integration line on the Raptor engine. Walking through one of the production floors, which had two parallel production lines and a dozen or so stations on each line, Morgan remarked that the linear production lines “very much came from Elon. Just walking down this production line, you inherently understand where the key bottlenecks are or what stations are running behind because you’ll see a gap in the line, and that allows you — five minutes from the first time that you walk in and got your coffee — you immediately understood where your problem is on the production line.” We were walking through the center of two Corsair production lines, with factory workers on each station. Each step was additive; as you walked down the factory floor, you saw each step of the boat’s manufacturing process. By the end of the line, the Corsair was nearly complete.

As we walk through the production lines, Morgan tells us that the factory was once a concrete mixing plant. When they moved in, “every single one of these surfaces had an inch of concrete on them.” They cleaned up the factory and made it “high, white, and bright,” a production line that both adds the air of a best-in-class work environment and allows for inefficiencies to show themselves. Later, Morgan let us in on a little secret: “the color tones are actually modelled off the Ferrari factory, the light gray and gray — you want people to come in and think, ‘wow, I’m very proud to come here and work.’”

The facility is highly organized. Every part has a bar code, a tracking number. “We can go and track every single thing down to the nut and washer. Because, let’s say I can build a boat every day, but I can’t get one harness” — the wiring that connects sensors, antennas, and engines — “or this composite structure every two days. Guess what? I’m only building a boat every two days.” Traceability, Morgan explains, is key to improving throughput and getting vessels built quickly. Harnessing — the wiring system — in particular had a lead time of eight to 12 weeks — “that’s just something that we’re not going to accept.” So they built a harnessing center in house; vertical manufacturing at its best.

Once a new boat is complete, Morgan explains, the testing team goes to Saronic’s lake in the outskirts of Austin where they do “mops almost 24 hours a day, seven days a week.” In testing a new boat, they “really push the needle on what this boat can do.” When I asked him how Saronic tests for enemy strikes on the boat, he reminds me that Saronic is dual-use, providing its boats and ships for both commercial and defense customers. The Corsair just completed 100,000 nautical miles of internal testing.

Part of what makes Saronic so impressive is just how fast the startup has moved. Both in the sense of corporate growth — the three-and-a-half year-old startup now has over 1,400 employees, the majority of them on the manufacturing floor — and in the sense of the sheer number of boats and ships Saronic has built. Morgan gives his own definition for boats: “we think of boat building as anything that’s 52 feet and below, anything that can be road transportable.” Shipbuilding, echoing Mavrookas’ definition, “is purpose built on a body of water, usually an ocean, to go and start building those things.” China did not will the American shipbuilding industry to obsolescence; American shipbuilding died for a reason. In part, American shipbuilding died because it failed to adopt 21st century manufacturing techniques — shipyards are old, and neither the private market nor the government thought to invest in updating them. But the “school of Elon,” to borrow Morgan’s phrase, teaches that it is possible for American industry to compete with China — and out-build our allies in Japan and Europe — by building smarter — by simplifying and lowering costs across the board.

For shipyards writ large to update, however, that requires pausing production lines, which are already struggling to meet demand. Port Alpha is Saronic’s vision for a “greenfield” facility, Morgan says, “purpose-built with the ability to go and expand as new techniques come out.” Saronic is yet to announce the site of the shipyard of the 21st century. The Port Alpha site will provide the infrastructure to build larger ships. Mavrookas told me that, “as you build more large vessels, more of that commercial market opens up, and eventually we want to be building cargo containers and bulk carriers and oil tankers.” Saronic’s CEO told me that Port Alpha and Saronic’s other manufacturing efforts will “create thousands of jobs, up to 10,000 jobs over the next decade.”

“In this facility” — the Austin facility that Saronic moved into in 2022 and has since outgrown — ”we have the capacity to build thousands a month” of the Corsair, Morgan tells me. “In Franklin, we’ll be building 20 Marauders a year. I’ll tell you, 20 a year is some of the highest throughput rate of any shipyard in the United States.”

The boat is being built by Sampriti Bhattacharyya and her team at Navier, a six-year-old maritime startup in Alameda, California. The company is named for the Navier-Stokes equation, which describes the motion of viscous fluids. Its flagship product, the N30, is the United States’ first all-electric hydrofoiling boat: a thirty-foot craft that cruises above the water at twenty knots on three actively controlled foils, reducing drag by roughly 90% and cutting operating cost from around four dollars per nautical mile to 38 cents. The first production unit was delivered in October 2024 to Rich Ross, a lifelong American boater who keeps it at his home on Hilton Head, South Carolina. Earlier that year, Navier launched a commuter pilot with Stripe, shuttling employees between Larkspur in Marin County and South San Francisco on a route that takes 90 minutes by car and 30 by water.

In October 2025, Navier unveiled three new vessels: the thirty-foot Quanta-D, the eighty-foot Valkyrie, and the 120-foot Morpheus. All hybrid-electric; all hydrofoiling. All built on what the company calls the Generalized Marine Vessel Platform, a single chassis-and-software architecture that can be configured as a luxury day boat, a commuter ferry, a cargo vessel, an unmanned patrol craft, or a sensor-laden escort. The Quanta-D, a hybrid-electric variant, is the vessel beginning sea trials this spring. Defense customers include Leidos, a multi-billion-dollar defense IT and engineering contractor that has been quietly building one of the deepest portfolios in maritime autonomy, and the Department of War. Navier has raised over $30 million publicly and reports multimillion-dollar revenue across defense, commercial, and recreational sectors.

✺

Sampriti grew up in India, went to what she calls a “teeny weeny” college — not IIT, nothing close — and got her first airplane ride to America on the back of one cold email, sent among several hundred to American labs and graduate programs. She had wanted to work in space since she was 13, when cable television arrived at her house and she saw footage of the Apollo missions for the first time. Space was her first love and she reached for it.

It took her to Fermilab, a particle physics and accelerator laboratory 40 miles west of Chicago, where she worked on experimental high-energy physics with the Department of Energy. Then to Ohio State, where she earned a master’s in aerospace engineering, worked on flight controls at NASA over the summers, and wrote her thesis on accelerator-driven subcritical reactors — a design she says she chose specifically so that America could be energy independent. Then to MIT, where she completed a PhD and built a fleet of small underwater drones called Hydroswarm, originally built for inspecting nuclear reactors and later adapted for ocean mapping, then for mine countermeasure and submarine situational awareness. The intellectual property ended up at one of the largest American defense primes. Sampriti was 28.

Somewhere along the way, working in space stopped being the goal. “The purpose of life is to create,” Sampriti tells me. The reach for the moon had been about the wonder. The wonder, it turned out, was closer to hand. The ocean covered 70% of the earth’s surface, had been mapped less thoroughly than the surface of Mars, and nobody was paying attention to it. She mentions Palmer Luckey, one of the few other founders she remembers being early on defense work when early was unfashionable, before American Dynamism was a marketing strategy.

✺

Hydrofoils are not new. They were patented in the early 1900s. The US Navy started building its largest hydrofoil class in the 1960s. The boats flew but the program died. The reason the program died is, to Sampriti, the most important technical fact about Navier, and she offers it unprompted. A hydrofoil is a boat with underwater wings that generate lift at speed, pulling the hull clear of the water. Drag goes with it. An active hydrofoil — one where the foil angle is continuously adjusted mid-flight to hold the boat stable in chop — requires three things: sensors to read the water, computers to process the read, and actuators to move the foils. In the 1960s and 70s, all three were, in her words, “really crappy.” The boats were unreliable and the maintenance was punishing. The last Pegasus came out of the water in 1993 and the Navy did not return to the concept.

What is new, three decades later, is not hydrofoiling. It is that sensing and compute and actuation got good and cheap enough to support the physics. On the N30, the flight control system adjusts the aileron flaps on each foil up to 50 times per second, reading wave conditions in real time. The hardware that does this work would have cost a fortune in 1965.

“We solve the physics tax,” Sampriti says. A conventional hull pushes water, and pushing water burns most of the fuel it consumes. A foiling boat climbs above the water, rather than pushing it, thus the tax disappears. The other thing the physics unlocks is stability. Most of the venture money flowing into maritime defense right now is going to companies building conventional vessels — boats that sit in the water like every other boat. Sampriti makes a specific claim about them: not one of them, she says, can hold a counter-UAS payload in sea state 4 or 5 (swells of roughly four to thirteen feet). The hulls roll and pitch too much in moderate swell. This matters because countering enemy drones — increasingly the dominant threat to surface vessels in contested waters — requires a platform stable enough to track and engage them. A boat that can’t hold its sensors steady is, for that mission, useless. A Navier boat can, because the same foil that saves fuel also damps the roll and pitch. “You would need a destroyer or like three times bigger vessel to get the same stability,” she says.

Navier’s objective is to move a unit of payload per unit mile in the fastest, most cost-efficient, most reliable way possible. Everything on top of the GMVP — the hull and flight-control system that are Navier’s foundational layer — is payload. The platform is purpose-built for three constraints: cost across building, scaling, and operating; range per unit payload per unit mile; and sea-state adaptability. Everything else follows. “Follow the mission,” she says. “Don’t get married to a solution.”

Sampriti puts it directly: “We are building the Nvidia chip of the marine industry.” The hull and the flight control software are the core architecture. What gets built on top, whether a luxury day boat, a cargo deck, a USV, or something else, is the application layer. “You can put a ferry superstructure on us,” she says. “You can put a cargo ship, a naval.”

This is where most maritime companies do something Navier refuses to do: build a bespoke vessel per customer. The traditional American shipbuilding industry, currently ranked 19th in the world by output, is structured around one-off vessels built by hand over multi-year timelines. The Navy’s Constellation-class frigate program, contracted in 2020 with a planned delivery in 2026, slipped three years to 2029 and ran $1.5 billion over budget before the Navy cancelled most of it in November 2025, capping a planned 20-ship class at the two hulls already under construction. According to a Congressional Research Service report, it takes around two years to build a new oceangoing Jones Act–compliant vessel — and US shipyards typically deliver only two or three of them a year.

Navier is trying to invert that by borrowing from industries that have already solved mass production. Its supply chain is stitched together from aerospace and automotive, not legacy maritime. Navier’s hull is almost secondary to its flight control system.

If the Generalized Marine Vessel Platform works as advertised, the consequence is structural. One chassis serving multiple markets means the commercial fleet scales the manufacturing that makes the defense fleet cheap. The flight-control code that reads wave conditions 50 times a second on a Stripe commuter boat is the same code that can run on the Quanta-D through the Strait of Hormuz. The foil geometries, the actuator hardware, the sensor stack, the redundant propulsion systems are all shared across the fleet. It is how Navier could, in principle, make a defense platform that actually costs what Sampriti says it should cost.

Most defense contractors subsidize commercial R&D with defense contracts. Navier, if their platform bet works, will do the opposite: subsidize defense production economics with commercial volume. That is a structurally different company from anyone else in the sector — and the mechanism by which the slogan from Sampriti’s recent Axios interview, “out-innovate, not out-produce,” has a chance of producing a real supply chain.

While the consensus in defense-tech discourse right now is “cheap vessels,” Sampriti thinks that is only half the battle. Fuel, she points out, dominates lifetime cost. A boat that costs more to build but consumes just a fifth of the fuel of an ostensibly “cheap vessel” is the actually-cheap boat over 20 years. Today, escorting a multi-hundred-million-dollar oil tanker through the contested Strait of Hormuz requires a destroyer. Sampriti’s pitch is a distributed presence of smaller, better, cheaper vessels — four or five Navier boats carrying ISR, counter-UAS, electronic warfare, and minesweeping payloads, continuously on station, refueling less often than anything in the category can.

Navier boats have already been delivered to the Gulf region, where they are being evaluated for naval operations in the Strait of Hormuz.

✺

Sampriti is a first-generation American immigrant building, among other things, for American naval power. The obvious question is why.

Much of the American Dynamism rhetoric is reactive, defined against a rising China, against peer adversaries, against the specter of American decline. Sampriti’s version predates the wave and runs in the opposite direction. Her vision is positive in structure rather than negative. She loves the country because of what it lets her do. She arrived in Illinois with no family, no network, no money, and it worked out anyway. “Nowhere else in the world,” she says, “I could have done what I have been able to do.” She mentions twice that American women could not have credit cards in their own names until 1974; the opening she walked through is, historically speaking, five minutes old, and worth protecting on those terms alone.

Sampriti believes the logic from there is self-evident. If America stops being the dominant world power, all the alternatives are worse. Therefore American power is worth defending. Therefore defense is worth building.

A few years ago, many tech investors told her that Navier was outside their thesis. She will not name them on record. “The same venture capitalists who are very patriotic today could not have imagined funding a defense tech company,” she says. “I definitely have the emails.”

Another VC told her he would have funded Navier if she had been doing just defense, because defense has higher margins. She declined. “Yes, you can make more money on defense,” she admits. “But to me that was pretty unpatriotic. What will actually make American defense win,” she says, “is a low cost maritime platform that is formidable at sea and available at scale.” Extracting money from DoW grants and contracts is not the same as delivering real maritime power.

There is something particularly American about the distinction. The country’s economic mythology is built on competition in markets — building things people want and then winning by building them better. The country’s defense mythology is built similarly: World War II–era industrial mobilization, the postwar aerospace boom, the chip industry that grew out of dual-use research. By Sampriti’s logic, the way to be patriotic in defense is to compete on the merits — to build a platform that outperforms in commercial markets and earns its way into government use. The way to be unpatriotic is to treat the Department of War as a procurement window where margins live untouched by competition.

✺

Sampriti’s long-term vision for Navier, and America, goes beyond defense. Global ocean logistics today is centralized in a few enormous, slow vessels — what she calls the mainframe era of shipping. She wants to end it. The replacement she has in mind is a network of smaller, faster, distributed boats: hydrofoiling vessels running point-to-point across coastal cities, ports, and trade corridors at a fraction of current cost. Ships today are big and slow because drag scales as velocity squared, and efficiency-per-ton requires size. That is why global logistics has been a choice between thirty-day ocean freight and one-day airfreight, with nothing in between. Hydrofoiling breaks the scaling law. If efficiency comes from the foil rather than size, you can build smaller and faster without paying the cost penalty.

Roughly half the world’s population lives in coastal cities. The majority of global economic activity happens in them. “We are thinking of spending billions of dollars trying to build air taxis,” she says, “and we are stuck and not building networks on the water because we haven’t figured out the physics of it.”

“Alameda” — a coastal city built on an archipelago across the Bay from San Francisco — “is such a great place to live,” Sampriti says. “If you are going to work at Oyster Point, you should totally be able to live in Alameda.” For anyone who has sat in Bay Area traffic on a Tuesday morning, or paid San Francisco’s steep rents, the geography starts rearranging itself against the reader’s will. Hence the Stripe commuter boat.

“The world moves at the cost and speed at which goods and people move,” Sampriti recites. She’s almost certainly said it before, in other interviews and presentations, but the reminder seems appropriate. Her company has even signed a hundred-vessel contract to build a new transit network.

Sampriti says she would “risk the last pieces of everything I have to build Navier.”

I consider believing it. She grew up, she says, in “chaotic” circumstances. She says she was rebellious and ran away from home more than once but doesn’t explain why. On a backpacking trip in the mountains, she tells me, she came close enough to being killed by a landslide that she now uses the incident as shorthand for a worldview: “life is a game of inches.” She has no family in America. She frames the work as a video game. Higher level, harder game, she says — “that is the point, you chose it.” She is not claiming to try to finish the infinite game. But she does not give up on what she believes in.

In less than 60 days, the Quanta-D begins sea trials. If the Atlantic crossing succeeds later this year, it will be the longest autonomous voyage by an unmanned hydrofoiling vessel and one of the longest by any vessel of its 30-foot size class. (The 2022 Mayflower autonomous ship, which made the first autonomous transatlantic crossing by a powered boat, was 50 feet.) Navier will have delivered, at scale, the proof of an argument Sampriti has been making since Hydroswarm: that the ocean is the frontier that has been waiting for a platform shift, that the physics tax holding back maritime can be repealed, that a thirty-foot boat with the right software can do work that used to require a destroyer (or was not done at all). And if it fails, they will fix it and try again.

Somewhere on the other side of the country, in a warehouse in Alameda, the boat is almost ready.

The Metals Company, based in Vancouver, British Columbia, was seeking permission from the US government to mine the 38 different elements contained in polymetallic nodules. Barron calls the nodules “batteries in a rock” because they contain many of the critical minerals needed to manufacture batteries, including nickel, cobalt, and manganese.

The target for this mining is a 104.5 million acre stretch of seabed between Mexico and Hawaii known as the Clarion-Clipperton Zone (CCZ), where US Geological Survey estimates suggest deposits contain more nickel, cobalt, and manganese than all known worldwide land-based reserves combined. The full CCZ is estimated to contain up to 30 billion metric tons of nodules — a deposit, at current valuations, worth up to $18.4 trillion.

The Metals Company plans to vacuum these nodules from the seafloor using robotic collector vehicles, haul them to the surface, and ship them to shore for processing into battery-grade metals for electric vehicles (EV) and battery energy storage systems (BESS), large-scale installations that store electricity for grid use. EV and BESS are the core hardware of electrification, the wholesale replacement of fossil fuel systems with electric ones.

Due to EV and BESS adoption, critical minerals demand is accelerating beyond what the existing supply system was designed to handle. The EV industry required 2.2 million tons of nickel, manganese, lithium, iron, graphite, and cobalt to supply newly sold EV batteries in 2024, when EVs comprised roughly 20% of global new car sales. By 2035, EVs could account for as much as 70% of global new car sales. The critical minerals supply system has no plausible path to keeping pace.

In the US alone there are 570 gigawatts of BESS projects in interconnection queues waiting to be added to the grid. That queued capacity is equivalent to enough to power roughly 513 million American homes. The International Energy Agency (IEA) has projected that demand for battery metals could grow by a factor of 30 by 2040 from 2024 levels.

A recently published report by the energy think tank Ember put the levelized cost of storage (LCOS) for utility-scale battery systems at $65 per megawatt-hour (MWh) in globally competitive markets. For comparison, new nuclear capacity in the United States runs between $130 and $230 MWh. New-build combined-cycle gas turbines have seen equipment prices rise sharply, driven in part by demand from AI data centers, pushing their global levelized cost to $102 per MWh, the highest on record, according to a Bloomberg New Energy Finance report published this year.

Meanwhile, battery storage is moving in the opposite direction. BNEF projects that BESS hardware costs will fall another 35–55% by 2035. At that level, the economic case for new gas-fired power plants collapses because storage can undercut gas on price while performing the same grid-balancing function. But this plummeting cost curve assumes a steady flow of raw materials, an assumption that currently rests on shaky ground.

The biggest risk to this energy transition is a lack of mines. The IEA estimates the world needs 80 new copper mines, 70 new lithium mines, and 70 new nickel mines to meet projected demand. Historically new copper mines take 15-20 years or longer to come online. Closing this gap through conventional mining alone is functionally impossible.

New battery chemistries are expected to reduce reliance on critical minerals, recycling is supposed to supply a growing share of demand, and substitution can ease pressure on constrained inputs. But none of it moves fast enough. Recycling cannot expand meaningfully until there is a large volume of end-of-life material — a large stock of dead batteries — to recycle.

Substitution brings performance tradeoffs, especially in applications where conductivity and energy density matter, and those limits show up quickly in practice. Demand forecasts for critical minerals may turn out to be somewhat overstated, but even conservative scenarios still imply a supply gap that is difficult to close with land-based mining alone. The problem is time.

The structural supply shortage of the materials we need to electrify our economies will redraw the map of global power. Instead of Saudi Arabia and other petrostates holding the world politically hostage, power could shift to what we might call electrostates that possess or control the critical minerals needed for electrification. The inversion is already underway. Argentina, Bolivia, Chile, and Brazil, which together hold roughly 65% of the world’s lithium reserves, have floated creating a “lithium OPEC” to control the market. Saudi Arabia has announced plans to invest $100 billion in critical minerals mining projects over the next decade.

The largest electrostate is China. Chinese companies control significant shares of global cobalt and manganese extraction in Africa and elsewhere and have locked up supply through overseas mining investments as part of a deliberate industrial strategy. A single Chinese company, CMOC, accounts for 24% of the world’s cobalt mining, while two Chinese firms, Isky New Minerals and Guizhou Dalong Huichen, each control more than 20% of the global manganese market China is also the dominant refiner for 19 of the 20 minerals analyzed in the IEA’s Global Critical Minerals Outlook 2025. The country manufactures more than 80% of the world’s finished batteries and controls over 98% of lithium iron phosphate battery cell production, the cobalt-free chemistry that has become the dominant and fastest-growing battery type globally.

What Barron was offering President Trump, in his own words, was “an amazing way of catching up from what is a very distant second place to China when it comes to critical minerals.” Four days after that Oval Office meeting, Trump signed an executive order directing the US government to expedite seabed mining licenses in international waters, unilaterally stepping outside the framework of the International Seabed Authority, the UN body that has ostensibly governed the ocean floor since 1994, and the Law of the Sea Convention that underpins it.

China’s foreign ministry condemned the move as a violation of international law, but this condemnation carried more than a little hypocrisy because Beijing has been more aggressive than any other national government in pursuing deep-sea mining. It is a bit like complaining someone is planning to rob the bank that your crew has already been casing.

Polymetallic nodules were first discovered in the Atlantic Ocean off the Canary Islands in 1873 by the British HMS Challenger expedition. The chief scientist on board wrote of the discovery in the scientific journal Nature, but the rocks were regarded as novelties and sent to the British Museum in London for display.

Deep-sea mining wasn’t seriously considered as a commercial proposition until 1965, when John Mero, a geologist at the University of California, published The Mineral Resources of the Sea. The book, combined with Cold War superpower resource competition and Space Age technological optimism, created widespread interest in deep-sea mining.

In 1967, Arvid Pardo, Malta’s ambassador to the UN, pushed for international regulations that would treat the deep seabed as the “common heritage of mankind” after learning about deep-sea mining at a cocktail party. In Pardo’s vision, deep-sea mineral extraction would finance utopian underwater cities where humans would be protected by air bubble curtains and farm fish, with trained dolphins acting as sheepdogs. He got something rather different. Due to Pardo’s efforts, the UN declared the seabed beyond national jurisdiction in 1970, with these principles codified in the 1982 UN Convention on the Law of the Sea — a treaty Pardo would later call “probably the most inequitable that has ever been signed in the world,” the common heritage of mankind having been reduced, in his words, to “a few fish and a little seaweed.”

Interest in deep-sea mining peaked during the 1973-1974 Arab oil embargo period, when commodities experienced a price “superspike.” Gasoline prices that had been hovering around 34 cents per gallon pre-embargo shot up to 84 cents per gallon, and the price of gold, freed from its fixed exchange rate after Nixon abandoned the gold standard, surged from $42 per troy ounce in 1973 toward $200 by mid-decade.

Aiming to increase supply, the defense contractor Lockheed and Standard Oil of Indiana (Amoco) began pilot mine testing in the CCZ (the nodule-rich stretch of Pacific seabed between Mexico and Hawaii) at that time; Lockheed still owns CCZ leases today. In 1970, the CIA recruited the eccentric billionaire Howard Hughes to provide a cover story that he was building a deep-sea mining vessel called the Hughes Glomar Explorer to explore the CCZ. The ship was actually built as part of Project Azorian to recover a Soviet ballistic missile sub that had sunk to 16,500 feet beneath the ocean near Hawaii. The Los Angeles Times uncovered the plot in 1975 and when journalists pressed the CIA, it issued the now famous response “it could neither confirm nor deny” the allegations.

The Howard Hughes scandal marked the high point of 20th-century interest in deep-sea mining, which would collapse along with the prices of metals in the post oil embargo period. Until recently, there was never enough of a sustained economic “why” to pursue it. The rise of electrostates and the US-China superpower competition is changing these dynamics. But economic and ecological risks remain constraints on development. The deep ocean is the largest habitat on earth and the least understood. Humans have mapped more of the surface of Mars than the floor of the Pacific.

At 13,000 feet underwater, in permanent cold and dark, ecosystems have organized themselves around the nodule fields over billions of years. Xenophyophores — single-celled organisms that can reach the size of a dinner plate and rank among the largest individual cells on earth — live on and around the nodule fields, many attached directly to the rocks themselves, using sediment to construct protective coverings. So do sea cucumbers, brittle stars, and fish species never formally catalogued. Of the estimated 10 million species thought to inhabit the deep sea, fewer than a quarter have been formally identified. Roughly 90% of the species that have been collected on surveys of the CCZ have been previously unknown.

Parapagurus crabs with corals of the genus Epizoanthus on their backs. These “blanket-hermit crabs” were seen throughout Dive 16 of the 2021 North Atlantic Stepping Stones expedition. The more movable ferromanganese nodule field encountered during the dive also had a selection of smaller fauna, such as sponges, stalked and unstalked crinoids, isopods, stalked tunicates, worm tubes, and chitons. Source: NOAA

Many of these species exist in an environment once assumed to be incapable of supporting life. The hydrothermal vent ecosystems discovered in 1977 rewrote the basic biology of what life requires to survive, demonstrating that an ecosystem could run entirely on chemical energy rather than sunlight. Those same vent sites sit atop seafloor massive sulphides (SMS), deposits where heat from the vents precipitates copper, gold, zinc, nickel, and rare earths from the surrounding rock.

The extraction methods available to date have been ecologically destructive. Nodule collector vehicles are lowered from surface ships on umbilical cables stretching miles to the ocean floor, where they crawl under remote guidance, vacuuming up nodules and pumping them to the surface as slurry. The deep sea is a system built around near-total stability featuring creatures adapted over millions of years to an environment without disturbance, without light, almost without food. Mining vehicles run like underwater tanks across that floor, generating sediment plumes that travel hundreds of miles. When the plumes settle, they blanket the seafloor in fine particulate that smothers filter feeders and disrupts the “marine snow,” the slow drift of organic matter from ocean surface to floor, that sustains the deep-sea food system. Scars from a 1970 test off North Carolina are still visible today and nothing has grown back since.

Mining seafloor massive sulphides could look like demolition. Remote-operated machines would cut and grind the deposits, breaking apart chimney structures and surrounding rock, then pump the material to the surface as slurry through long riser pipes. In doing so, the miners would remove the physical foundation of vent ecosystems, which depend on those structures and exist nowhere else. The cutting process also stirs up sediment plumes, while discharge from the surface can spread finer particles through the water column over uncertain distances. Disturbing the deposits can also release dissolved metals into the surrounding water, changing local chemistry and ecosystems.

Any environmental concerns must be balanced with the reality that open-pit mining is more ecologically destructive than the methods proposed by Western deep-sea mining companies; open-pit mining projects are mostly located in countries with weaker environmental protections than the international frameworks governing the CCZ. The relevant comparison here is between different forms of mineral extraction, because extraction is necessary and environmental costs can only be mitigated, not eliminated.

Deep-sea mining’s economics have been equally challenging. A commercial-scale operation demands billions in upfront capital: specialized ships, seafloor robots, vertical lift systems, and processing infrastructure that, in most cases, has yet to be proven at scale. Even under favorable assumptions, production costs for nickel equivalents run $2,500 to $3,000 per tonne, which only works if commodity prices hold. They rarely do. Cobalt and manganese markets are thin enough that a single large operation could flood supply and crater the price of the metal it’s selling. The deposit rich enough to justify the capital cost may be too rich to sell into.

But the projected demand for critical minerals is so great, and the competition for control of these resources so intense, that the constraints which once foreclosed deep-sea mining may no longer hold.

“Copper is the new oil,” according to Robert Friedland, a legendary mining industry figure and one of the first investors in Apple. Copper is why the economics of deep-sea mining are becoming newly compelling.

As battery chemistry evolves, dependence on metals like cobalt and, to a lesser extent, manganese is already being engineered out. Copper, on the other hand, has no real substitute. It’s also embedded in virtually every system that carries an electrical current. Aluminum can displace it in some applications, particularly long-distance transmission, but its lower conductivity and higher losses limit its use in motors, electronics, and dense grid infrastructure. Copper remains what industry analysts call the “metal of electrification.”

By 2035, copper demand could exceed supply by 6 to 10 million tonnes a year, widening toward the mid-tens of millions by 2040 as demand approaches 50 million tonnes. Closing that gap through conventional mining alone would require 80 new mines and $500 billion or more in capital, with each mine taking 15 to 20 years to develop — a timeline that makes the math functionally impossible.

At sustained deficit levels, some commodity analysts project copper prices rising toward $15,000 per metric ton, against a historical range of $6,000 to $9,000. At those prices, sources of supply that were previously uneconomic begin to look different. Geological estimates suggest seafloor massive sulphide systems could contain over a billion tons of copper in-place, theoretically exceeding all known land-based reserves, with potential annual output from developed deposits exceeding 10 million tons.

The copper deficit is ultimately a problem of construction — enough mines, enough capital, enough time. Heavy rare earths are a problem of a different kind, one that no amount of new mining outside Chinese territory has yet solved. Dysprosium and terbium, the elements used in high-performance magnets for motors, wind turbines, and defense systems, have no substitutes and almost no supply chain outside Chinese control. The deposits exist elsewhere, mainly in Myanmar, but Myanmar’s output flows almost entirely into Chinese refineries. China controls roughly 91% of global separation and refining capacity for magnet rare earths, meaning material mined outside China typically must be sent there for processing.

Extracting even a modest volume of heavy rare earths from the seabed becomes compelling when the alternative is the shutdown of your country’s advanced manufacturing. Japan has been pushing this for years. Government-backed researchers identified large deposits near Minamitorishima, Japan’s easternmost territory, and a 2024 pilot-scale extraction program has since transitioned into a multi-year feasibility study, proving that heavy rare earth enrichment in deep-sea mud is technically, if not yet commercially, viable. The deposits are diffuse and require processing large volumes of sediment, but they offer a path to supply beyond China’s control.

China has leveraged REE as part of disputes with both the US and Japan and will do so again. As former Chinese supreme leader Deng Xiaoping put it: “The Middle East has oil; China has rare earths.”

Until recently, China’s strategy was to secure control over critical minerals through land-based investments, particularly in Africa. Since 2000, it has extended roughly $180 billion in loans across the continent, much of it tied to infrastructure and resource projects. It has since been forced to absorb tens of billions of dollars in losses, restructurings, and write-downs as projects underperformed or governments pushed to renegotiate terms.

In 2021, 3.5 million documents from the private Congolese bank BGFI were leaked to the French outlet Mediapart in an investigation known as the “Congo Hold-Up.” The documents showed that at least $138 million in public funds drawn from the central bank, the state mining company, and other institutions were funneled through networks linked to former DRC President Joseph Kabila and his family during the period when major resource deals were negotiated. The leak triggered a political and legal backlash in the DRC, with authorities reopening contracts and courts moving to assert greater state control over key assets — including one of the world’s largest cobalt deposits —while allegations involving the Chinese operators were investigated.

There is no hard evidence of CIA or other Western intelligence service involvement in the leak, but the access to this level of secret information and the fact that the leak was through the Western press is a strong indicator that intelligence services were involved. A year after the leak was published, China reversed course on deep-sea mining, a position it had resisted for decades. Beijing’s calculation had changed because land-based mineral strategies in Africa were proving expensive, politically unstable, and increasingly exposed.

China now holds more deep-sea exploration licenses than any other country and has built a large fleet of research and survey vessels operating across the Pacific and Indian Oceans. In June 2024, the survey ship Xiang Yang Hong 3 deployed the Kaituo II prototype mining vehicle for polymetallic crust and nodule tests at depths exceeding 4,000 meters in the western Pacific, in an area southeast of Taiwan where Chinese and Philippine continental-shelf claims overlap. Beijing Pioneer Hi-Tech Development Corporation, a state-owned enterprise under China’s Ministry of Natural Resources, has also announced a 2025 test collection in a western Pacific license area adjacent to Japan’s continental shelf near Minamitorishima.

China’s deep-sea mining push mirrors its approach to fishing, where it has been vacuuming up much of the world’s supply. The country operates the world’s largest distant-water fishing fleet, with estimates ranging into the thousands of vessels; between 2022 and 2024, Chinese vessels accounted for about 44% of global fishing activity and operated in over 90 countries’ waters, a scale that has fueled repeated accusations of illegal, unreported, and unregulated fishing, as well as environmental damage and labor abuses.

The appeal of deep-sea mining to China, which prizes self-sufficiency above all else, is apparent. The ocean asks nothing of you. Unlike developing nations, the ocean won’t attempt to nationalize your assets or default on a loan or hold elections with unpredictable consequences for your investments.

In the March 2026 issue of Qiushi, the Chinese Communist Party’s top theoretical journal, an editorial declared “The 21st century is the century of the ocean; whoever wins the ocean wins the future. China is one of the earliest nations in the world to develop and utilize the ocean ... We must deeply implement Xi Jinping’s vision to build a maritime power.”

Minerals are not the only thing China’s survey ships are looking for. In March, 2026, Admiral Mike Brookes of the US Office of Naval Intelligence told a congressional commission that Chinese research vessel activity provides data that “enables submarine navigation, concealment, and positioning of seabed sensors or weapons.” That same month, a CNN investigation found that eight Chinese state-owned vessels tied to deep-sea mining research spent relatively little time in China’s licensed mining blocks and considerably more time in the strategically sensitive waters around Taiwan, Guam, and elsewhere.

If the Chinese pursue deep-sea mining as part of an integrated strategy to control the world’s oceans, the US must treat it as a mandatory theatre of competition.

On the West Coast of North America, startups are attacking the dual problems of deep-sea mining’s economic and environmental challenges.

The Metals Company’s PATANIA III collector vehicle uses hydraulic suction to skim nodules from the seafloor rather than the bulldozing motion of earlier prototypes, reducing sediment disturbance by roughly 90%. TMC has also committed to preservation reference zones — areas of comparable seafloor left unmined — to maintain an ecological baseline and allow scientists to measure whether recovery is possible on any human timescale.

The Bay Area startup Impossible Metals, which announced plans to raise $1 billion last year, is working from a different premise. Its EUREKA vehicle hovers above the seafloor, using computer vision to identify individual nodules before retrieving them with a precision arm, one at a time, avoiding nodules with visible organisms attached. In trials it has operated with near-zero sediment contact. The tradeoff is throughput — selective collection cannot match the volume economics of hydraulic systems, and the company is betting advances in autonomy will close that gap.

Thomas Peacock, an MIT mechanical engineering professor, recently testified to Congress that sediment plume risk was minor or “roughly the equivalent to the grains of sand in a fishbowl.”

Both companies still introduce light, noise, and electromagnetic disturbance into environments dark and quiet for billions of years. The constraint is epistemological: there is no way of knowing what damage you are causing to an ecosystem nobody fully understands.

China isn’t asking these questions, and that asymmetry has a cost. Western deep-sea mining operates under public and regulatory scrutiny that Chinese state-backed operations do not face. If early commercial projects generate footage of sediment plumes blanketing the seafloor, the political backlash could produce moratoria that freeze the Western program for years while China continues unimpeded. The environmental methods TMC and Impossible Metals are pursuing are a condition of Western participation in deep-sea mining. Steamrolling the seafloor is a strategy available to Beijing but it is not obviously available to Washington.

The Trump administration is pushing forward regardless. NOAA and the Bureau of Ocean Management are accelerating permitting. In late March, the US and Japan signed a memorandum of cooperation to jointly advance deep-sea mining.

What happens next may not be decided by Barron or any of the entrepreneurs who spent a decade trying to convince the world the ocean floor was worth developing. Rio Tinto, Glencore, and BHP — the companies with the capital and operational scale to run deep-sea mining at volume — have watched from a careful distance. If the executive orders succeed in validating the legal framework and de-risking the first commercial licenses, the likely outcome is that major industrial players absorb or displace the startups that proved the concept.

Who extracts these minerals, and under what legal framework, will determine more about the next century than most of the decisions being made in Washington right now. A CCZ developed under American legal frameworks produces a different world than one developed under Chinese state direction, with output flowing into Chinese refineries, Chinese battery factories, and Chinese defense supply chains.

The floor of the Pacific is the last great untapped resource extraction prize on Earth. Gerard Barron had to carry a rock to the Oval Office to make that case. China will be mining the sea floor next to Japan’s continental shelf later this year.

I sat down with David to discuss his early entrepreneurial career, the rapidly evolving defense technology sector, and Silicon Valley’s shifting investment landscape. What follows is a transcript of our conversation.

CB: The period in which you started your career — the late 1990s and early 2000s — must have been very exciting with the tailwinds of the dot-com boom. What was it like to be an internet entrepreneur at that time?

DU: I was very lucky that I started working early. I’ve had a 1099 or a W-2 every year since the eighth grade. I started working for people that were some of the early internet networking pioneers — people who popularized this idea of peering on the internet, which is where networks come together and exchange traffic. I learned a lot about routing, networking, Unix, Perl, programming, all these things that really gave me a strong foundation for how the internet worked. But also a philosophy of people having access to more information, people being able to build and create solutions to problems, people having an engineering mindset. I think the internet was the first time in my life where I felt like, if you had an idea, you could make it a reality. It’s similar now with AI, where, for people that are not technical, all they need now is an idea and the agency, and they can turn an idea into reality. The internet was the first time in my life I had that same feeling, that if you wanted to learn something, you could find information and learn it. If you wanted to put something out there for other people to learn about, you could publish it on the internet. It was a very exciting time for me, and it really paved the way for the rest of my career. I was very lucky to get exposure to that from 14 or 15 years old.

CB: Tell me about DNS. It’s Domain Name System, right? That’s the acronym? So what exactly is it, and how did you come to specialize in it?

DU: The DNS is the system that connects names to numbers on the internet, like a glorified phone book, but it’s the only truly distributed system we have on the internet. It’s very distributed. People can publish their own DNS records. Every country gets to sign their own country code, top-level domain. I loved it because it was very loosely coordinated. DNS only works through agreement. There’s no government that runs DNS. There’s no multinational government. There’s no UN or WTO for DNS. It really only works through coordinated agreement. I really fell in love with that aspect of it. It felt very apolitical, but it’s this fundamental infrastructure layer of the internet — that without DNS, there would be no way to find resources, to find things on the internet. Nobody’s going to navigate by IP address. It just also felt like such a fundamental technology — there had to be a lot of room for improvement and innovation.

Before I started OpenDNS, the most popular DNS server on the internet was something called BIND, which was notorious for also being one of the least secure pieces of software on the internet ever published and widely used. It constantly had security vulnerabilities, constantly had exploits. It was written at a time when people didn’t really care about security on the internet, and nobody actually wrote a better version of it, until I came around. Now there’re a lot of robust implementations of DNS, but I basically decided to create a DNS service that would be much faster, much more secure, much more reliable, much more robust, and really, designed for scaling to the modern internet. DNS is just this foundational component of technology that powers the internet.

There’s another one out there called BGP. I didn’t want to create a BGP company, but DNS is one that we all use. It’s a constant source of headaches on the internet, and I just felt like it could be much better. In 2004 I moved to the Bay Area, and in 2005 I started OpenDNS to build a better DNS. And by the time we sold it in 2015, the whole internet networking world had woken up to the idea that DNS could be much more secure. Fast forward another 10 years — there’s a ton of really robust and secure DNS implementations. But at the time, OpenDNS was really revolutionary.

CB: Could you walk me through the process of — over the course of 10 years — building a company that you end up exiting for over $600 million?

DU: People always think these things are overnight successes. My overnight success took 10 years, and people forget that it’s really a hard journey. The path to success is non-linear. We started out as a consumer company trying to sell a cybersecurity service to consumers, but it turns out that while we built a great product, consumers would rather go buy a $5 Starbucks than pay $5 a month for cybersecurity. They really just don’t put a value and premium on cybersecurity. They do once they’ve been breached or compromised — then they value it — but they don’t value it ahead of time. Five years into the business, we were really struggling. We had lots and lots of users. We had a free service that was very successful, but we really had trouble making money. So in 2010, five years after we started the company, we did a complete hard pivot on the business and really shifted to the enterprise. We kept the same product, kept the same service. In fact, we had realized at some point that a lot of enterprises were using our service but just not paying us, and so we just started offering more and more enterprise features: better reporting, better access and account control and account management, better security controls. Then the business really took off. But it really took us five years to pivot into that business. We never would have gotten there if we hadn’t started with the consumer business, because we leveraged all the data we learned and all the data we’d collected to do our security research. It really does take a long time, and that experience set me up probably better than anything to be an investor and even to understand the journey that entrepreneurs go through. You’re not going to have every quarter be better than the last one. You’re going to get punched in the face, you’re going to have all these things that happen to you — but if you believe in your mission and the product, and you have a good team, and ideally, you have investors willing to stay with you and support you, you can ultimately build a great company.

CB: How have the upheavals of the last couple of years impacted the cybersecurity sector?

DU: It’s a really exciting time to be refocusing on cybersecurity. I took a break from it on the investing side for my first five or six years at the firm. One of the issues with cybersecurity has always been its asymmetric nature, meaning that the attackers have unlimited tries. It costs them nothing to try to attack you. They only need to be successful once to have a breach. On the cyber defense side, you need to constantly protect against every way in. You need to constantly pay for the best talent, the best solutions, the best tools. Most organizations fall below that cybersecurity poverty line — they can’t afford the best people, and can’t afford the best tools, and have difficulty deploying and managing those things. AI finally gives the defenders a better fighting chance and really helps to level the playing field against that asymmetric dynamic. With AI agents and tools and continuous monitoring and continuous vulnerability assessment and all these capabilities, you can have much more robust defenses, much more robust network visibility and controls, see what’s actually happening, and better manage your access control on firewalls. AI allows you to get the benefits of the best intelligence when it comes to cybersecurity, as well as to have much more capability on the offensive side. It’s an exciting time, because companies are starting now that really give all organizations — not just the ones who can afford to spend an unlimited amount of money on cybersecurity — the ability to have a fighting chance against adversaries. That required a technological step-function change to happen, and AI has given it to us. For the last 10 years, cybersecurity on the defense side has been very challenging, and now I think it’s really a new era to really rebalance the scales.

CB: You also have these emerging quantum technologies. How do they interact with cybersecurity?

DU: For us, from an investment standpoint, quantum still presents as a research effort — but one worth pursuing. If it happens and if it works, a lot of what we take for granted around encryption and encryption capabilities may end up turning out to not be as secure as we thought. So it’s important research to be doing. We haven’t focused on it from the investment side, mostly because we think it’s still too early to be commercially viable. But once quantum unlocks that, it really changes the way people think about encryption, about encrypting data on the network, and about information security. In theory, if a working quantum computer really exists and has enough capability, it will be able to break a lot of the common encryption used today.

CB: What would need to happen, from an investor’s perspective, for quantum to become viable?

DU: I’m not an expert in that. It mostly still feels like a science project — It’s hard for us to follow. What I find is that even the most advanced technologies, when they’re real, they’re not that hard to explain. Take nuclear fission, for example. We know how nuclear fission works. We know how fission reactors work. We have a lot of them. They’ve existed for seventy years. It’s very complicated, but not that complicated, right?

CB: Where do you see the greatest upside in the defense sector?

DU: There’s no shortage of opportunities right now as we think about how to reinvigorate and rebuild the defense industrial base – as folks in the Department of War say, rebuild the arsenal of freedom. We are at this technology watershed moment where we now have the ability to do commodity manufacturing of low-cost and atrittable systems. Low-cost drones, low-cost missiles, and things like that. We have now seen people shift from building larger and larger systems to building smaller, more agile, and more expendable systems.

We also have this technology watershed moment with computer vision and autonomy and electronic warfare, where we can now send small systems out autonomously to acquire the target and try to intercept or attack it without any human intervention. Even without radio communication in an electronically contested environment or an RF-contested environment. Because we have those capabilities, we’re seeing rapid innovation — and because we’ve had the benefit of learning from recent conflicts. We see what that looks like in Ukraine. We see what that looks like in the Middle East and the Iran war. Things like Shahed drones were not part of the calculus of what people thought the fight of the future was going to look like. But they are clearly part of the calculus now, with entire regions GPS-denied, or denied other positioning systems. You need other ways of finding your position and figuring out where you want to go. Because of all these things, we’ve rapidly seen innovation in the software side of the defense industrial base — the small, credible systems side of the defense industrial base, and the autonomy side. All of that creates opportunities that are, I would say, upstream of those products. Upstream is one word for it.

Another way I usually talk about it is shifting left. What are all the things you need to make drones, to make autonomous systems? You need computers and guidance systems. You need low-cost motors — motors for drones, whether high-end motors like you might see on a Shahed or very low-end motors you might see on a quadcopter. How are you going to make those? Where are they going to come from? You’re going to need optical systems — where are you going to get the optics? If they’re made in China, and we have a conflict in the INDOPACCOM region, we’re probably not going to be getting those shipments. We’re probably not going to get the shipments of our computer chips that we need for the guidance systems. Where are we going to get rocket motors? Where are we going to make rocket fuel? And what about all the chemicals needed to make propellant? All these things just create more and more opportunities— left of the industrial base: the manufacturing base, all of the supply chain components that are needed.

I think we’ve ventured into a realm of unlimited and unbridled opportunity. We can’t replicate what they’ve done in Shenzhen, China. We can’t do this the same way we did it in World War II, by just repurposing an automotive factory line. We have to do it in a way that is representative of the time we’re in — leveraging factory automation, leveraging as much software as possible. It’ll rhyme with the past, but it won’t be exactly like the past. And so we’re working with the founders who really have a view and a vision for how to capture this opportunity and deliver what the aerospace and defense market really needs today.

CB: As someone who’s allocating capital to companies which ultimately aim to sell to the government, how does the defense procurement model work?

DU: We’re coming on the tail end of a procurement system that has become incredibly calcified and is finally being decalcified — for lack of a better term — calcified through bureaucratic rule-making and all kinds of things designed to prevent mistakes. That means that we have a process that over-specifies requirements. The government should say, “we want a missile that can go 1,000 kilometers, carry this kind of payload, and fit into this kind of a launch tube.” That’s what they should describe. Instead, they get very, very specific on exactly what kind of propellant it needs. Is it a liquid propellant or solid propellant? They might even call it by a certain name — you have to be able to make the Tomahawk. This contract is for a Tomahawk. Well, what if somebody makes something that works like a Tomahawk but isn’t one? That should be allowed to be bought too.

The over-specification of requirements has really bogged down the R&D and innovation side. You have companies today, like Anduril, that are building before the requirements are fully set — and as a result, they get to help shape them. We have a process that over-optimizes for compliance, making sure that nobody can protest an award or a contract, that nobody can say any rules got broken. A lot of the existing rules really favor the incumbents. They take a long time. I read recently that there’s something like $2 trillion worth of awarded contracts that won’t be delivered to the Pentagon for over a decade. That’s just too slow and too long. Those processes favor the incumbents, who are equipped to deal with such a slow process, when our warfighters need capabilities quickly. They need iteration.

We’ve entered an era where everything from the procurement process to the appropriations process is finally being changed to enable a much more nimble defense industrial base that produces things quickly. One of the big success stories of Ukraine is that they’ll work on a drone capability, put it in the field the next day, get feedback, iterate, and a week later have another version of the drone out in the field. That is so far removed from our process historically here in the US. To have the best capabilities and the best technologies, we need procurement processes — and the ability to field new equipment — that focus on iteration and feedback, not on calcified bureaucracy. We’re putting compliance way ahead of capability and speed.

CB: Much of the procurement in Ukraine is coming from the battalion level as well, which is so much closer to the source.

DU: They have a sense of urgency we don’t have. In a crisis, a lot of rules get thrown aside. But we are in a very slow-moving crisis right now in the US. We have expended a massive amount of our ordnance and missile capabilities in the war in Iran, and our resupply time is on the order of years, not months. We cannot operate in that kind of environment.

CB: Erebor has recently received its banking charter. What kind of impact will this new bank have on the hard tech industry?

DU: I think with the loss of Silicon Valley Bank, it’s become very clear that startups — particularly startups in the manufacturing, defense, energy, space, and supply chain categories — need a banking partner that understands their business. One that understands that they’re not going to have five years of financial records that they can share, because they’ve maybe only existed for a few months. When you’re dealing with everything from inventory to tooling to all the working capital needs that you have, you need a banking partner that understands that dynamic and is willing to work with you. And traditional banks, I would say, really do not. Historically, Silicon Valley Bank filled that void. But as Silicon Valley Bank went away, it exposed a huge gap in the market. The idea that startups need a partner is not novel. We had that, and we lost it. A dedicated bank like Erebor for the hard tech community, the national security community — a bank that, frankly, is very focused on supporting US companies, one that will not just seize your assets if the EU sends you a demand letter, the way a very large, established bank might — an American bank focused on the innovation economy is just a critical need, particularly in the hard tech, defense, and national security space.

CB: How long does it take to actually get the factories up and running?

DU: It takes a long time, which is why we need to act like we’re in a crisis right now. We are, but we just don’t feel it. Take the government shutdown. We’ve watered down what a crisis feels like. We actually need Congress to act with urgency. They have the power of the purse, and we need to behave like there’s a real crisis. Because it can take years. We have a company, Radiant Nuclear, that’s building a factory to build nuclear reactors, and they have so many orders that they really could put together the plans for a second factory. There’s so much demand out there, but that means they need to have the fuel supply chain built. They need a regulatory environment that’s clear. They need to have multi-year commitments from the DOE and the NRC to know they are actually going to be able to do these things, and that requires real government support and partnership.

For far too long, the government regulators have become these “no” people, when really they should be partners and enablers. In many countries, the regulator is an enabler. They support businesses. They help you build your business, get the licenses you need, operate in compliance with the law. Regulators in the US focus way too much on bureaucracy, too much paperwork, too much CYA behavior. Thankfully that is starting to change, and that is essential, because it does take so long to stand up these factories. I’ll give you one example. You might have a 10-year contract to field missiles for the United States military, but if the dollars that fund that only get appropriated to you annually, when Congress passes a budget, why would you ever buy more than one year of inventory at a time? You have no guarantee you’re actually going to get your order the next year or get paid the year after that? Why would you buy the inventory? That’s one of the reasons why our supply chain is so slow. Once the government says, “Okay, we’re ready now we want to buy it,” it then takes time to order the inventory, get the inventory, spin up manufacturing, rehire all the people in your factory. We need a Congress that understands those problems and is focused on multi-year commitments to fund programs — on multi-year commitments to award dollars that motivate the private sector to build these factories and the supply chain resiliency we need in this country.

CB: From a funding perspective, could you walk me through how this sector is different from the software that you’ve traditionally invested in?

DU: If you’re going to be an investor in the American Dynamism categories, you have to be prepared for a much, much slower J curve to revenue growth — meaning these companies are going to focus on R&D for a lot longer. They’re going to have to deal with more dynamics than a traditional sort of enterprise software company might. They have to deal with regulators. They might have to deal with hardware. Most of our companies sit at the intersection of atoms and bits, where hardware meets really advanced software, which means you’re going to have inventory needs, machining needs, factory production line needs. The capital needs are much more intensive. The revenue might get pushed out. But then once these things start working, the revenue catches up very, very quickly and grows much faster.

When investing in these categories, you have to make sure you understand the business. Oftentimes, we have to institute financial discipline much earlier in these companies, to really think about what fundraising looks like, how you make sure that these companies don’t run out of money. That’s one of the biggest things that’s changed in the last few years. We won’t take all the credit for it, but with American Dynamism, we’ll take a lot of the credit: we’ve unlocked a lot more downstream capital. There’s a lot more equipment financing. There are more investors excited about these areas, which makes it a much safer area to invest in. It means there’s going to be more capital after the seed round; after the Series A, there’ll be more capital available to support these companies and help them get through that part of the J curve — until they can really get into the market and get real revenue. If the government’s their customer, getting real programs of record; if other folks are their customers, getting real contracts and delivering real solutions.

CB: Agentic AI has seen widespread adoption in the last couple of months. What’s the tangible impact of that development on the software industry from an investor’s perspective? It looks like there’s also been a considerable downstream impact on Private Equity.

DU: It’s a profound change. As I mentioned earlier, as related to cybersecurity — or even my experience on the internet, AI and large language models, especially as they relate to coding, are now unlocking the ability for people with ideas and agency to manifest those ideas into reality in a way that was very, very difficult before. I think it’s actually very exciting for the hardware companies in the venture ecosystem, because it means they can now very rapidly take these hardware prototypes and introduce much more advanced software, much faster, without having to spend months and months tooling up an entire software stack that can be automated and generated very quickly. That means people can go work on other problems. They can hire people to work on the much more difficult factory problems, the hardware problems. And software is just going to get much, much better. I think it’s very exciting.

In the private equity world — and here we’re primarily talking about the enterprise SaaS world — I think we’re going to see that things that were previously valued at one price are now revalued at a new price. It doesn’t mean a lot of these enterprise SaaS companies are going away. I don’t think they’re going to go away, but it certainly means their growth may not be what we thought, and it may mean some of these products will go away. But lots of other things will stick around. Take a company like Samsara, which gets punished in the public markets. I don’t think Samsara is going away. They sell enterprise software into large fleets. They have an IoT solution and hardware component with it. There are lots of companies that have been unfairly punished, and maybe in the private equity world, the buyout world, they overpaid for some things they’re going to find out they overpaid for. But that’s life in the big leagues. They’re going to have to deal with that. Thankfully, that’s not my problem.

This interview has been lightly edited for length and clarity.

Dr. Edward Luttwak is an American author and strategist born in Romania and raised primarily in Italy. The author of numerous books on geopolitics, economics, military theory, and history, Luttwak earned a degree from the London School of Economics and a PhD from Johns Hopkins University. His first book, Coup D’etat, published when he was 26 while he was working as a consultant in London, was an international bestseller and inspired the subsequent 1972 coup attempt in Morocco. His other works include Turbo-Capitalism, The Grand Strategy of the Roman Empire, and Strategy: The Logic of War and Peace. Luttwak has worked as a consultant for the White House, the National Security Council, and various branches of the US military. He has also advised multinational energy firms, among other clients. He has served in both the British and Israeli armed forces.

I sat down with Luttwak to discuss the post-heroic era of history, innovation and military procurement, and the decline of democratic politics. What follows is a transcript of our conversation.

CB: With so much happening in the world, where should we begin?

EL: I suggest one thing that is on my mind. A phenomenon that exists all over Europe in a very powerful way, and beyond Europe, in many other countries — one that nobody talks about. The post-heroic era. It is a phenomenon that hits you in the face the moment you think about it. In certain parts of the world, such as Europe, the whole territory of which was always fragmented into different states. In other parts of the world, such a large territory became one empire — like China or even India, for the Mughal Empire, and in the Middle East the Ottoman Empire. Most of these territories historically were not under the control of small states. Only Europe was divided into small, quarrelsome states. The only big state was Russia, and Russia was not present in most episodes of European history until very late, under Peter the Great.

These Europeans became small, quarrelsome states that had wars every 20 years, and those wars resulted in people dying. When the war ended, the warriors returned, and the warriors wanted women. The women wanted warriors. They didn’t want to marry these stay-at-home guys. And the more dynamic people who came back from the war — they picked up a few things, and then they would make more children than the number that died in the previous war. So the demographic machinery of Europe was: fight the war, have people killed, and many more babies are born. Second, much of Europe, Northern Europe, was built entirely of wood. Every time there was a war, entire cities would disappear in fire and flame, and then they rebuilt twice what had been there before. So the whole engine of European civilization was war — and that’s why there was a timing to it. Every 20 years it was time for a war.

There were exceptions — for example, Napoleon. Napoleon was so excessive in fighting all these wars, and when he finished with the wars, he came back once again, with Waterloo, once again when he was already defeated. Because of that extreme amount of warfare, there was no real war until 1870 — and even that was isolated — Franco-Prussian. But otherwise, then you went back in the 20th century to the normal rhythm, a war every 20 years. At the end of every war, there was a huge upsurge of the economy, and then a second one from the children growing up. A demographic push — more people, more people. The last time it happened was 1945, and people did not realize this had happened all through history. So they invented special words, like the German Wirtschaftswunder, economic miracle.

The Italians had their own. The French said les trente glorieuses, the glorious 30 years after ‘45, when the economy was growing and growing — and then, of course, it stops. It stops after 1975. 30 years later, European economies started not growing. What’s happened is this. In the 19th century, when you had proper military service — organized and everything — a 15-year-old boy started thinking of the day when he would go into the army, when he would get his rifle. And once he gets his rifle and his training, he thinks of the next day: that is when you go on campaign. That word is from the same Latin root as Champagne. You leave the barracks and you start marching through the open country, and it has to begin in springtime, because there are many horses, and they have to have grass. So as you’re marching, eventually you’re marching with golden wheat growing next to you. That is how Europe lived until 1945. That war was a huge war, did a lot of damage — and somebody invented nuclear weapons. That interrupted the machinery. There was enough energy left over in 1945 – the Wirtschaftswunder miracle, and there were babies. The birth rate exploded. All these people came back.

The Germans lost so many soldiers killed, but there was still a birth rate increase. The French had prisoners during the war. They came back. They made children. Italians, everywhere — there was a big bump in births. That meant that 20 years later — now we’re in 1965, you had lots of young people entering the labor force, and they all got jobs. Okay? Now, what happened is that suddenly the machinery was taken out, the engine was taken out of Europe. What is the last fertility number for American females?

CB: It was below two, I believe.

EL: Once it’s below two, it means your population is shrinking and getting older. I’ve been a military contractor all my life — a military advisor, military planner, and a combatant, fighter. I was in different wars and so on, different wars for governments. I was in the British army first, and then the Israeli army, but also as a contractor. What’s happened — this is my hypothesis — is that the reason so many very different kinds of people, very different cultures, have lost any desire to fight in any war for any reason is fertility. It is as if the wars of European history were all fought by the spare male child that a woman would have. Normally four children. Normally there would be two boys. If one of those boys dies, the family continues. That is my personal theory, which I advance to explain the fact that suddenly, everywhere in the world, they don’t want to fight. Now, meanwhile, we have military institutions that assume the opposite. In fact, NATO is huge. The sum total of all NATO personnel is 3.5 million. They boast another 3 to 4 million reservists, which means they have many more people than Russia, many more people than China. The whole PLA is 2 million.

CB: The rate of innovation seems to have declined. When you look at the period prior to World War II, for example — the North American Aviation Corporation could put together a prototype of the P-51 in 100 days. It took Lockheed 48 hours to come up with a new bomber design that the British RAF had requested, the blueprints drawn from nothing.

EL: This still happens in Israel.

CB: What happened to innovation, and how do you bring it back? How do you accelerate innovation?

EL: What’s your surname?

CB: Becker.

EL: German. So, tovarish, Becker, everything has to do with the structures of capitalism. You’re mentioning a company called Lockheed. Who do you think started Lockheed?

CB: The Lockheed brothers.

EL: Exactly. And Northrop was started by Jack Northrop. So aviation companies were set up by people who loved airplanes, whose only modus operandi was to quickly build a prototype, fly it, and try to fiddle with it. Over time, the aircraft pioneers were replaced — naturally, because their work became Northrop Corporation, Lockheed Corporation, and so on. But something else happened. The engineers were demoted more and more and more, because the companies were run by financial analysts, by lawyers, by political lobbyists. The engineers counted for less and less. The maximum of this phenomenon was Boeing, and they got to the point where they couldn’t build airplanes. Boeing went into a complete dive, and only then did they bring back somebody who understood the company. They were not delivering aircraft — and I don’t mean aircraft for some poor people in Afghanistan. Not to the President. The presidential airplane was not delivered.

The Israelis ordered tankers from Boeing in 1981. The first one was delivered the next week. In other words: first they got rid of the engineers. Then everything became the logic of a contract. Now, what’s the virtue of a contract? The virtue of a contract is that it should last forever — 20 years, 25 years. That’s why there’s enthusiasm right now in Europe — because they’re getting extra money, 5% of GDP, for defense. And immediately, what do the aircraft people do? They do multinational programs.

For example, there’s this project — the British and Italians and the Japanese are supposed to develop a fighter aircraft. This enables the project to last 25 years. I mean, the F-35 is a good airplane. Israelis use it and everything else. But it took them 20 years to develop. Now, once you are developing an aircraft over more than four years, five years, all the microprocessors change. So every computer element has to be changed. You have to stop, go back to the beginning, take out all the subsystems — this box, that box, and so on. It’s a sociological change. Why did that happen?

Selling airplanes used to work like this. You make a prototype, you and your brother. Then you call an Air Force General, and you say, “Joe, I have a plane.” And the guy says, “If I fly it, will it crash immediately, or only later?” And then the general would come and fly the airplane. And then he goes back and tells the other generals, “this flies really good.” And then they say, “Okay, we’ll give you a contract for a few prototypes.” They would work day and night and they would make it. Now it is an elaborate corporate process, and it is so willed, so desired, so wanted by the United States Congress, which passes the laws under which they write regulations. They only produce a few hundred laws, only a few hundred, but the regulations are many thousands.

CB: It reminds me a bit of the Ukrainian procurement model, which occurs at the battalion level, as opposed to the ministerial level.

EL: This is the Israeli army. Everything the Israeli Army Corps acquires is acquired by the officers of the army corps — that includes the 65-metric-ton infantry combat vehicle, the heaviest vehicle in the world. You know, 65 tons, and it has no turret. No turret because it’s an infantry combat vehicle. So it’s actually more armor, and I rode it through the Gaza Strip during the fighting. And you see everything from screens inside the vehicle. You don’t have slits and any of that. The secret is the people who use the weapons, the people who fly the airplane, are the ones buying the airplane — not officials, lawyers, lobbyists, acquisition specialists and purchase specialists and selling specialists and marketing people. If you don’t remove all these barriers between the designer and the user, you’re going to end up with the end of aviation — the F-35 is very close to it. It took 20 years to develop a very normal airplane with stealth. And the value of stealth is so little these days that the last purchase was the F-35 — where you sacrifice speed, range for stealth — and they also bought the F-15, which has no stealth at all. The US Air Force order last time was an equal number of stealth and non-stealth. Why? Because stealth is not important. Have you ever been to war?

CB: No.

EL: I’ve been in war, and the most important thing in war is that nobody should see you. And of course, you don’t go there and start to advertise yourself by sound. What radars do is advertise. Some years ago, in 1972, the Israelis started working on an anti-radar drone, the cost of which was maybe one or two percent of the cost of the cheapest radar in the world. And that drone just flies around. When the radar touches it, it goes to the source of the radar.

CB: Because of Yom Kippur, right? The Arab armies were equipped with Soviet SAMs.

EL: In the 1973 war, the Yom Kippur war, the Russian Air Defense Organization — PVO strany, it was called — was composed of very serious people who responded to American air superiority by making it a whole separate service. And they developed everything. By 1973 they had a missile, the SAM-6, that was really a good missile. They shot down many Israeli aircraft, so the Israelis decided they had to have a countermeasure for every anti-aircraft missile, not just SAM-6. The SAM-5, for example, has a range of 250 kilometers. You can be in Damascus and shoot down an airplane over Tel Aviv. And that was my Mongolia trip. That’s how I got to see Mongolia. SAM-5. So anyway…

Now, the contrary story, which will be very short: I’ve decided that the Marines cannot reach the Chinese island bases along the coast — they have no way to reach them. If the Marines could reach these bases, a company of Marines — force recon or just regular Marines — would come there and kill all these Chinese soldiers, take over the island, because they’d come by surprise. The Chinese have been on watch every year, every day, every night — but now they’d come by surprise. They would definitely take them. But how the hell do they get there? How will the Marines get there? The Osprey aircraft they developed — the multi-plane aircraft — has a huge radar cross section. It would be seen, even shot down — no chance of making it.

However, when I told that to the then-Commandant of the Marine Corps, General Berger, he said to me, “Edward, we don’t care whether it can be shot down or not shot down, because we can’t get them out of the garage. They’re so complicated that they don’t fight, and we have to work a week to get them out” — in effect. The vulnerability of the Osprey doesn’t interest me, because they can’t fly. So I said, “So what are you going to do?” He said we commissioned very small landing vessels, which the Naval Sea Systems Command is supposed to deliver for us, to carry an infantry company — 100 men — a small boat, fast, open boat for landing. And I said, “What happened?” He said, “They’re not doing it. They instead offered us the medium landing boat. The small boat they promised in three years, the medium takes seven years, and the medium is bigger, more easily seen.” And I said, “You give me three Marines and we will go to the headquarters and force them to give you the boat you actually need.” He said, “Oh, if I do that, they won’t give me anything” — in other words.

I had a brilliant idea: a submarine, purely electrically powered. Every submarine configuration, if you look at it, has a big engine — and there’s a space below the engine, I don’t know what’s used for. And then there’s a big engine that occupies the main part of the hull. Then there’s a forward part, the sonar, and so on. I would have no engine at all, because all these engines end in an electrical motor. They always generate electricity. That’s how it’s done, because if you use a mechanical transmission, it’s too noisy, right? So it’s an electrical motor. I say: keep the electrical motor, take everything else out, and I feed the electrical motor with lithium batteries. After the landing, you pop up a cable from the submarine, and the ship comes and recharges it. So we have a hull. We have no engine in it, only the original electrical motor at the very end. It occupies a very small space. 100 Marines fit very nicely there. Below them are the batteries. And then I evolved the idea and put the batteries between the pressure hull and the outer hull. Cylindrical lithium batteries.

I had been unable to present it to the Marine Corps. The Marine Corps refused to receive the design. Everybody in the Marine Corps knows me — they’d say, “We have to come through proper channels.” And what is the proper channel? It’s that the Lockheed guy, who is ex-Marine Corps and left last year, calls the current guy and makes the appointment. They refused to see the project. By now Commandant Berger, who started everything, has moved on. When I called Berger to arrange an appointment for me — he’s the former commandant now — I just wanted to present my idea. I paid my own money to have a fully engineered design. It’s simple engineering. Berger says, “yes, if I call, you will definitely get an appointment, and you’ll get it tomorrow morning. And the person you get an appointment with will then not do anything. And if he tries to do anything, it will fail, because procurement channels are separate.” That’s the Commandant of the Marine Corps.

That is how you get no innovation. If you want to strangle innovation, adopt the current American system. In fact, the United States government should give free briefings to explain the American procurement system to the world — because if they copy it, it will be so much easier.

CB: What should be done about the F-47, the Next Generation Air Dominance program, and the B-21 Raider, our next generation stealth bomber?

EL: I think they’re wonderful aircraft and very beautiful, and they should make them in plastic and sell them in toy shops. But they could also make some in metal, for hobbyists or people who want something more serious. As for combat aircraft — forget it, because the sheer complexity of them is sufficient to make production a slow process. During that slow process, everything electronic will change. So either they ignore that — imagine that you get an aircraft today with the electronics of 1970 or something like that — and things are moving really fast, so it’s absolutely not possible to do it.

You can’t start with the paper project. You must start with a prototype. The airplane is what made America a great power. I once had a very well-paid contract to study intelligence before World War II, Japanese, German, Italian, French, British. The Italians won, totally. Not slightly better — totally better. Now, who were the Italian intelligence officers? There were two infantry colonels who had no particular education in intelligence or background in intelligence. They got the job because the Italian Army, hierarchically, was superior to the new Air Force and the Navy — so they got a job, and the Italian Army had no particular interest in intelligence at all. For them, it was not a big issue. They’re defending the Alps, or they’re in Ethiopia. Either way, it’s a matter of practicality — of knowing the countryside. To fight in the Alps you need an excellent map.

So these two guys are asked the following question by Mussolini in January 1940: who will win the war? Because Mussolini wanted to be in the war; he wanted to know who would win the war. And the answer was: the United States will win the war. At that time, the isolationists were in control, so the army was very small, there was no money. That’s how they wrote the analysis. Americans live on two different coasts of a broad continent; therefore, to do anything, they have to fly. In Europe and other places, very few people fly. In America, the mail goes by air. Everybody has to fly. Therefore, the Americans have something like 30 aircraft companies, and they’re all different, and so on. And therefore, if the Americans get involved in the war, they will produce huge bombers and will flatten German cities. It doesn’t matter how clever German generals are at maneuvering and outmaneuvering because every German city will burn. They actually said “flat” — it was the British who invented the incendiaries. Many German cities were made of wood. And therefore: Mussolini, don’t enter the war — because even if the Germans are winning the first year, the second year, the third year, the end of the war is the destruction of every German city. Now, these infantry officers were not equipped with sophisticated intelligence methods to then say, “Oh, well, of course, there’d be a surprise, there’d be a political shock,”— to make a scenario to explain how the country that lives on two sides of a faraway continent will become involved in this European war.

They also said the Americans have aluminum. They have plenty of electrical power to make aluminum, and they have many interesting designers. There are so many different aircraft, so it took a whole layer of mistakes to lose all that, because the F-35 is a very poor aircraft. It’s not supersonic, it’s not fast, it’s not big, doesn’t have the range or the capacity, and it has everything for stealth — and stealth doesn’t matter, because you can knock out every damn radar the first day with cheap drones. The Israeli anti-radar drone is as old as your grandmother. The Harpy. It was one of the first things they did. It’s like 30 years old. It flies around, and radar paints it — as it’s called, “painting” — and it destroys everything.

CB: Do you still own the cattle ranch in Bolivia?

EL: I sold it last year because neither my son nor my daughter wanted to take it on. And there has to be an owner. In Bolivia, the ownership is the owner, not a piece of paper. You don’t go to court in the Amazon. But I love doing it. I love cows. I now understand Hinduism; sitting there in the evening, slowly the light is fading, and suddenly — hundreds and hundreds of cows walking, following a leader, who is another cow, down to the lake to drink water. Complete silence, hundreds of cows moving. You don’t hear anything, because of the grass…

CB: We have a $200 trillion, approximately, global debt burden. And I wonder what might happen if we were to have a margin call of some kind.

EL: The American debt is American politicians giving more gifts to their supporters than the taxpayer will pay for — and instead, you print the bond, and you sell it to foreigners. The American debt was not due to a war, not due to a famine, not due to a disaster. No, it was to give gifts to people. The man who first explained democracy, contemporary democracy, parliamentary democracy, was an Englishman called Walter Bagehot. He wrote The English Constitution in the 1870s. Universal voting was introduced in England with the 1870 reform, and Walter Bagehot wrote: every common man can now vote and decide who is the prime minister. That’s fine. That’s very good, but one day, they will discover that they can use the vote to compel the government to give them money, which the government will do by printing money. When that happens, it will lead to the end of democracy. It’s only a question of time. Bagehot is one of the great political thinkers. He predicted modern America. We did not have a disaster, a catastrophe, a monsoon. It was simply politicians giving more money to the electorate than they collected in taxes. That’s their idea of being generous.

CB: When do we have our Caesar moment in America?

EL: When the democratic system falls, there is Caesar. Well, it’s not Donald Trump. You know why I know it’s not Donald Trump? Because when he wants to say that many people were killed, he uses the word “decimate” — and Caesar would know that to decimate means killing one out of 10. There’s a female spokesman at DoW who will say, “we decimated.” This is the thing that annoys me, personally, much more than anything else: using the term decimate to mean killing many when it actually means killing few. You ask when is our Caesar moment. When does democracy break down and Caesar arrive? And the answer is: there are competing Caesars. We now have political figures who claim the status. But Caesar means that the military are plausible. I’m going to present you with a little issue, young man: in a post-heroic era, there is no Caesar.

Who are the real, proper heroes of our time? They are exactly people like Musk and Thiel and so on — the mega innovators. Obviously, it’s been years and years and years that they did anything to be rich. Being rich is not important for them. To make decisions that cost them hundreds of millions of dollars is no problem. They don’t respect money. Think of Julius Caesar — if the Roman legions had all become pacifists, which is the case with most people in uniform worldwide. Military service is not connected with war anymore, even in the United States. To have Caesar, you need a victorious war. And they don’t want to fight, and they definitely don’t want victory. Victory is a dirty word. I wrote a book — one of my books that you have overlooked in your carelessness — is called The Meaning of Victory. And I wrote it years ago to reintroduce the concept of victory, because America was fighting its wars merely to reach a settlement.

This interview has been lightly edited for length and clarity.

Project Omega is working to commercialize betavoltaic technology. By extracting valuable materials from nuclear waste and turning them into power sources, the company aims to address longstanding challenges in nuclear energy while advancing new applications in defense, space, and computing. I spoke with Sheehan about his entrepreneurial background, the economics of nuclear waste, and the future of nuclear-powered energy systems. What follows is a transcript of our conversation.

CB: I understand you started several companies. Walk me through your background and first few entrepreneurial ventures.

SS: Out of high school, I took the civil service exam, trained as a medic, and spent time as an EMT and firefighter.I didn’t initially plan on any entrepreneurial pursuit, but I was doing contract work on the side that turned into a software company. This was back in 2006, 2007, and that was what taught me how to start a business, just trial by fire. I was definitely too early with this thesis, but I realized that the software side of things didn’t have the same impact as deploying hardware. I really wanted to dedicate myself to deploying hardware that could transform the energy industry, transform manufacturing, and transform the physical world. After that business and college, I studied chemistry and physics at Yale University and got a PhD.

My second company, Catalytic Innovations, was spun out of Yale. Initially, that company focused on systems that produce hydrogen on site, but hydrogen wasn’t a huge market at that time. We pivoted into metals, and I learned a lot about the industry, zinc and aluminum smelting. We were able to sell that business, and then I started Air Company.

Air Company made synthetic jet fuel, but we started out making vodka because we needed to make our unit economics work, and fuel is $4 a gallon, but vodka is $400 a gallon for a premium product. The idea was to use that as a stepping stone. After I left that company, I started working with a handful of people who were focused on electrons for data centers, and on how we generate enough electricity for our increasing demand. With Project Omega, we aim to reduce our dependence on Russian uranium, or any sort of foreign nuclear fuel source, and solve the nuclear waste problem that we’ve had in this country for decades. The government told all the operators that they would get rid of nuclear waste by 1998, and it’s still sitting there.

CB: Could you break down the nuclear waste problem? What is it and why is it an issue? Why is it difficult to handle?

SS: Nuclear waste is a mixture of many different materials, but it’s primarily reusable uranium. 96% of it is just reusable uranium, and it still contains over 90% of the energy that it had when it went into the reactor. The colloquial term, nuclear waste, is called in the industry “used nuclear fuel.” Like a secondhand car is slightly used, pre-owned nuclear fuel is mostly reusable uranium.

What happens when you run a nuclear reactor? You split apart uranium atoms. The resulting neutrons hit other uranium atoms, and you get a lot of energy that comes out from breaking apart atomic nuclei into others called fission products. Some of those fission products absorb neutrons very efficiently, which makes the fuel only usable for a limited time. That used fuel sits next to reactors all around the country. The US had a plan to centralize it all in a repository called Yucca Mountain. That plan did not work out for many different reasons.

Now, what we do is we take that used nuclear fuel — I mentioned it’s mostly reusable materials — and we separate them out all. We use some of those fission products for things like the power sources or battery replacements that we developed and announced in January, and we take other products from that and use them as fuel in reactors.

CB: What can you do with nuclear waste and its derivatives?

SS: You have a piece of nuclear waste sitting in your house with you right now, in your smoke detector. Your smoke detector uses Americium-241, which was separated out from the nuclear waste and put into a smoke detector. It doesn’t have a very strong gamma emission, so it’s not hazardous to be around in the quantities that are sitting in your house, but the alpha particles emitted by Americium-241 will hit smoke — if you have smoke in your house — and set off your fire alarm. There are numerous applications for these fission products that come out of nuclear waste. Because 90% of the energy is still there, and you can make fresh fuel. And France does that today. They use Mixed Oxide Fuel made from nuclear waste. They reuse their spent nuclear fuel, and that bolsters their energy independence. So that’s one of the reasons that they haven’t had as many challenges with energy prices compared to, let’s say, Germany, with all of the turmoil of the war in Ukraine.

CB: What are the major limitations on battery technology in 2026?

SS: We’re limited by lithium ions right now when it comes to batteries. What we’re aiming to do with some of these fission products that come out of the spent nuclear fuel is make battery replacements. Think about an AA battery that lasts for 30 years. Those battery replacements can be used in different strategic applications. There are a lot of defense applications where a battery that doesn’t die on that timeline could be very helpful. In space, we have used radio isotope power systems to power the Voyager probes, to power satellites. It’s helpful in undersea applications as well. Getting power where it’s very hard to get power, is one of the advantages of these “nuclear batteries.” We don’t like that term too much because people think of a lithium-ion cell when they think of a battery — we call them power cells internally in our business. But battery technology is limited by chemistry, and we’re able to completely change that by using nuclear power, not the chemistry of moving ions back and forth.

CB: What do you think would be the most profound impact of this battery type?

SS: One of my favorite applications of these is for AI. A reason why nuclear is having such a resurgence right now is because the power output of a nuclear reactor is a match for the demand input of a data center. Our power cells are also really good at powering individual GPUs. One of the applications of that is edge computing: being able to do complicated computational exercises in contested logistics areas, or in places where it’s hard to get power or compute. There’s “big nuclear” powering a data center, and then there’s “little nuclear,” which is what we do in powering individual chips.

CB: How would you explain the sudden pivot in attitudes towards nuclear energy? It used to have this widespread persecution that seems to have abated.

SS: Old reactor technology, or improperly deployed reactors, have caused issues in the past. Everybody knows about Chernobyl. The challenge is communicating to the public that that sort of thing can’t happen anymore. That’s like back when electricity was in its early days, and you had people getting shocked to death by exposed wires in houses and things like that. We’ve figured out how to fix that. Radiation is similar to electricity in a lot of ways — back when we started putting electricity into houses and using electricity in a more widespread way, people were afraid of it. Radiation, when used properly, saves lives. Radiation in your smoke detector would save your life in a fire. The radiation that you get for different treatments in hospitals saves lives every day. Improperly used electricity, and improperly used radiation, are both dangerous, but I think people are more used to electricity, whereas they’re not as used to radiation.

CB: Where do you see the largest drivers of growth in terms of energy demand, and from which sectors?

SS: We need to reindustrialize in the United States. We need to mine rare earth elements and refine them. We need to manufacture parts. We need a manufacturing base in the United States. We outsourced all of that to China in the ‘80s and the ‘90s, thinking that we will be friends forever — and that’s just not the case. We’re in a situation where we still rely on, essentially, our adversaries. I think there’s going to be a huge electricity demand for manufacturing of all sorts. Nuclear is the direction we should be going, because it lets us have cleaner water and cleaner air than most other methods of electricity production. One of my favorite nuclear facts is that you get more radiation living next to a coal plant than you would living next to a nuclear power plant.

CB: Can you explain what betavoltaics are as a category and how it relates to what you guys do at Project Omega?

SS: The best way that I like to explain betavoltaics is by making an analogy to a solar panel. Light from the sun hits a solar panel, and the solar panel generates electricity. Betavoltaics operate with the same fundamental concept. Instead of radiation from the sun hitting a solar panel, you have radiation from a radioisotope hitting a radiovoltaic, and it generates electricity. You could think about a betavoltaic as a solar cell with its own sun.

CB: How saturated is the field of companies that are working on problems relating to beta voltaics at the moment?

SS: There are not many companies that are making radiovoltaics. We recently received a DARPA award, so there’s been a little bit more buzz lately. However, making reactors is probably a lot more saturated. There are also other companies that are working on radioisotope thermal generators — so, instead of a solar cell with its own sun, the radioisotope is used to generate heat, and that will heat up a heat transfer liquid or heat transfer medium. There are a number of companies that are working on different radioisotope solutions. One of our goals at Project Omega is to make these radioisotopes here in the United States. We need to manufacture the radioisotope in the first place to be able to power any of these applications.

CB: Where are they made?

SS: France makes a lot of radioisotopes because they’re the only western recycler, and China makes a lot of them, but they don’t export, and we don’t want to buy from China anyway, because we need to be more self-sufficient and produce things here in the US.

CB: How does one go about securing a DARPA grant?

SS: They’re public calls, so anybody can respond to them. We responded as part of a team — we’re a small business on that team, along with a large company, other small companies, a lab, and a university. Usually, awardees of DARPA projects are teams, groups of experts coming together to solve a problem. But there was a public call. Anybody could have submitted an application.

CB: How does the Department of Energy interact with companies like yours?

SS: The DOE signed a contract with all of the nuclear power plant operators in the ‘80s and the ‘90s called the standard contract — it’s public, you can see it. There’s an office of the standard contract in the DOE. The DOE said to all of the operators that if you build a nuclear plant, we will take all of the spent fuel, all of the waste away in 1998, and for a variety of reasons the DOE was unable to do that — and that’s where we come back to Yucca Mountain as one of the attempts to fulfill the obligations under the standard contract. Every year, the operators enter into settlements with the DOE for breach of the standard contract, because the operators still have all the spent fuel sitting on their reactor sites. The spent fuel is technically owned by the operators, but the liability is borne by the Department of Energy. We will work with both to take that spent fuel and use it for the United States and for our energy independence.

CB: Going back to the origins of the company — how did you formulate the original thesis?

SS: I think that this wouldn’t have been possible if it weren’t for all of the momentum in the nuclear industry, and specifically, the Executive Orders in May 2025 that explicitly highlight the need to recycle used nuclear fuel here in the United States. That was one of the key enablers for our business. The thesis started with recycling. There are a lot of articles that came out last year that talk about the nuclear resurgence — but what do we do with all the waste? Waste has been one of the vectors of attack from people who are anti-nuclear in general. I’d say one of the major criticisms of the nuclear industry is around its waste, even though it is very well managed today.

We started with that, and we did a lot of customer interviews. We talked to a lot of people that could utilize different aspects of the nuclear waste, whether that’s people who make fuel for reactors, or people who utilize isotopes in different ways. We found that one of the ways to make recycling economic is to utilize these fission products in power applications, and it turned out that there were a lot of strategic and critical applications that they could solve. We really took an economics-driven approach: how do we make the economics work for nuclear recycling, without just the uranium? Uranium is 96% of the mass of the waste, but uranium fluctuates the commodity market. You build a big refinery and you only have one product coming out of it, the uranium, it’s a very risky investment. We’re making sure that we have multiple products that can make money coming out of this refinery.

CB: Where are you in your product development stage?

SS: We’ve announced our first prototypes. These are small nuclear power cells that use strontium-90. We’ve announced those in partnership with Pacific Northwest National Laboratory. We’re now scaling those prototypes for specific applications for customers. Space is one example, scaling up those prototypes so that they can power a satellite. If you want to go to the dark side of the moon, you still need a power source for your satellite, or if you’re looking at going to Mars, or you’re going to deep space.

CB: On the topic of long-distance space exploration — how much have you thought about this?

SS: Realistically, it’s hard to use solar to generate power on Mars. That’s where nuclear power really comes into play. It is critical to have small, portable, and safe nuclear power for space. And that’s what we make. We’re not a reactor, but like I said, our technology has no moving parts. It’s a chunk of isotope with a chunk of semiconductor, which is robust.

CB: You’ve raised about $12 million to date. Is that number still accurate?

SS: $12 million is what we publicly announced in January, but we’ve raised more since then. We’ve brought on additional partners for our growth and building out our larger facilities.

CB: Do you have any growth plans that you’d be able to share at this time?

SS: We are building facilities; we’ll announce those very soon.

This interview has been lightly edited for length and clarity.

Like many Americans of my generation (I’m a Millennial, mea culpa), I grew up watching The Simpsons. Nuclear power, per the show, was lorded over by Mr. Burns, a business tycoon-cum-Nosferatu figure so old his social security number was a single digit. The waste that oozed from Mr. Burns’s plant eddied in Springfield’s waterways like a green, day-glo goo, spawning the town’s mutant three-eyed fish. Mr. Burns is a caricature as evocative as it is divorced from reality.

In truth, high-level civilian nuclear waste — the kind that most people worry about giving them an extra ear or seven fingers — has never harmed anyone in the nuclear industry’s entire history. (And, civilian nuclear waste is not a liquid, it’s a solid.)

Nuclear energy is often demonized, but it’s the best shot we have at powering our upcoming technological revolution. As the grid groans under the dual strain of increasing power demand and diminishing 24/7/365 power generation, nuclear’s firm, clean, reliable power offers the most coherent medium and long-term solution. In fact, nuclear is the most reliable source of power on the grid by a wide margin, boasting an average capacity factor of 92.3%. Compare that to geothermal (65%), combined cycle natural gas (59.9%), coal (42.6%), hydro (34.5%), wind (34.4%), and solar (23.4%). Nuclear is also the most energy dense source of power we have available; nuclear fuel is 2 million times more energy dense than any other chemical.

There’s only one problem: our fuel supply. Currently, we consume more uranium than we produce. At present, the US is reliant on imports for 90 percent of its uranium needs. Over the last few decades, we have ceded uranium production to other countries, primarily Canada, Australia, Russia, Kazakhstan, and Uzbekistan. And in May 2024, we banned uranium imports from Russia. The geopolitical supply risk has discouraged utilities from inking long term fuel contracts. As a result, the spot price of uranium has shot through the roof.

Before we dive into how to use our spent fuel, we need to understand what it is.

Nuclear fuel is made from low-enriched uranium (LEU) that is formed into gummy-bear-sized ceramic pellets of uranium oxide — the form of uranium we use for reactor fuel. The pellets are packed into hollow fuel rods that reach as long as fifteen feet, which are bound together into a larger fuel assembly, a fuel assembly being a bundle of fuel rods. As per the Government Accountability Office in 2021, “When operating, a typical reactor holds about 100 metric tons of fuel that are generally stored in roughly 200 to 800 fuel assemblies.”

Once stored in a fuel assembly, the uranium in them undergoes fission, whereby atoms split and collide, triggering a sustainable chain reaction within the fuel pellets, which is thus contained inside the fuel rods. This reaction has two effects: it creates nuclear isotopes and other radioactive materials (more on that later) and it generates heat. With that heat, the reactor boils water, which produces steam that spins a turbine to generate electricity.

Once the assemblies are “spent,” i.e., used up, they are moved into pools while they cool off — a process that takes about five years. After that, the fuel assemblies, while continuing to cool in the pool, are loaded into steel canisters. Once removed from the pool, the steel canisters are drained of water, welded shut, and dried off. From there, they are loaded into a steel and concrete cask about 20 feet tall, which sits on a concrete slab outdoors so it can be monitored by safety and security systems. It is physically impossible for the waste to “leak” out of these casks. The fuel assemblies and the uranium pellets inside remain solid, and solids cannot leak through other solids — especially solid steel. Think of nuclear waste as big metal straws, filled with uranium pellets, dropped into the center of giant steel and concrete empty water bottles.

So, how much nuclear waste is there? According to the Government Accountability Office, America has about 90,000 metric tons of spent fuel that are stored “on-site at 75 operating or shutdown nuclear power plants in 33 states.” This inventory grows by roughly 2,000 metric tons each year. To put this in perspective: If you gathered every cask containing America’s civilian nuclear waste and arranged them side by side, they would fit on a single football field.

Nuclear is the only power source that takes full responsibility for its waste. Wind turbine blades and solar panels end up in landfills. Fossil fuel plants “store” their waste in the atmosphere. Nuclear accounts for every ounce, stores it securely, and monitors it indefinitely. That’s why the waste is, in many ways, the best part about nuclear power.

But there’s one major problem with nuclear waste: we waste it. What we call “spent fuel” is actually 95 percent reusable. The fission process leaves behind isotopes and other radioactive materials which can be harvested to produce resources like medical isotopes and more nuclear fuel. It’s more accurate to think of nuclear waste as “lightly used fuel” rather than “waste” or “spent fuel.”

If we reprocessed our civilian waste, we would reduce its volume by 90 percent while extracting energy equivalent to five times greater than Saudi Arabia’s oil reserves. We would also save a massive amount of money. The Department of Energy (DOE) estimates that the cost for federal liability for storing nuclear waste will amount to around $40 billion. Beat here about how reprocessing would turn that waste into an asset. To take advantage of our waste’s left over material, we could reprocess it, which involves taking the spent fuel as described above and chemically treating or heating it to separate out useful material. That material could then be repackaged into pellets, fuel rods, and assemblies and loaded into reactors — or it could be harvested for materials to make batteries for lunar rovers or cancer curing isotopes, like Actinium 225.

But the US doesn’t currently have a waste reprocessing policy or a waste reprocessing industry. Why?

The Painful History of Nuclear Waste

What began as a technical engineering problem — how to safely dispose of nuclear waste — devolved into a political morass that would haunt the nuclear industry for decades. During the Manhattan Project, scientists developed ad hoc methods for handling radioactive waste — much of it liquid — at the Hanford site in Washington State. But when the Atomic Energy Commission (AEC) inherited these operations in 1946, internal reports flagged inadequate waste management practices.

The acceleration of the Cold War put greater urgency on handling the waste problem.

President Eisenhower, in an attempt to find a silver lining in the mushroom cloud, began pushing for civilian nuclear technology that could deliver abundant, clean power. The result of his vision, the Atomic Energy Act of 1954, laid the groundwork for America’s commercial nuclear fleet. Three years later, the first civilian nuclear reactor came online at Shippingport, Pennsylvania. Within a few years, utilities were rushing to build nuclear plants to meet electricity demand that was increasing 5-10% annually — a pace that would double consumption every decade. This boom would generate unprecedented volumes of commercial nuclear waste, and the AEC had no plan for managing it.

So, in 1956, the AEC made the reprocessing technology pioneered by national labs available to private industry, provided facilities and workforce development, and called for proposals for commercial reprocessing facilities. But reprocessing never achieved commercial viability due to technical challenges, uncertain worker safety, and a lack of waste to use as a feedstock for their activities.

As the reprocessing industry struggled to get off the ground, the AEC cast about for both waste storage methods and a waste repository, and struck a solution. Their initial experiments with salt storage were, as one Oak Ridge scientist said, “outstandingly successful,” as the salt could handle massive amounts of heat and radiation without melting. Early experiments at the mine in Lyons met with results that researchers found “most encouraging.” The AEC began to consider Lyons as a potential national repository site.

At first, the Lyons community was open to Project Salt Vault. But the Kansas Geological Survey (KGS) was not convinced that the AEC had done its homework. In 1971, Otto Rueschhoff, a local oil driller and salt miner, got wind of the AEC’s plans and raised an alarm: the mine they were looking at, Rueschhoff explained to KGS, was riddled with pockets liable to leak waste into the water table. When Kansas Governor Joe Skubitz found out that “the Lyons site is a bit like a piece of Swiss cheese, and the possibility for the entrance of fluids is great,” he announced that the “Lyons site is dead as a dodo for waste burial.” The AEC quietly abandoned the Lyons project, but the damage was done.

Challenges to the AEC’s authority continued to dog the commission as it pursued reprocessing and waste storage. By the 1960s, public fears of low-level atmospheric radiation exposure from atmospheric weapons testing ballooned, just as commercial nuclear power was expanding rapidly. Americans found themselves living closer to reactors, downwind from test sites, and increasingly aware of radiation in their food and water. This new proximity transformed abstract nuclear anxiety into concrete dread, and nuclear waste grew in totemic stature as an embodiment of these concerns.

These public anxieties entwined with a broader shift in American society, both cultural and institutional. The cultural aspect was marked by the emergence of the modern environmental movement, which repatterned American assumptions about big government, big business, and consumer society. A generation who came of age with Rachel Carson’s Silent Spring (1962) and witnessed oil spills, smog, and polluted rivers rejected the postwar consensus that economic growth should be pursued at all costs. The AEC found itself in the crosshairs of a movement that equated economic growth with civilizational suicide. This movement also produced a wave of environmental legislation — the birth of the Environmental Protection Agency, the Clean Air and Water Acts, and the National Environmental Policy Act — empowering citizens to challenge government and industry through litigation.

This new body of law provoked institutional change. While the AEC had accustomed itself to broad leeway, it now found itself subject to outside actors hostile to all of its activities. Indeed, both the West Valley reprocessing plant and the Lyons storage project encountered substantial environmental impact statements and other permitting challenges that the AEC had never encountered before. Eventually, the political challenges to nuclear became near-insurmountable. Though civilian reactors had roared onto the scene to meet grid demand over the 1950s and 1960s, the bumper crop of designs meant that reactors themselves struggled to come down the cost curve as well. This resulted in expensive delays, betraying nuclear’s promise of power “too cheap to meter.” Then the economic and energy crises of the 1970s flatlined power demand growth, robbing utilities of the business case for new reactors. Dozens of nuclear plants were canceled, leaving some customers footing the bill for power that never came online.

The travails of nuclear power impacted the viability of a sensible waste management program, especially as public hostility towards nuclear increased. Consumer rights advocates, non-proliferation hawks, and environmental crusaders all succeeded in painting nuclear as a villainous technology. Any radiation was seen as poisonous and hazardous to human health.

These trying times pushed the AEC to adopt cask storage — our current approach to waste management — as a stopgap until it could straighten out reprocessing. But the AEC wouldn’t live long enough to solve the waste issue. In 1974, the AEC was dissolved and replaced with the Nuclear Regulatory Commission, which was shorn of the AEC’s mission to promote nuclear power. Then, in 1977, nuclear’s main defender in Congress, the Joint Committee on Atomic Energy, was dissolved. The curtain fell on reprocessing that same year. President Jimmy Carter indefinitely suspended commercial reprocessing in the United States. By the close of the decade, the few reprocessing sites that had made a go of it all shuttered their doors.

But if not reprocessing, then what? The nuclear power industry had spent decades expecting reprocessing to handle their waste, and now that hope was over. The search began for a national repository for the large casks that had been sitting around at nuclear power plants. The Nuclear Waste Policy Act of 1982 revived geological storage as the official solution, prompting the DOE to issue guidelines for two permanent repositories. But the Act had no answer for the hardest question: where would the nuclear waste go?

The answer came five years later. The Nuclear Waste Policy Amendments Act of 1987 designated Yucca Mountain, Nevada, as the sole site for the nation’s permanent waste repository. Originally, the plan for waste management involved burden sharing, with a sister waste repository to be located on the East Coast. Instead, the project fell on Nevada, in part because officials became convinced that there was no need for another site.

Resistance to the Yucca Mountain site was immediate. Local tribes, environmental groups, and Nevada Senator Harry Reid opposed the project, calling the 1987 amendments the “Screw Nevada Bill.” Tribes cited concerns about the use of sacred lands, while environmentalists opposed the project on principle. Reid’s motivations were more political: why should Nevada shoulder this burden alone? It seemed quite convenient for eastern congressmen to foist Yucca on his constituents alone. Reid wielded incredible power in the Democratic Party and in Congress, enough to sandbag Yucca for decades.

The environmental and permitting challenges also plagued the project, making it harder and harder to license Yucca for use. DOE drilling and tunnel-making also revealed that, like at Lyons, Yucca’s geology was more complex than expected. In 2010, with $15 billion down the tubes and nothing to show for it, the Obama administration defunded Yucca. Today, civilian nuclear waste sits in dry casks at over 70 sites nation-wide. Utilities collect money from the government in damages since Washington never delivered on a national waste repository. The downfall of Yucca, coupled with the post-Cold War glut of cheap uranium fuel from Russia, sent all hopes of reprocessing nuclear fuel into hibernation.

We live in the boring present where lightly used nuclear fuel sits in big concrete casks scattered across the country bothering no one. It’s not a crisis, but it is a huge missed opportunity.

Why Now Is the Time

Nuclear enjoys more bipartisan support than it has seen since the late 1940s. Recent polling shows that “Support for nuclear energy outweighs opposition by 1.5 times,” and the partisan support gap for nuclear power is smaller than that of wind and solar. If ever there was a time to resolve the waste issue, it’s now.

We can make reprocessing a reality today. A burgeoning industry already exists, including companies like Oklo, Exodys, SHINE, and Curio. These aren’t brand new start-ups; some of these firms are over 15 years old. Some plan to use pyroprocessing, a technique developed by Argonne National Lab in Illinois that heats spent fuel to high temperatures to harvest useful materials like fresh fuel and medical isotopes. Pyroprocessing is expected to be cheaper and more efficient than waste recycling methods used in France or Japan. With a reprocessing industry in full swing, America could power a new wave of reactor technology, spur advancements in medical isotopes, and develop new battery technologies — just to name a few.

To put this future within reach, the government needs to behave like a market catalyst, clarifying regulatory routes to make the path toward commercial reprocessing reliable, explicit, and safe. Congress has begun adjusting the regulatory framework for reprocessing with the bipartisan Nuclear REFUEL Act. And President Trump has signalled his support for reprocessing in one of the four executive orders he signed on nuclear power. His Reinvigorating the Nuclear Industrial Base order calls for building supply chains that “maximize the efficiency and effectiveness of nuclear fuel through recycling, reprocessing, and reinvigorating the commercial sector.” This is one of the strongest signals a president has sent in favor of reprocessing since Ronald Reagan.

More can be done. President Trump must also modify National Security Memo 19, issued by the Biden administration in 2023. NSM-19 merely restated the pause on commercial reprocessing that has been in place since the Carter administration, though it did create programs under ARPA-e, the self-styled “disruption wing of the DOE,” to investigate cheaper methods of reprocessing. Trump could amend the memo to finally greenlight commercial reprocessing — but the details matter. The amendment should focus on uranium and transuranic separations that can be licensed under Part 70, the NRC’s nuclear materials licensing framework. This would dovetail with the bipartisan REFUEL Act and open the door to a closed fuel cycle in the United States.

Moreover the national labs could start qualifying fuels for recycled materials. Qualification is the process by which we subject various nuclear fuels to a variety of tests — putting them through wild heat swings, stress testing their cladding, etc. These tests then establish the regulatory standards for what qualifies as commercial grade fuel that can be loaded and unloaded from reactors. This is a laborious process because it needs to be. The DOE also provides a path toward fuel qualification through fueling test reactors, a process that could synergize reprocess and new reactor technology. Whichever way, the sooner we get started on qualifying reprocessed material for commercial reactor fuel, the easier it will be for reprocessing companies to create products that reactor operators can buy.

Similarly, establishing clear regulatory guidance for workforce training standards, licensing fees for reprocessing, and decommissioning requirements would map out the full life cycle of reprocessing. Companies could then cultivate a competent workforce with uniform skills. Secondly, they could smoothly begin operations with explicit, consistent, annual licenses in place. Lastly, they would also know how to break down and remediate their work sites should they ever close their doors. While all fuel qualification and other guidance may appear “in the weeds,” it’s nailing down these kinds of issues before they become problems that can make or break a nascent industry.

But that still leaves the hardest part: finding a place to put the waste. Even after reprocessing, some waste will need to find a permanent home. The Trump administration has signaled that it wants to cut deals with states to host waste. According to recent reporting, the DOE “will invite interest from states on deals for nuclear power that would also offer incentives for nuclear waste reprocessing and uranium enrichment.” Several states have already expressed interest. Additionally, the DOE is soliciting interest for Nuclear Lifecycle Innovation Campuses. Imagine nuclear innovation hubs that power data centers and advanced manufacturing while supporting nuclear from cradle to grave — from enrichment, to reactor construction, to waste reprocessing.

All wealthy societies use incredible amounts of energy. This tight coupling suggests that wealth and stability track closely with how well a society harnesses thermodynamics. To reprocess our lightly used nuclear fuel innovation in energy, medicine, and exploration would mean climbing up the energy ladder. It would mean true, lasting energy dominance.

Neros manufactures first-person-view (FPV) drones for military purposes. Its flagship product, the Archer, was developed after Monroe-Anderson and his co-founder Olaf Hichwa visited Ukraine in 2023 to provide combat-ready hardware to be tested in battlefield conditions. In 2025, Neros was selected as one of the US Army’s primary FPV drone suppliers in the Purpose-Built Attritable Systems (PBAS) program. Neros is one of many American companies poised to benefit from the Pentagon’s $1.1 billion Drone Dominance initiative, launched via executive order last summer, which seeks to bolster the drone industrial base and procure hundreds of thousands of low-cost, US-made, attritable drones for the military by 2027.

I sat down with Soren to discuss his early immersion in FPV drones, the founding and growth of Neros, and his experience as a young CEO leading a growing defense company. What follows is a transcript of our conversation.

CB: When did you first develop an interest in drones?

SMA: I grew up in Anchorage, Alaska, and I was obsessed with building things and taking things apart. I was flying remote control airplanes and driving RC [remote control] cars. When I was 11, I started seeing FPV drones pop up online. This was pretty early, and it was a very small group of hobbyists online who were figuring out the technology. I thought it was super cool. My dad actually ran a mapping company and was interested in using drones for aerial photography, for map data. I would fly drones for his company because I knew how to fly RC airplanes, and he would get me drone parts in exchange. That’s how I got my first set of FPV goggles, for my 12th birthday. I started by putting a camera and video transmitter onto a DJI Phantom 1, the first ever DJI drone, and flying it around FPV. I kept crashing it because it wasn’t really made for that, so at 12 I built my first racing drone, and then from there, I got totally obsessed with it. It was more technical than other RC stuff; the building challenge was way harder. But to put on the goggles and feel like you’re flying was amazing. That was the thing that captured me.

CB: What did you build next?

SMA: I built a racing drone from a bunch of random parts. The way racing drones are generally built is that you buy all the separate parts and put them all together. You get motors, speed controllers, flight controls, and frames from different companies. It’s like this whole mix-and-match. I copied a build that someone else had done on YouTube, where they had this long series. I watched it three or four times, and I just tried to copy what they did.

CB: What are the attributes that make a drone better at racing?

SMA: The main thing is they have a really high power-to-weight ratio. What’s kind of amazing about the technology stack that enables this — modern batteries, modern flight controllers and modern motors — is that you can have a drone with a 10-to-1 power to weight ratio very easily. They accelerate insanely fast, and can go over 100 miles an hour, and get there really, really quickly. The racing drones are also different from other drones because they don’t put any limit on the angle the drone can achieve. Normally, if you buy a drone, you know it’s gonna limit you to a certain amount of angle — you can’t go past it. And then, when you let go of the stick, it’s gonna stabilize itself. With racing drones, you can do whatever you want: flip it upside down, roll it around. When you let go of the stick, it will hold the last input. You can put the drone on a 70-degree angle, and it’s gonna start going forward at 100 miles an hour. Then you can turn it upside down. That super agile, full manual control is one of the other main things that differentiates a racing drone.

CB: When did you decide to turn this hobby into an entrepreneurial endeavor?

SMA: I started going to competitions, and there were pilots who were racing on these sponsored teams. I saw people who were just a little bit older than me doing it and I thought it was incredibly cool. I set my sights on becoming a professional pilot and competing at the highest levels. After freshman year of high school, I stopped doing any school sports so I could practice drone racing more. It was around that time that I started flying professionally. Through high school, I felt like I was learning more from drone racing than from school. There was this interesting technical side of it that I was getting, and I was starting to compete at a very high level. I ended up starting a company when I was 16, selling parts for drone racing. I had my hands full, so I ended up taking more classes online so I could spend more time on drone stuff. I probably missed 50 percent of high school, when you combine all the time I was gone for drone races — and then Covid happened in the spring of my junior year. I was out for that whole spring, and then I didn’t go back in person for senior year because of Covid. I stayed remote, kept doing drone stuff, and graduated a semester early.

CB: How is it that DJI and China became such dominant players in the FPV drone market?

SMA: It’s a few different factors. Number one, China became dominant in all consumer electronics. That happened because America decided we didn’t want to do this dirty manufacturing at home. American corporations went overseas and literally taught China how to build these factories. DJI’s rise benefitted from the rise of Foxconn and the whole consumer electronics industry in China. They were really early in drones, but by the time DJI got going, China had by far the best smartphone industry in the world. They had the amazing advantage of the Shenzhen industrial base, where you can instantly source any part, everything is cheap, and there’s so many consumer electronics manufacturers.

Number two, they started very early in the drone space, and they had a really viable drone consumer product. That compounded because they had the highest volume, which made for bigger and bigger investments. They were also subsidized by the CCP. The CCP understood the strategic advantage of having the number-one drone company in the world. It was a totally unfair advantage for DJI, and no one is going to be able to compete with that outside of China. Over time, they built such an incredible technology stack that they actually absorbed multiple smartphone chip companies into DJI to make custom silicon for drones. This is kind of crazy. It took Apple a long time to make custom silicon for iPhones and for Macbooks, right? And DJI did that for drones, as a company, because they have this insane advantage bestowed by the Chinese government.

CB: Tell me about your Thiel Fellowship. How did you apply successfully?

SMA: I had an interesting experience, because Thiel Capital actually backed Neros a few years ago. So it wasn’t my first interaction with the Thiel universe. I’d known about the fellowship for a long time and decided to apply right at the end of when I was age-eligible. I can’t speak to the specifics of the process, but I found it extremely valuable. The group of people they select is phenomenal. I’ve made some very close friends through the program. Neros was a little bit later-stage than most of the fellows’ companies, so I had a slightly different experience where we were already up and running.

CB: In developing the Archer, can you walk me through the testing process?

SMA: Product development is heavily based on real-world feedback; this is something we’ve focused on since the beginning. When we started Neros, the first thing we did was build 30 drones, bring them to Ukraine in our suitcases, and start getting actual feedback. Doing that type of thing is what has accelerated us to having an extremely useful, battle-proven, mature product that we’re really confident in and that users really love. It sounds kind of obvious, but most American defense companies don’t actually do this. They don’t prioritize getting their stuff into real-world combat at scale. I think that’s what’s differentiated us the most — that we made it a priority and did so many iterations based on that feedback. We also have a very rapid testing process internally where everyone on the flight test team is a world-class pilot and trained as a scientist. They’re out in the desert every single day, pushing our gear to its limit, doing brand new development, testing or qualifying an upcoming product. It’s that constant cycle that makes it possible.

CB: I read that one of your primary goals at Neros is to secure the supply chain so you can build FPV drones at scale with little to no Chinese input.

SMA: From that first trip to Ukraine, what we saw was companies building at the scale of tens of thousands of drones per month — but everybody was doing it on Chinese parts. When we started in 2023 you could not buy an FPV drone that didn’t have Chinese parts. It was literally impossible. We knew that if we’re going to scale this technology — which we had learned about from the battlefield in Ukraine and from our deep connection with users — but to actually scale it, we had to go and make all the components ourselves, going down to the really deep levels: the chip level, the material level, on components — to eliminate China from the supply chain. We saw that as a necessary thing for using these drones in defense applications. We started on that problem earlier than most did in America, and now we’re able to do high-rate production — at least the highest rate in America — while also maintaining that secure supply chain. That’s because of the work we’ve done over the last two and a half years.

CB: What specifically are those components?

SMA: The major components are motors, sensors, all the electronics, the flight controller, the radios. We spend a lot of time on radios, especially building something that’s jamming-resistant. You have radios on the drone side and on the ground station. The ground station is a system people kind of overlook. Then you have the airframe of the drone, small things — antennas, stuff like that. For all of those electronics, there are so many different components that traditionally would come from China. You have to go and kind of re-engineer everything around a different set of microcontrollers, since traditional ones come from China. We’ve done all of these things from the ground up. We control what microcontrollers and other components are going into these drones, so we can ensure a secure supply chain through all of that.

CB: What are the battlefield lessons of Ukraine? I would break that down into two categories: the tactical role of the FPV drone, and the evolution of countermeasures that are used to stop those drones. How did you convince the Ukrainian forces to take their time to give your drones a test?

SMA: We got in contact with a group doing pilot training in Ukraine, and the message was, “Hey, this is who we are. How can we help? Tell us the problems you’re having.” We found these groups were really excited to talk to us and try out the hardware. There were a lot of things they liked and a lot of things they didn’t. We always found a strong percentage enjoyed talking to us and working with us. That became the way the company operates. We have a team dedicated to working with the military units that receive our drones. Over time, it’s grown. I’m in Ukraine very frequently; it’s just part of what the company does now. And we built those relationships, and we have units that really love our product. It’s been a natural evolution, and totally critical to our development.

CB: On the tactical front, what have been the biggest developments over the last few years as you’ve been watching the conflict?

SMA: The biggest thing that FPV drones deal with is electronic warfare. Radio systems have evolved a lot and been a really big focus. How you avoid jamming is one of the key questions. This is also why fiber-optic drones have become a really big deal: optic connections can’t be jammed by a traditional RF jammer. That’s been one of the biggest shifts, is this sort of cat-and-mouse game of new radio, new jammer, or fiber-optic drone. Now you need a different way to counter that drone, and then you’re left with things like shotguns and nets. That’s really primitive, but it is the last line of defense. We’ve seen that constant evolution of UAS and counter-UAS.

We’ve also seen the ranges of drones and targets get much longer. In ‘23 you could very often see a five kilometer effective range, because there were all these valuable targets across the front line. Now, because there are so many drones across the front line, there’s a 30-kilometer-wide zone where almost nothing can move. If you try to put a tank into that zone, it’ll be so quickly destroyed by a horde of drones that it doesn’t really work anymore. In that zone, you either have literally, soldiers on foot — Russia’s new tactic is sending soldiers on foot or on motorcycles in small groups to try and break through this line — or you just have drone-on-drone warfare, where there are a lot of drones flying over and then you have interceptor drones going after them. That’s the biggest shift, I would say, is the fact that this zone is widening, all because of drones. If you want to strike a valuable target, you are probably going 30, 40, 50 kilometers in to hit something like a logistics line.

CB: Any interesting breakthroughs in kinetic interceptors?

SMA: What’s been interesting to me is watching the evolution of Shahed interceptors in Ukraine. Since the full-scale invasion, Russia has been utilizing what was originally the Iranian design, the Shahed drone, and then they set up the domestic production of these drones. They use them to bombard Ukrainian cities. It’s always been a big problem, but about nine months ago, interceptor drones started working as a counter to Shaheds — and it’s all happened in the same way that FPV drones did. Someone in Ukraine has an idea and starts trying it. It doesn’t work at first, but you keep iterating, and eventually you get to a solution that can work. Now Ukraine has gotten really good at using these small, cheap interceptors to go up and take out Shahed drones. Where you used to need a very expensive missile to take out one of these drones, you can now use a drone way cheaper than the actual Shahed itself to take it out. With what’s happening in the Middle East, America is looking at this and realizing that we have to have this technology too — and so does every other country in the world. FPV drones are going to be a ubiquitous tool, and I believe that cheap interceptors will be as well.

CB: Tell me about the products you currently have in development at Neros.

SMA: Right now, all of our drones that you can buy today are manually piloted. We are working on an autonomous variant of the product. We’re really focused on making autonomous features that are extremely useful to the human operator without trying to replace them. The question we’re asking is: how do we give the human operator leverage? What I’m most excited about is some of the stuff we’re doing there, and the autonomous drone, the first version of which will be coming out this year.

We’re also continuing to go deeper and deeper into the supply chain and the whole FPV ecosystem. We’re going to own the entire FPV stack end to end and figure out how to build it all without China’s supply chain. Right now, we’re focusing most of our efforts on getting that full ecosystem done and making sure we do that the absolute best we can before we get distracted by other product lines.

CB: What’s the impact of the Drone Dominance initiative, and where are we as a country in fulfilling its objectives?

SMA: The goal is to provide a clear signal to US industry — clear demands — and help kickstart this process of scaling manufacturing. I think it’s doing a very solid job of that. When you have something as high level as Drone Dominance, coming down from the Secretary of War, saying “one of the most important things to the Pentagon is scaling small drones, and here is $1.1 billion to back it,” it really creates a sense of urgency within the industry.

I think what’s going to be very interesting to see is, I think a lot of companies that got selected for the first round of Drone Dominance will struggle to deliver the volume of drones — even though it’s a really small volume of drones. I think a lot of those companies will struggle to deliver. We’re gonna start to funnel down to the companies that can actually deliver. And I think for the second gauntlet, which hasn’t happened yet, we’ll be going into that with the context of who is successful in delivering Gauntlet 1. That is going to be, is going to be very interesting,

CB: Where are you in your growth journey and funding journey as a company?

SMA: We just moved into this new facility. It’s 250,000 square feet. We sized the facility to be able to do a million drones per year. The goal that the company is working towards right now is being the first American company that can build a million drones a year. We care about that because at that scale, we start to have some impact on credible deterrence for the country. We want to start to move the needle in terms of deterring conflict.

On funding, we did our Series B last November. To date, we’ve raised about $121 million. The company has about 150 people right now, and we’re growing. We are hiring. This year is going to be an amazing year of growth in terms of team size, production, and building out the factory. It’s a really exciting time for folks to jump on board.

CB: How old are you?

SMA: I’m 22.

CB: How do you approach managing people who have more years on you or more experience in different categories?

SMA: Almost everyone at the company has more years of experience than I do. I only try to make an executive call or hold a really strong opinion on a small number of things. Most people know most things at the company better than I do, and I should often defer to the experts. I also think that my intuition and direction have been pretty good so far and have gotten Neros to where we are. But I’m informed by all the expert opinions around me.

CB: What have been your most valuable sources of mentorship as you’ve built Neros?

SMA: A lot of mentorship has come from the backers of Neros. We have a pretty amazing cap table. I’ve learned a ton from Shaun at Sequoia, John at Vy Capital, who was our first-ever investor. We have some amazing folks behind us and have gotten a lot of mentorship from them. I also try to learn from the incredible people at the company every day. Everyone here is excited to learn from each other, and that’s a really important aspect of our culture.

CB: If someone’s reading this who might be interested in joining Neros, how do they find out what you’re looking for, and how do they get a hold of you?

SMA: We have a bunch of jobs posted at neros.tech/careers, and there’s a role for almost anybody up there. But if they don’t see a perfect fit, reach out to info@neros.tech, or to me personally on LinkedIn. I may or may not see it, but we’re always looking for unique individuals and high-agency, high-performance people. People can come prove they’re the best in the world here. And that’s exciting for a lot of folks, so I highly recommend reaching out. And even if there’s not a role, we will probably find one for you.

This interview has been lightly edited for length and clarity

All of that may come as a surprise to anyone who sees Mormons as an odd, yet simple offshoot of Protestant Christianity. But the reality runs much deeper than that. The heritage of success in business and technology is tied not only to Latter-Day Saint history, but the religion’s very theology. From Bryan Johnson (Don’t Die) to Clayton Christensen (Innovator’s Dilemma) or Keith Johnson (the head of Sequoia’s family office), and across the founding stories of companies like Adobe, The Trade Desk, JetBlue, Stance Socks, Vivint, Ancestry, and countless others — the spirit of the quintessential American religion has left its mark.

An American Moses

America is a nation built on the dual concepts of restoration and reinvention. Making what is old into something also new. Americans reinvigorated and perfected little-r republicanism, little-d democracy, and inalienable rights.

American religious life is no different. While we transplanted Old World religion, we also made way for the reinvigoration of American prophetic voices. Where the ancient Moses had tablets, as a member of the Church of Jesus Christ of Latter-Day Saints, I believe that the American Moses had a pen.

In 1831, Joseph Smith, under a prophetic mantle, took up that pen and wrote words that fell from the lips of God, as far as we’re concerned. In what he wrote, you find articulated a fundamental ethos that rings true in the heart of every so-called Mormon: “Men should be anxiously engaged in a good cause.”

While we may not like being called Mormons, members of The Church of Jesus Christ of Latter-Day Saints have ultimately felt that anxious pursuit driving us forward. It’s how we became, arguably, the most influential American-made religion, spanning millions of members across 150 nations. Almost any entry over the last 200-or-so years of that history could justify its own book.

Put Your Shoulder To The Wheel

Chased from a half dozen towns, murdered and persecuted along the way, we built a city out of a swamp in Nauvoo, Illinois. At its peak, it reached a population of 12,000, rivaling Chicago at the time. After our first prophet was murdered, our second one, Brigham Young, led 148 fearless pioneers across a 1,300-mile trek to Salt Lake. Then, we turned around and spent the next twenty years helping 65,000 other people join us there. We spent over $50 million — adjusted for inflation — helping poorer European immigrants make the journey, spreading out across the west, not just Utah, but also Idaho, Arizona, and California. And, these frontiersmen succeeded: the first millionaire in California, Samual Brannan, was a Latter-Day Saint businessman literally selling picks and shovels to gold miners.

We imported industries wholesale to Utah to reinforce our self-reliance. From iron to textiles, we saw physical toil as a spiritual commandment. Our own scriptures reinforce that every commandment from God is spiritual; never temporal. We didn’t build industry to get rich, we did it to build the Kingdom of God, our own home-grown Zion. In fact, arguably the first full-line department store in the US was the Latter-Day Saints’ Zion’s Cooperative Mercantile Institution (ZCMI) founded in Salt Lake City in 1868 that pooled industrial output and local buying power into one enterprise.

Out of a barren desert, my ancestors built Deseret (the name we wanted for what is now “Utah” before the feds forced us to abandon the name), a Book of Mormon word that means honeybee — the perfect mascot for a hard-working, industrious hive.

By the 1860s, out-of-state travelers said Utah looked more akin to established Eastern towns like Chicago or New York, rather than the western outposts they were expecting. Brick homes, wide streets, mills, stores, schools. In fact, at a time when twenty percent of the US population was illiterate, only 2.5% of Utah had the same disadvantage. Our economy was well-rounded and strong, resisting the swings of the mining boom towns that then dotted the American West. Utah was built for broad stability, not quick fortunes.

Adding Investment To Industry

The first generation of Latter-Day Saint entrepreneurialism after Utah’s settlement revolved around industrial categories: iron, sugar processing, utilities, and railroads. All were risky capital-intensive endeavors. When the Union Pacific and the Central Pacific railroads finally met, connecting the east and west coasts, they did so in Promonotory, Utah. The arrival of the railroad in 1869 brought cheap products from the east, crowding out local industry. The Church was left carrying millions in debt with industries that were much more exposed to competition.

In the late 19th century, Utahans, determined to create heaven on Earth, continued to build everything from Church-owned enterprises, across media, insurance, and construction, to adjacent projects in the state. Founded in 1940, Hill Air Force Base became a major Air Force base with 22,000 personnel during World War II, offering critical maintenance. Thiokol, founded in 1929, manufactured everything from rubber to missile propulsion systems, including Minuteman ICBMs and the airbags on the Mars Pathfinder. Geneva Steel, founded in 1941, was at one point producing 60% of the steel in the Western US.

The same tenacity that powered my ancestors through spiritual and physical obstacles in the quest to find our American Zion prepared us to tackle financial obstacles at the turn of the century. Weathering the Great Depression and World War II redoubled our focus on self-reliance. One old pioneer motto came in handy: “use it up, wear it out, make do, or do without.” The Church created the Church Security Plan in 1936 to supplement government relief, provide food, jobs, and resources, and encourage self-reliance. By the 1960s the Church was solvent, but not wealthy. Then, N. Eldon Tanner helped introduce modern financial discipline into the hearts of every Latter-Day Saint.

Tanner was an oilman and politician in Canada, but was chosen as a modern-day Apostle in 1962. That was an unusual call at the time, given most Apostles came from academic and legal backgrounds. Upon becoming a senior leader of the Church, he was “thrust into financial affairs.” He took the raw fundamentals of self-reliance and grit among the Latter-Day Saints and supercharged it with corporate-style accounting, an aversion to debt, and a diversified investment portfolio.

Tanner pushed the Church towards an additional golden rule: “We do not spend money we do not have.” From debt-ridden to conservative spending to building up long-term reserves, the Church’s financial picture changed dramatically. Over the next 60 years, the Church built on a foundation of industriousness and financial discipline following scripture. The call from God was to “organize [ourselves] and prepare every needful thing.” So we did.

By the 2020s, the Church had built a net worth of $265 billion. That includes 290,000 acres of ranch land and citrus in Florida, one of the largest cattle ranches in the country, billions in real estate, a $55 billion equity portfolio, and on and on.

The entire purpose of the Church’s aggregation of financial assets had nothing to do with amassing wealth. The late president of the Church, Russell M. Nelson, may have been the head of a $265 billion organization. But he had a net worth of a few million as a former renowned heart surgeon, and was earning a $120,000 annual salary working full-time for the church, all while wearing a $30 Timex Casio wristwatch and vacuuming his own house.

So what, then, does the Church do with the $265 billion empire it has built? The leaders of the Church will be the first to reemphasize the mission: “help people learn about and live the teachings of Jesus Christ, to share that message with the world, to strengthen and unite families, and to care for the poor and the needy.” The Church spent $1.45 billion in 2024 alone on humanitarian aid and welfare efforts across over 3,000 projects in 192 countries. Beyond physical needs, the Church invests in supporting its members’ spiritual needs. That includes access to over 200 temples. Not just casual chapels, but sacred sanctuaries we believe are literally “the house of the Lord.”

We built industries, amassed financial assets, and gathered a reserve that can support millions of people spiritually and physically. But we didn’t stop at shaping the physical world. The digital world has been just as fertile a playing field for Latter-Day Saint grit.

“The Glory of God Is Intelligence”

In February 2024, Sam Altman said something that made the ears of every Latter-Day Saint perk up: “I grew up implicitly thinking that intelligence was this really special human thing and kind of somewhat magical. And I now think that it’s sort of a fundamental property of matter.”

Latter-Day Saint scripture teaches that all matter is eternal and can be neither created, nor destroyed. It can simply be organized, right in line with the law of conservation of mass. Where religion is often framed as being at odds with science, my religion offers a fundamental cosmology that has remained the same since at least 1833 (if not since, you know… forever).

Far from a congregation of Luddites, Latter-Day Saints have embraced science as frequently as possible. One of the most noteworthy entries in Utah’s technological history came from the University of Utah’s computer graphics lab in the 1960s and 1970s. Our own home-grown Traitorous Eight, this group yielded an insanely dense pool of technology titans from Ed Catmull (founder of Pixar) and Jim Clark (founder of Silicon Graphics and NetScape) to John Warnock (founder of Adobe) and Alan Kay (one of the most influential technologists to come out of Xerox PARC).

They may not all have been “Mormon” (Alan Kay and Jim Clark weren’t), but their technical training sprouted from an institutional environment fostered in Utah where, at the time, 70% of the population was LDS. That environment also enabled adjacent technology history to grow up around it. In 1969, in large part because of that graphics lab, ARPANET (a fundamental precursor to the internet) set up its fourth node at the University of Utah! (The first three were all in California.)

One graphics lab alumni, Alan Ashton, developed WordPerfect in 1979, which eventually became the dominant word processor in the 1980s before Microsoft Word came on the scene. That same Utah environment birthed Novell, an early pioneer in networking founded in 1980, reaching seventy percent of enterprises at its peak, and Cirrus Logic, the first fabless semiconductor company.

Most of that technical prowess in Utah over the second half of the 20th century grew out of a comparative advantage. Members of the Church were disproportionately literate, multilingual (almost half of Latter-Day Saints serve on a two-year mission, often learning a second language), globally comfortable, industrious, and thrifty. But every aspect of that character grew out of our heritage.

Our Heritage

Members of The Church of Jesus Christ of Latter-Day Saints make up just two percent of the American population, but we’ve taken seriously Christ’s admonition to be the “salt of the earth.” Our hope is that our heritage is clear from the way we live our lives. We work hard to build things and impact people as positively as we can.

In the first book of the science fiction series, the Expanse — written by a non-Mormon author — there is a side plot where, in the year 2350, “the Mormons” have built the LDSS Nauvoo; an interstellar ship — a “portable Eden” — running the length of six of the longest aircraft carriers in the world.

In a world of musicals like The Book of Mormon and other “cultural” artifacts like the reality TV show, The Secret Lives of Mormon Wives, a more true to life portrayal of the Latter-Day Saint character would be calling out the grit, self-reliance, and commitment that it would take to self-finance an intergalactic ship. The LDSS Nauvoo represents an embrace of, if my co-religionists will allow me, a techno-Mormon future. And the utopia we’ve been tasked with creating on Earth is just as much a techno-utopia as it is a spiritual utopia.

Last month, Pax closed its first fund after raising $52 million from investors. I spoke with Michelle about her vision for Pax Ventures, her love of running, and her observations on Silicon Valley’s shifting relationship with the defense sector. What follows is a transcript of our conversation.

CB: I saw you ran D1 track and cross country. Are there any particular highlights from that part of your career, or persistent lessons that come from competing at that level?

MV: I’ve been running for a very long time. My mom ran in college, my older brother ran in college, so I was introduced to running at a young age. I actually ran for the same coach and college that my mom ran for, which I thought was very funny. Minnesota is a very small, insular community. And I would say I always had this very deep competitive streak. When I graduated, I had to keep doing races to give myself something to compete for, and ended up doing a lot of marathons, and went after this very obscure Guinness world record for the fastest aggregate time running a marathon on all seven continents. There’s a small number of competitors in the pool of women that have run on every single continent. But it was this nine-year, lifelong goal. It takes a while to find all these marathons and train for them and run them. And I would say I need to have a race on the calendar, or a goal to be striving for. That’s how I motivate myself. There are different types of motivation. If you know yourself, you know what type of motivation drives you. Mine is very goal oriented. I need a goal to force myself to train.

CB: You need to be in training mode.

MV: Yeah. Or going after a goal mode, going after a certain time, or some kind of record.

CB: What’s your proudest race?

MV: My race in Antarctica, I won the whole thing. I beat all the men. It’s probably the only race where I get to say that.

CB: How did you end up in venture? You were at Palantir, later Andreessen Horowitz. What drew you into that field as a competitive person?

MV: I worked at Palantir and then worked at a couple of different startups after that. At the last one, I was on the founding team. It was a real estate startup. It’s dead now. I applied to business school, as sometimes lost people do when they’re trying to figure out their life, and got into MIT. When I went to MIT, I thought I was going to start a company, and I wanted to get back into national security. I knew the Anduril team really well. It was just starting to really take off. This was 2020, before Russia invaded Ukraine. So I got to business school. I started a defense tech club. I did a year-long fellowship with the Defense Innovation Unit, meeting all the founders at the time — a much smaller number than now — worked in defense. I didn’t expect to go into investing, but because I was meeting all of these founders, I was then also meeting a lot of investors and hearing how people thought about the space.

Everyone hated defense. They’d say things like, “Anduril is not even a real company.” People were hating on Palantir, in 2021 everyone was like, “tech-enabled services aren’t a thing. Palantir is just consultants.” I would get so annoyed. I started developing this defense and dual-use investing thesis, mostly because I was arguing with all these investors all the time. And then I got very lucky. Andreessen Horowitz announced they’re starting this American Dynamism team, and I was in the right place, right time, and ended up being the first partner they brought on to that team. Turns out I love investing. I love company building. And now I get to help founders a lot at that stage. It was this perfect blend of getting to do the investing side that I love, but also getting to build something of my own.

CB: Why did Silicon Valley’s relationship with defense sort of die out in the post-Cold War era, and how would you explain the cultural shift that has taken place in recent years?

MV: There’s a few different ways to answer that. One is on the tech side itself. The government used to be the main innovator of technology — creating the internet, GPS, and a lot of the tech we use today. They really started this era of the computer software revolution. Now we’re seeing a lot of pull from the private sector. The government is still providing the contracts, but they’re pushing the R&D and innovation to the private sector and letting the private markets participate in that part of technology creation. I think that’s actually a very good balance.

Government should still dictate some of the requirements and reward the things that they want and are working well. You can look at SpaceX — a lot of their early contracts were from the government. Palantir got lots of early government contracts. But the ability to have R&D dollars come from the private sector allows for a broader pool of competition to bring more interesting technologies to the forefront. So I think the shift for Silicon Valley is, in some ways, going back to its roots of building for defense, but opening up the markets a little bit more, and opening up the founder pool a little bit more, which I think is great.

On the cultural shift side, there was definitely a period in the 2010s where people got allergic to all things government, all things defense. My sense of what started to happen is that people began realizing the period of relative stability they had taken for granted was starting to wane. And now people who are very talented and don’t want the country to stagnate are getting involved.

CB: As somebody who’s worked inside startups and allocated capital, what would you say is the main differentiating factor between a successful founder and an unsuccessful one? If you can point to just one — or maybe there’s multiple?

MV: VCs constantly try to pattern match, and it ends up being the outliers who are the ones that create totally new categories. What I have found is that it takes extreme grit and determination and a different viewpoint on a market that comes from experience and knowledge and understanding, and the resilience to go make that happen. People at Palantir — which often spins out founders — would describe it as people that are just willing to chew glass for longer than other people, which I think is very true. Starting a company is very hard. Everything that can go wrong will go wrong. People glamorize being a founder, but it is the loneliest, hardest job out there. You have to really, really want to make this thing happen. You just have to have the pain tolerance to make it happen, which is probably the number one founder trait.

CB: Is money a good motive in itself? Or does it distract from the big picture?

MV: Ambition, and wanting to be very successful, can work as a motivation. I would argue that it shouldn’t be the only thing motivating you, but I don’t think it’s a bad thing to have. For example, I would like to have a very nice house. It’s a helpful additional motivator for me. But it feels very meaningful to be supporting founders working on important things.

CB: In some of your thesis materials, you mentioned Pax Technica. You’re observing the world shifting in a new direction, and you want to capitalize on that, so tell me a little bit more about that element of your thesis.

MV: The name Pax comes from a reference to Pax Americana and Pax Romana, historical periods of relative stability defined by who had the best technology. The Romans had the best engineering. Britain had the best industrial might and the best navy. And then Pax Americana had the best technology and nuclear power for a long time. Geopolitically, for the first time in my life, we have a great power competition. I think the next era of relative stability will be defined by who has the best technology. I want to back the founders who are building that technology to ensure peace and prosperity.

CB: If you were to group the most consequential technologies in development right now, from an investor’s standpoint, what are you looking at? What opportunities do you see?

MV: The most exciting categories, which I have invested in, are where software and AI intersect with the physical world. That wasn’t the case 10 years ago. Software was eating software. Now, I think for the first time, the hardware has caught up, and they’re advancements that allow for software to be deployed at the edge. Software to be deployed in manufacturing facilities, autonomous machinery and robotics can come in to improve economics or work in places that previously weren’t able to be innovated in.

For example, one of my companies is a mining company, and they are bringing a lot of automation into mining operations. That’s allowing a big copper mine in Utah to be reopened. They didn’t have the skilled labor, they didn’t have the equipment, the economics didn’t work. Now they’re restarting this gigantic copper mine and employing a ton of people in the area, and that is only possible because of the technology that exists today.

Starlink is another enabling technology, similar to maybe GPS, in some ways. Where GPS ushered in this whole new wave of companies when it came out; Uber couldn’t have existed without smartphones and GPS. But when GPS satellites were launched for the government, no one thought “Aha, like, now where there’s going to be some sort of marketplace for cars.” Somebody innovated because of an enabling technology that existed. I think Starlink is the next enabling technology. What can you do when you have connectivity everywhere, and you’re not limited by the physical infrastructure? That means internet on planes, cars, boats, in the ocean, in mines— places where you couldn’t bring a bunch of cables.

So you have deployment at the edge, AI becoming much better. Chips have become smaller and more energy-efficient. There’s a lot of innovation happening there. I have a couple of nuclear investments, which I think are really exciting. And then, autonomous systems in general — I think the next age of autonomy is upon us, whether that’s in defense, whether that’s in robotics and manufacturing, whether that’s in transportation and supply chain. We’re going ot see a whole new set of really exciting categories.

CB: What are the main inefficiencies in mining that AI in particular will solve?

MV: There are two major areas. The first is the management of building infrastructure itself, which is actually the biggest bottleneck to big infrastructure projects like starting up a new mine. My whole family are civil engineers who build bridges, and the amount of planning and logistics and operations required to get a big infrastructure project up takes years and tens, hundreds of millions, sometimes billions of dollars. A lot of that can actually be automated through sophisticated and very complex flow charts of dependencies — mapping all the inputs that feed into a project, and showing how a change in one thing affects everything else. That software didn’t exist before; nobody had built it for physical infrastructure. So one part of the operations of building — where software can come in and understand all the nuances and dependencies of putting something up.

The other is the operations of the mine itself. And so if you’re moving things around, and you don’t have the drivers to drive the trucks, this company, Mariana Minerals, uses autonomous like, hauling vehicles to move things back and forth. They use autonomous extraction, autonomous drilling, and AI helps with the discovery of different deposits and where to drill. Throughout the entire value chain, there are little places where you can continuously find more margin, more yield, more improvements, which makes the economics of the whole operation much better.

CB: I saw you also invested in, I think it was a lithium processing company — or was it extraction?

MV: It’s the same company.

CB: Is that primarily happening in the US? The extraction?

MV: Yeah, all their other current projects are in the US. They’re refining lithium from wastewater in East Texas.

CB: Last month you closed your first fund. Was it $50 million?

MV: $52 million is the official number, but I’m saying $50 because it’s a little bit rounder, easier to understand. I technically closed it in December, but announced it last month.

CB: What goes into raising your first fund as a venture founder?

MV: A lot of meetings, way more meetings than you ever expect or want to have. A lot of “kissing frogs,” is sometimes what people will say. It’s like figuring out which LPs (investors in funds) actually want to invest in what you are doing, and that’s a long process. You get a few early believers, and then you can start investing some of that money, and that helps you build a track record to then go out and get more and more LPs. It’s like a fundraise for a company, just much longer. You need a lot more yeses in a fund than in a venture round.

CB: How do you make sure you’re in alignment with your LPs on your investment vision and capital allocation in general?

MV: The portfolio construction, the pitch thesis — it’s all me. You have to find LPs that want to believe in the same thing. I have found that it has to be driven by me. I cannot let LPs wag the dog, otherwise you lose what makes it special, which is having some kind of unique insight and access and ability to get into deals you think are going to be great. The point of the fund is that the fund manager makes those decisions. LPs might say “Oh, I wouldn’t have expected that to be interesting,” and that’s exactly why they’re betting on the fund.

For example, Anduril was a very controversial investment at the seed stage for many fund managers. Even at one of the funds that invested, the LPs were asking, “Why did you do that? You shouldn’t invest in defense. That’s a terrible category,” and that fund manager had say, “I believe in this company.” Now, of course, everyone says they knew all along it was going to be great. But the magic of venture is you see something before the market sees it, and you see something before other people do. You want LPs to understand that. That’s why they’re betting on you.

CB: What are the most exciting opportunities in energy and nuclear at the moment?

MV: I think we’re finally going to see the nuclear renaissance come to life this year. July 4 is a big milestone date for nuclear. The Trump administration signed an executive order trying to have at least three new nuclear reactors go critical by July 4 — America’s 250th anniversary. And there’s actually four in contention right now. Because of this push for nuclear, it’s also opened up this whole supply chain for the nuclear market. There’s a couple of new uranium enrichment companies. There’s General Matter, there’s Standard Nuclear. New companies are being formed around uranium mining and discovery, finding new pockets of uranium. This whole ecosystem is growing.

Nuclear fission is not new technology. We’ve been doing it since the ‘60s, and it’s safer than the general public would understand — even safer now because of advancements in the way fuel gets processed. There’s this thing called TRISO, which is extremely safe fuel. It’s kind of expensive, but it’s meltdown proof. It’s like little nuclear beams that go into the reactor. It’s baseload, so it’s clean, continuous energy — just turn it on. It works. 10% of the country right now is powered by nuclear, and I hope that number goes up significantly as we turn on more reactors.

In general, we’re seeing a wave of new energy innovation beyond nuclear. Solar is still doing great. Battery development has gotten way better in terms of battery density and the power you can get out of them, and the cost is now at a point where you can almost have baseload energy with just solar and batteries. Data centers, obviously, are pulling the energy market forward in many ways. We’re due for an upgrade to all of our energy systems, so I’m very excited about that category, too.

CB: What do you think are the opportunities in space?

MV: The main markets in space are either defense-related or communications-related. Obviously the government is spending a lot of money in space. I think the new Space Force budget just came out — it was around a 70% increase from last year, or something high. Don’t quote me on that number, but it’s something very big. As we get more and more satellites in space, new ecosystems will form around how to manage all these satellites, how to move them around. Starlink is the major example on the comms side. But many other companies are putting up their own constellations. I think we’ll see more infrastructure-layer companies come out to manage this whole space networking problem. Ground stations will need to be modernized to help manage all the information coming down from space. I think it would be good to have more than one launch provider. There’s a couple companies that are very close to being able to do more regular launches, but obviously SpaceX is pretty dominant.

CB: Having launched Pax Ventures, what are you looking to emulate from Andreessen Horowitz, and where are you looking to take a more novel approach?

MV: Andreessen Horowitz has a lot of platform teams and can provide a lot of horsepower for companies. Those platform teams can sometimes be too much horsepower, or not timed right for early-stage companies, which need a different set of things. Sometimes the companies I meet are figuring out who to use as their law firm. They’re figuring out how to set up their internal HR systems. They might need a deck designer ahead of their first institutional round. They need their first real hires. They’re hiring advisors. I love that stage, and want to have the right ecosystem around me to support founders dealing with zero-to-one problems. From there, I want to help them raise those next rounds of capital from the multi-stage funds, who will then pile on the resources that are more stage-appropriate for companies as they’re scaling. I want to be very helpful — still be very hands-on and high-touch — but really focus on the needs of companies at the early stages, which are always different and unique.

This interview has been lightly edited for length and clarity

Cathedral is a technical term, though. It specifically refers to a Catholic church which is the seat of a bishop and the central church in a diocese. Taken as a group, however, such churches correspond only loosely to the popular idea of cathedral-ness. Interesting large buildings, in a more general sense, are what most people associate with the noun ‘cathedral,’ and are a better yardstick for human achievement anyways. I think this is especially true in light of what I like to call the conservation law of religiosity: If you look at a society and see no obvious organized religion, it hasn’t evaporated into the ether, it’s merely gone subterranean. The corollary to this is that no tall building has ever stood without having some sort of belief system embedded into its structure.

In the United States, skyscrapers have supplanted cathedrals as the centers of our cities. Our modern behemoths of steel, concrete, and glass — not cathedrals — are viewed as symbols of American technological progress. That does not have to be the case. There is an even more peculiar large building than the Central Park Tower, a modern skyscraper par excellence, constructed in the alpine foothills of the Wyoming section of the Rocky Mountains by cloistered Carmelite monks. This peculiar building is made of multi-ton slabs of stone, carved into Gothic forms by an arsenal of machine tools whose movements are dictated by computer programs. It is not strictly a cathedral, but much like them, it strikes one as a sparkling anachronism.

The New Mount Carmel Monastery sits roughly three hours away from Yellowstone National Park in Northwest Wyoming. Surrounded by near-pristine forest and the Rocky Mountains, the brothers at the monastery can be found variously tending to cattle on horseback, praying in quiet contemplation near streams, or programming toolpaths for their robotic CNC arms. Despite the infusion of neoindustrial cowboy aesthetics, their lives are structured in a way that would be recognizable to the founding hermits of the Carmelite Order. That Order traces its origins to hermit communities on Mount Carmel (in present-day Israel) in the 12th century, and officially took on a rule under Albert of Jerusalem in the early 13th century. The Wyoming branch of the Carmelites was founded in 2003 by the then-Bishop of Wyoming, David L. Ricken, as the “Monks of the Most Blessed Virgin Mary of Mount Carmel,” in a setting which naturally mirrors the solitude-at-elevation of their spiritual origins.

They celebrate the Carmelite Mass in Latin with Gregorian chants, maintain a strict horarium — schedule — replete with ora et labora, and cite the importance of manly virtue as expounded by Saint Teresa of Ávila. The John Wayne references alongside Saint Teresa might be a bit of a surprise to the founding hermits of their order if they could read them, though. To quote their writing on manliness and manual labor, “8 hours of work, 8 hours of prayer and 8 hours of sleep provide a perfectly balanced lifestyle for monks to attain holiness...Men need a challenge. John Wayne explained it best when he said, ‘I define manhood simply: men should be tough, fair, and courageous, never petty, never looking for a fight, but never backing down from one either.’” For the monks in Wyoming, those eight hours of manual labor are split between a variety of tasks, of which construction is only one part. For instance, they also roast and sell coffee to help fund their construction projects and the normal costs of operating the monastery. The emphasis on balance in their way of life means that they are patient with respect to overall timelines. However, the technological efficiency of their construction methods has already enabled them to finish a substantial amount of their master plan, and they think their crown jewel building — the chapel — could be completed by as early as 2030.

It’s all too easy in a hyper-modern world, especially in hyper-modern cities like San Francisco or Los Angeles or New York, to look at friars garbed in brown habits and write them off as antiquated or myopic. A more lucid picture is to regard monastic orders as a stable font of civilizational progress that have existed for millennia and will likely exist for millennia more. Roger Bacon, a friar in the Franciscan order, gave us much of the scientific method as we know it today; Gregor Mendel, part of the Order of Saint Augustine, gave us the field of genetics; Guido of Arezzo, a Benedictine monk, gave us the backbone of Western music. They have quite the track record.

Calling the brothers of New Mount Carmel anachronistic is a naive misunderstanding. Under any historical perspective, their engagement with and advancement of techne makes vastly more sense than, say, semiconductor manufacturing springing up from the soil of Bay Area orchards. The Carmelites in alpine Wyoming are the inheritors of a deep tradition that excels at maintaining its integrity while effectively adapting to local flavor. Their meditative yet pragmatic approach is precisely what one would expect from the serene, harsh environment of the Rocky Mountains. On the other end of the spectrum, the technology industries of our major cities excel at rapidly generating novelty ex nihilo, and by virtue of creative destruction cannot meaningfully participate in a deep tradition.

Similar, if you squint, to the Carmelite monks is Monumental Labs, a New York-based company which operates CNC stone-carving robots as a service. In contrast to Monumental, our Carmelite monks do not have to compete for architectural restoration commissions or vanity projects of the techno-riche. Instead of raising venture dollars, they rely on donations, tithes, and the freely-given labor of competent young men. The Carmelites know their aim, and merely choose CNC stone carving and 3D modeling as the most effective way to achieve that aim in our era. Despite completely lacking the requisite esoteric knowledge of machine tools and stone when they began, they found their way by devouring old reference books and learning what they could from projects like the restoration of Notre Dame after its disastrous fire in 2019. The state-of-the-art techniques were merely seen as instrumental to attaining their overarching goal of praising Jesus Christ.

The project of the New Mount Carmel Monastery is not an act of nostalgia. Gothic is not a style so much as metaphysic in stone. Each rib, arch, and pointed vault is like some frozen gesture of reaching out toward the divine. That the Carmelite monks have resurrected such forms with precision robotic tools suggests that technology, rightly ordered, can be drawn back into the service of transcendence. The same KUKA robotic arm which might otherwise be used to assemble the next generation of Meta products is, in their calloused hands, carving echoes of Heaven. The key difference is teleology. The monks pray to the God of Abraham, while Meta prays to the god of Nick Land’s technocapital singularity.

Seen this way, the monks’ project exposes the emptiness of modern architectural ambition. Despite their scale, skyscrapers like the Salesforce Tower say very little beyond themselves. In contrast, the monastery proclaims, “I am not a building to be optimized or monetized, I am one to be prayed and labored into existence.” Hundreds of thousands gaze at Salesforce Tower on the skyline of San Francisco every day and feel nothing. A single person could walk into New Mount Carmel in Cody, Wyoming and feel the weight of the cosmos pressing on their chest.

I’ve heard from my own friends an oft-quoted factoid that the knowledge of how to construct Gothic ribbed vaults has been lost. This is characteristic of the self-defeating disdain perpetuated by the so-called architecture schools that Tom Wolfe lambasted half a century ago in From Bauhaus to Our House — and our civilizational reluctance to build things that step outside the domain of pure utility. The existence proof to the contrary of that factoid is in Wyoming, it’s hundreds of feet high, and it’s made of American limestone.

The United States is not the only major government to recognize the Moon’s economic potential. Russia and China have agreed to jointly build a research station on the Moon’s south pole, where 3He has a “comparative abundance.” They seem to view a presence on the Moon as a type of first-mover advantage, securing access not only to the isotope but also to water ice (critical for on-site fuel production) and even rare minerals.

A contest over lunar resources could drive the first permanent human presence outside of near-Earth orbit. As the engineer and author Robert Zubrin argues, Earth’s closest satellite will be key to any strategy involving further bodies like Mars. The White House even calls for the “initial elements” of a lunar “outpost” by 2030. Space colonization has transformed from science fiction into a strategic imperative.

However, before rockets reach our nearest neighbor to initiate lunar mining or settlement, they will have to escape the gravity well of international law. Neither capital nor pioneers will flow to a Moon ‘owned by everyone’ — with colonies legally murky, hard to defend, and ungoverned. There must be an extrinsic factor that accompanies American migration to the stars: public backing that provides legal stability, prudent defense, and responsible governance. These are the foundations of sovereignty. Without a way to exercise it in the no man’s land of space, there will not be an American Moon economy.

The Property Problem

The 1967 Outer Space Treaty banned sovereign territorial claims on the non-Earth celestial bodies (for all intents and purposes, the Moon and the planets in our Solar System). At a time when only a few nation states had the capacity to leave Earth — and two years before Apollo 11 astronauts set foot on the Moon — the treaty drafters never considered that private corporations might operate in orbit. In 2015, Congress codified the right of private actors to own and sell space resources. That seems to have motivated the startup Interlune, founded in 2020, but American law still stops short of grounding rights to extracted lunar materials in the formal right to mine them.

The capital-intensive business of mining requires a measure of exclusivity. On Earth, holders of real estate assign that exclusivity, guaranteed by the jurisdiction the property falls under. But property presupposes prior governmental ownership, either retained or transferred to private hands. Governments cannot claim celestial territory or own it. How, then, can they grant or secure exclusive use? Without the full protection and enforcement of property rights, private Moon development becomes more impractical than it already is.

After all, the basis for the bull case is a loophole in a Cold War-era treaty. China has shown its willingness to create territory in supposedly non-claimable areas. Should China attempt to similarly encroach upon American Moon installations, the United States may not have a clear international legal basis to respond in kind. The lack of sovereignty in space complicates workers’ rights too — it is unlikely the highest-value human capital will still commit itself to the stars when its freedoms and protections are an open question. Any private settlement or mining endeavor must underwrite billion-dollar expenditures and reputational and moral liability just to make a productive Moon possible. The status quo asks them to do that amid geopolitical tensions, in a hostile physical environment, and with an ambiguous legal justification 239,000 miles from Earth.

Optimists might argue that pioneering companies are able to pan for gold in the wilderness of uncertainty, that success is only a matter of will. Nonetheless, when it comes to international waters, American deep-sea mining operators have floundered. In the case of the oceans, international law has not necessarily bound the country (the Senate never ratified the United Nations Convention on the Law of the Sea). The government has even created a licensing regime for American companies to engage in commercial recovery. Yet, that regime sits barely used: operators struggle to raise capital. Many of the same dynamics that define the space frontier apply here: geopolitical competition, technical challenges, and enormous risk. The potential of the deep seas has not been enough to generate economic activity in spite of those barriers. In the even more complex realm of space, then, loopholes — like licensing in contravention to international law — will not be enough.

Dreams of American space colonies require sovereignty to exist where international law forbids formal state ownership. Call it quasi- or de facto sovereignty: in the context of space, a mechanism that allows the government to coordinate and constrain private sector activities, defend national interests, and enforce public welfare where de jure sovereignty is proscribed. History might provide the model.

In the Age of Exploration, international law did not exist to restrict far-flung land grabs, but state balance sheets did. Commercial actors could step in where the government could not afford the endeavor at hand. Accordingly, European monarchs chartered corporations to “pursue long-distance commerce and conquest without direct government finance or control.” Where third parties engineered, administered, and profited from far-flung supply chains, the government could focus on strategic interests: tax revenues, geopolitics, national influence.

There were excesses in the quasi-sovereignty exercised by charter-holders, of course, such as the British East India Company’s private military operations. However, chartered companies allowed the notion of sovereignty to function at scale and across oceans, and they brought order to the settlement of what Europeans understood as terra nullius (‘nobody’s land’). In North America, the British Crown granted charters to “specif[y] the land that an individual or corporation had the right to settle.” Famously, the Massachusetts Bay Company charter stipulated local lawmaking authority. Such colonies were the infancy of self-rule in America: corporate charters grew into the actual sovereignty of the state and federal governments.

Of course, the charter colony is not directly transferable to space. A modern charter cannot grant territory that the state does not own. However, the precedent transfers: public-private partnership to establish settlements at scale and at distance, when the government alone cannot. The charters combined private profit with public legitimacy, and private space colonization requires some manner of governmental backing to succeed. Like Massachusetts Bay received from the king, American lunar settlements need legal standing, defensive backing, and a governance structure to be viable. Public equity stakes could provide them.

The Space Settlement Corporation

The “Space Settlement Corporation” (SSC) could fill the legal void in space with de facto sovereignty. Congress would endow the government corporation with the funds and standing authority to buy preferred shares in any corporation seeking to mine or settle the Moon. This preferred stock would guarantee board seats, reflecting the SSC’s status as proxy for the American people. Statute would ban majority stakes, and appropriations would prevent overweighted public ownership. In effect, the American people would become a stakeholder in and beneficiary of any lunar mining or settlement project.

If adversaries threatened the public’s bottom line, the United States would have clear justification to defend burgeoning colonies. Board representation would enforce the public interest in labor rights, responsible extraction, and relations with foreign-aligned facilities. Further, the government would assume some of the inherent risk in space mining or settlement. The SSC would coordinate rival American operations, preempting competition and introducing exclusivity. This vote of confidence in corporate space pioneers would unlock greater private investment and accelerate Moon development.

The SSC would not just vote by proxy, however. If established to clarify the chaos of space settlement, it must be an active governor of American activities in space. As a condition of its backing, the SSC would require veto authority over actions that compromise the national interest or colonists’ welfare and the power to remove corporate officers who disregard the same. Its ultimate teeth would be removing the legitimacy it provides. If a mining enterprise 239,000 miles from Earth decided to flout federal law, the SSC could liquidate its stake in the operator. ‘Rogue’ colonies could no longer expect governmental support against rival claimants, defense from foreign interference, or mediation with their workforce. No operation could survive long in such a Wild West.

It is not unprecedented for the government to demand equity when an industry is vulnerable. When General Motors went bankrupt in the late 2000s, the Obama Administration converted over $50 billion in federal loans to a majority stake in the “New GM.” Despite “hands-off” pledges, the Administration reserved the right to “set up-front conditions to protect taxpayers, promote financial stability, and encourage growth.” Whereas the “Old GM” died in a hospital bed, the nascent Moon economy may suffocate in its cradle if not nurtured by industrial policy. As America’s deep-sea mining struggles show, the free market stalls when asked to singlehandedly develop a frontier. When it stalls on the lunar frontier, American space superiority suffers. The SSC is a moderate response: government intervention without nationalization.

This strategy would follow the trail blazed by the second Trump Administration. In June 2025, the government finalized approval of the partnership between U.S. Steel and Nippon Steel. As part of the agreement the companies entered into with the government, the public retained a “golden share,” providing “special say in how the partnership is run.” The strategic and economic importance of maintaining domestic steel production justified the involvement. Likewise, gilded stakes are necessary in space colonies. Emerging economic competition on the Moon could soon veer into military competition. Government involvement is crucial to preventing and mitigating that possibility. Moreover, partial public ownership would recognize the people’s already vested interest in colonization while securing the majority of upside for the private sector. Taxpayers alone underwrote the early exploration of the cosmos. Should it become a back-pocket sovereign wealth fund, the SSC would deliver a return on that investment.

The expansion of the United States was often a victory of private individuals who sought out new opportunities. Before Texas became an independent republic or the nation’s second-largest state, it was a sparsely settled expanse where Americans laid the agricultural foundations of a powerhouse. The Union is stronger today because the private sector stepped in where the public sector alone could not. An English-speaking Moon might make the Union stronger still.

The Space Settlement Corporation would lay the groundwork for the future integration of the Moon into America proper. Future reform or replacement of the Outer Space Treaty may establish official methods to claim territorial sovereignty. In that scenario, the United States would need a viable footprint on lunar soil. Current law creates deadly ambiguity for dreams of space settlement. Public-private partnership would clarify and order that murkiness until the treaty regime can finally catch up to space realities.

The same dynamics are playing out in Antarctica, where treaties ban territorial claims and the Madrid Protocol restricts minerals development until 2048. But China has been building a growing network of research stations on the continent. Even if treaties do not allow true sovereignty in Antarctica, the People’s Republic recognizes that “presence equals power.” If international law were to change, settlements would provide leverage for territorial and resource claims.

The age of astropolitical competition has arrived. Helium-3 may solve the most important technological problems of our time, and a stable supply chain of lunar resources may grant America another century of terrestrial and extraterrestrial dominance. For our spice to flow, perhaps the government needs to take a stake.

Matthew Kinsella is the Chief Executive Officer of Infleqtion, a quantum technology company based in Louisville, Colorado, that produces neutral-atom quantum computers, clocks, and sensors. Kinsella first backed Infleqtion as its lead investor in 2018, while serving as a managing director at Maverick Ventures. The company raised several rounds of capital from Boka Capital, Morgan Stanley, and In-Q-Tel, among others, and has formed partnerships with organizations including Safran, NASA, and Nvidia. After nearly two decades in venture capital, Kinsella joined Infleqtion full-time as CEO in 2024. In February 2026, he took the company public on the NYSE under the ticker INFQ at an implied valuation of $2.2 billion. I sat down with Matthew to discuss the implications of this new technology, quantum’s relationship with traditional computing, and its impact on geopolitical competition between the United States and China. What follows is a transcript of our conversation.

CB: What is quantum computing and why does it represent a massive paradigm shift?

MK: One important distinction to make: I tend to think about quantum more broadly than just computing. There are huge paradigm shifts that are taking place across multiple different technologies based upon our ability to harness the quantum properties of atoms. Computing is one of those. But there's a whole other swath of products: time keeping devices, clocks, antennas, inertial sensing equipment. These are equally as impacted by quantum as computing.

If you think about the information technology revolution, even going back to the 1870s when we figured out how to harness electricity and RF waves, you're using physics, but you're using it in a bulk sense. You're sending trillions of electrons through a conductive wire for electricity, right? And then it basically all comes from there. RF antennas are the ability to interpret what those electrons are doing based upon the vibration of an antenna. Communications are embedded signals inside of those electrons. And then finally, computing is sending electrons through bits — zeros or ones. That's the basis for all of binary computing.

With quantum, we're able to harness individual atoms and unlock their power. It's going from bulk physics to individual physics. In harnessing the power of the atom, we can start taking advantage of its quantum properties and turning those into useful products. Some of those properties are the energy transitions of atoms. An atom looks like a solar system, you've got electrons orbiting it. What does quantum mean? It's that quantum jump from one energy field to the other. The energy transition of an atom — of its electrons — is the most stable frequency reference nature has to offer. So you can build clocks that are many, many orders of magnitude more precise than traditional clocks. Because what is a clock? It's a frequency reference, right? That's gone from a pendulum swinging to the vibrations of a quartz crystal, to now the energy transition of an atom. You're just trying to find something that's stable, right? When you take that energy transition of an atom and control it with lasers and lock the laser frequency to that energy transition, you now have a very fast ticking clock that ticks super precisely and never drifts. So we can build very, very precise clocks. That's the least complex thing you can do with quantum.

You can also then turn those same atoms into antennas. The way classical communications — radio, wireless communication — work is you embed a signal in a radio wave. Then you use an antenna that vibrates to the frequency that radio wave is set to, and you extract the vibrations. You extract the electronic signal based upon the vibrations of that antenna. That's worked amazingly well for 120 years. The issue is that the antenna needs to be about the same size as the wavelength you're receiving. Otherwise, it won't receive the wavelength. It's resonant to that wavelength. We can turn these atoms and the electrons of the atoms into antennas. That fundamentally breaks that correlation. We can receive very low frequency, long wavelength signals that would normally require an antenna the size of a football field to receive, to something the size of a sugar cube. It’s absolutely game changing technology in the RF world. You can also build inertial sensors with the ability to navigate the world with such extreme precision, you would never even need to access the GPS network. GPS is increasingly prone to being spoofed or denied. There’s all sorts of GPS jamming and denial relating to Ukraine, and you lose the ability to navigate. If you can recreate that service at the precision levels of quantum, you can wean off of GPS. So that's the broad, quantum sensing umbrella that's enabled by unlocking the power of the atom.

CB: Why is it useful for computing?

MK: It goes back to that whole binary concept of computing. Even the most powerful GPU cluster on the planet — it's still boiling all problems down into binary logic. And you can do unbelievable things. Who would have ever thought of all the amazing things you can do by boiling everything down to a trillion zeros or ones. But at the end of the day, that's not how nature works. It's a heuristic for how the world works. Nature works in quantum, so if you're trying to solve the problems of nature, you can't really use a heuristic. That's why you hear things like, “in order to solve this problem, you'd need a computer the size of Jupiter.” The reason is that quantum mechanics are inherently uncertain. It's all based on probabilities. If you're trying to do things like model the electron interactions of two molecules that are combining, that's a quantum mechanical process. That's the type of thing that any classical computer would break on. Those are the types of problems that quantum computers can solve quite easily because they are quantum mechanical in nature. They utilize atoms that are in what is called superposition, which means they can simulate all potential combinations at once. And then you can now start to unlock those types of problems of nature — like drug discovery, materials science, etc. —  that we've never really been able to point compute at in the past.

One misnomer people have is that quantum computers are going to come in and replace classical computers. That's not what's going to happen. They're never going to do the things classical computers do. Rather, they're going to open up a whole new set of problems working alongside classical computers. This is the heart of Infleqtion’s partnership with Nvidia. A lot of it will be solved on the classical computers, but the really sticky quantum parts will be kicked off with the quantum computers.

CB: Does this intersect with Silicon Photonics?

MK: It does, in that everything we do is based on lasers. In order to harness the quantum mechanical properties of these atoms, you have to hit them with lasers. So how do we build our clocks? We shoot a 778-nanometer laser at a rubidium atom, and that's the frequency that excites the atom and creates that stable frequency reference. We make our antennas by hitting them with a different type of laser that puts them in what's called their Rydberg state. And if you think about the electrons orbiting the atom, it's as far out in orbit as it can get before it goes away and becomes an ion. Think of a really big atom, and that makes it sensitive to that RF spectrum. And then the way we build our computers is we trap clouds of rubidium or cesium atoms inside an ultra-high vacuum cell, and then we individually address each atom with a laser. We hold those atoms in place. Each one of those atoms becomes a qubit, then we put them into superposition and then entangle them, and those are the basic building blocks of our computer. The point is, everything is based upon lasers, and as we can integrate those lasers into Silicon Photonics, we can bring the cost down and the performance of these systems. Instead of having lasers, you're actually working with photonics on silicon, which are basically just lasers that are built on silicon.

CB: What's going to change in the wider world due to this technology?

MK: For the sensing products that I mentioned, it’s more of an upgrade cycle from classical technologies. So not necessarily enabling new applications but making us resilient against the loss of GPS is an absolutely massive implication of quantum sensing. Because what is GPS? It's basically 30 satellites orbiting the Earth in geospatial orbit. And these satellites have clocks in them, and they send a very weak signal down to Earth of what time it is. It's like a nested doll situation. These clocks all synchronize with these room-size clocks on Earth's surface. You can think of it as these nodes on Earth's surface sending their best guess at time up to these less precise clocks  in orbit, which then send time down to everything, ranging from your iPhone to the nodes in the electricity grid. The key service from GPS is time, and we use that time not just to figure out how to navigate the world, but to synchronize all of our critical infrastructure: the RF networks, the electricity grid, the financial markets, the air traffic control system. If these things didn't have access to GPS to figure out what time it is, they would not be able to synchronize. And if they can't synchronize, they can't work.

The implication that's most near term is resilience against the loss of GPS. With quantum you can make clocks that are more precise than those room-size clocks I was talking about, in a small form factor, and put them locally. We don't need to rely on GPS anymore for our critical infrastructure to work, or to get the timing signal to make sure your phone works when you're out in the field. So that's one very near-term implication to it. And there's all sorts of use cases in the defense world for the other quantum products. The RF sensing and communication that I was mentioning to you is absolutely game changing for a number of reasons. For example, submarines have to dangle a one-kilometer antenna off the back because the only wavelengths that will penetrate salt water are these really long wavelengths. We can shrink that down to something very small. If you're trying to communicate over large distances, you need to use those low frequency, long wavelength signals. To receive those in the field, you have to put up a huge antenna that makes you a target. We can do that with something the size of a sugar cube. And importantly, because they're not electronic in nature, they emit nothing. The average lifespan of a radar system in Ukraine is something like eight seconds. You turn it on, it's detected, it gets blown up. We turn our RF systems on — not detectable, not blown up. They continue to operate. So huge implications for national security.

Quantum computing opens up an immense range of new possibilities for what we can throw compute at. The ones I'm most excited about in the near term are in the material science world. It sounds very niche, but anything that's made of physical materials can be rethought. Imagine if the battery in your phone would last a year versus a day. It's very challenging to do the electronic and molecular modeling to recombine these photovoltaic materials to make better batteries. Those are the types of things you can do. Write out what you want, and a quantum computer will give you the recipe when they're powerful enough. With SpaceX going public, what are they spending all their time trying to figure out? It's all materials science issues. Do we have better jet fuels? What do we build Starship out of that can withstand multiple reentries? They don't have the materials right now that are lightweight but robust. Discovering new materials and utilizing them — these are things that quantum will enable.

CB: China has made it very clear that quantum is part of its technological ambitions. And apart from the cooperation on nuclear submarines which it primarily stipulates, AUKUS has a designated pillar where quantum plays a major role. What's the geopolitical conversation surrounding quantum?

MK: Well, it's very much a race, and it's divided between China and the US and allies like the UK and Australia. The UK has actually taken a leadership position in quantum. They've been really pushing hard from a national security perspective, going back to 2013. In many ways, the UK is a bit ahead of the US in its quantum journey. The UK recently announced a £2 billion investment program for quantum, called the Quantum Missions. And then at the end of it is something that they're calling ProQure — basically a billion pounds of funding, £500 million of which is for quantum computing over the next four years, £250 million of which is for quantum sensing, £150 million of which is for quantum networking, and then another £100 million for tangential stuff. And at the end is another billion pounds that's set aside to procure actual quantum systems after 2030. The US has yet to announce anything of that level. But the US has deemed quantum one of the six technologies that the US can't and will not lose from a national security perspective. They call it quantum battlefield information dominance. Because of those precision levels I was mentioning to you, the quantum sensing is probably just as important as computing, because if you can have a better grasp of position, of navigation, of timing on the battlefield, especially in absence of GPS, which is almost certainly the position we'll find ourselves during a hot war, that side will have an enormous advantage. And so that's what a lot of this push for quantum comes down to. It's enabling better precision without GPS.

CB: Where is China? Is it possible to know?

MK: They're taking a very different approach than the US. The US is betting largely on its own capital markets to win. China's taking a very top down, directed approach, allocating funds from the nation's balance sheet to quantum research and commercialization, and ultimately putting quantum sensors into the field. So where are they? It's really hard to tell. The US and our allies have a pretty open process of releasing papers when new breakthroughs happen. And what you see from China is, once we release something, a week later, they'll release something similar. So it's unclear whether they're slightly behind us or ahead. I will bet on US innovation any day of the week. I do believe we are ahead in quantum computing, and China is definitively ahead in quantum networking. They've been able to send a signal from the ground to a satellite, back to the ground, in an entirely quantum-entangled way, which makes it completely unhackable. But the US and our allies are ahead on quantum computing, I believe, and there are other areas of quantum sensing where I'm pretty sure we're ahead.

CB: What are the implications of quantum for encryption?

MK: Encryption is critical to all computing and all communication. If I were just to send a signal out into the world and try to call you, if you couldn't prove that you were Carson with your digital signature, anybody could answer.

CB: Anyone could dial in on this call right now.

MK: That's kind of the fundamental need for encryption. It's just to make sure you can trust the party who's on the other side of whatever transaction you're doing. And that can range from a bank transfer, to a zoom call, to anything. Now, the encryption standards that we've been using for decades are what are called RSA-256, or SHA-256. What it is, basically, is a large string of numbers — 256 or more bits of data — where you have two keys, a public and a private key, right? You probably heard those terms, and the public and the private key are actually the two prime numbers that when you multiply them together, you get this very long string of numbers. Multiplying those numbers together is very easy for a classical computer to perform. So once you have both the public and the private key, you can get the encryption hash. But it's impossible for you to say to a computer, “Here is this long string of numbers. Give me the two prime numbers that when you multiply them together, you get this number.” The only way a computer can figure that out is by trying every single possible combination of every single number out there to see if they multiply together. Do I get this number multiplied together to get this number? So that's why you hear about these. It would take the life of a universe for the most powerful GPU cluster to break this type of encryption, because they just have to try every possible combination. 

It turns out that I'd mentioned these types of problems that are quantum mechanical in nature, where you need to simulate all sorts of different outcomes at one point in time, and superposition can enable you to do that. Well, this encryption standard isn't quantum mechanical in nature, but it's not dissimilar in that you're trying multiple chances to see if you get an outcome. You can basically simulate all those combinations at one time to get that outcome. And so you can take that life-of-the-universe-timeline for a classical computer to try to break encryption and shrink it to a week, a day, a month, depending on the power of the computer. Wer’re not there yet, but those are the types of things that quantum computers are going to be able to do, and that's why all modern day encryption is at risk. That's why everyone's talking about Q-Day. When will quantum computers be able to do this? And in the meantime, we need to get to what's called post-quantum encryption, different standards to encrypt data that are resilient against quantum computing as well.

CB: How far away are we from Q-Day?

MK: There were some interesting announcements from Google earlier this week, specifically related to Bitcoin. Bitcoin is based on a very similar type of encryption standard, and the number of qubits that we think we'll need keeps shrinking because the software and the error correction and the quality of these qubits keep getting better and better. It used to be like 20 million qubits would be needed. Now, Google just announced that they think they could do with 100,000, and we at Infleqtion have shown 1600 qubits already. And so the way I think about this is we'll have a kind of minimum viable usage of quantum computers, where they'll start to do things in that material science world that we talked about by the end of 2028. It’s not going to be like a ChatGPT moment, but it'll be like, “Oh, wow, we were able to do something that we couldn't do with a classical computer.” And then, you know, Q-Day keeps getting pulled in. But right now, I think it's probably 2030 to 2032, or something like that — you'll probably start to have quantum computers capable of breaking encryption. But don't hold me to that number. It's very uncertain, and it keeps moving around. But it's going to happen.

CB: What does that do to everyone's bank account, for example? Does that just mean that it's not safe?

MK: If we get quantum computers — the ability to break encryption — before we transition to a post quantum encryption world, the answer is yes. The good news is everyone can see it coming. NIST, the National Institute of Standards and Technology, which governs encryption standards, has said, “everybody get your act together.” They’ve put out post-quantum encryption standards to be adopted by 2030. People will migrate, but the basic way that humans behave is they wait until the last minute.

CB: Prior to becoming CEO of Infleqtion in 2024, you were actually a founding investor. What was your original thesis, from an investor’s standpoint?

MK: Back in 2018, it wasn't clear a quantum computer was ever going to work. What I found so intriguing about Infleqtion’s version of quantum was that it's a very flexible type of quantum technology that can be used for that whole swath of products that I mentioned to you before. I could see a world in which we were deploying those in the not-too-distant future. So instead of making this long-term call-option bet to seed a company that might build a quantum computer someday, we crafted a Nvidia-style monetization strategy, where we had this powerful neutral-atom core, and we started pointing it at some near-term markets to monetize and commercialize —  in this case, clocks and RF antennas and inertial sensors — selling those, generating gross profit dollars, and then funding the R&D to see if quantum computing was a thing. Nvidia pointed their GPUs at gaming, then at crypto mining, then at physics problems, and ultimately, the crown jewel of large language models came along. We're addressing thesensing market today while building quantum computers that will ultimately be our crown jewel, just like large language models. It's a very different strategy than what most quantum computing companies follow. But to me, it was the intuitive one, because we could earn a return along the way, and it wasn't just this binary yes or no. It was more of an island-hopping mentality to building a company.

CB: Could you walk me through your various funding rounds?

MK: Our Series C was a $100 million raise at a $700 million post-money valuation. Before that was our Series B, and then going all the way back to my initial investment, the seed, which was a $6 million raise on a $12.5 million post money valuation. It was a big journey from the $12 million valuation we started at to the $2.2 billion valuation of the IPO.

CB: Are you the first quantum company to go public?

MK: We are not. We are the fourth quantum company to go public. Three of them went public back in 2020: IonQ, Rigetti, and D-Wave. But we are the first neutral-atom company to go public. We’re the first to be focused on anything beyond computing. And to be clear, the reason I took the company public was largely for the financing. To me, the existential risk to our business — and most deep tech businesses that have actually proven the technology works — isn’t really technology or execution, it's financing. We were well capitalized after our Series C, which we closed at the beginning of 2025, but as I thought about scenarios in which Infleqtion failed five years down the line, most of the potential scenarios came down to running out of capital. Let’s say you get to 2028, it's time to raise another round of financing, and maybe the markets aren't accommodative to deep tech. I've seen it happen multiple times in my career, so I felt like the best time to remove that existential threat was to do it from a position of strength.

CB: How do you go about courting investment?

MK: Maverick led the first round, and then, similar to how most VC-backed companies work, the CEO and the CTO of the company at the time went out and raised the capital. As an investor and a board member, I played a big role in those raises. It was not easy. The seed, the Series A, the Series B and even the Series C were all hard-fought battles. We got great investors along the way, like Boka Capital, In-Q-Tel, which is the CIA's investment fund, the National Security Strategic Capital Fund out of the UK, which is basically the In-Q-Tel of the UK. When I joined full time, raising the Series C became a primary focus. It took a long time to come together, but we were able to get some great investors, like Morgan Stanley, Counterpoint, and Glynn Capital. Maverick has invested in all of our rounds. So how do you go about doing it? You just pound the pavement and tell your story to a lot of people, and eventually you'll find somebody who resonates with your story and wants to invest.

Hodak is one of a very small number of people seriously thinking about what is perhaps the largest unsolved problem in science: describing the nature, and the why, of subjective experience. Science traditionally seeks to describe the physical world — the world of atoms, protons, neutrons, and materials, and the conventional scientific method has proven enormously effective at doing so. But humans (at least, and possibly other animals) also have a rich subjective inner world, which we call “subjective experience,” or “consciousness.” When you close your eyes and picture a bright red apple, that apple “exists” somewhere (in that you could change your mind and picture an elephant instead), but it also does not “exist” in the physical world. So where is the apple, when you are imagining it? What is it made of? And why do humans (and almost certainly other animals too) have the capacity to have subjective experiences at all? Scientists have long dismissed the problem of consciousness as either impossible (i.e., too conceptually difficult to be worth trying to solve) or too “woo,” and not a serious research question. But as an eager engineer, Hodak is more optimistic than ever about understanding the physics of consciousness, and hence closer than ever to “engineering experience directly”: to reliably and precisely cause people to have particular experiences on command, like experiencing super-bliss or seeing certain colors without a traditional stimulus. His current vehicle, Science Corporation, a medical technologies company, is using cutting-edge neuroscience to cure diseases, enhance human experience, and create tools that explain consciousness. By building products that work directly with the brain, Hodak is chipping away at this problem of understanding subjective experience that scientists have long thought was intractable, or thousands of years away.

Hodak is of a school of thought that believes that even the most important scientific breakthroughs happen through trial and error, catalyzed by the pressure of market forces — like the rapid adoption of weight loss drug Ozempic, and the proliferation of artificially intelligent chatbots. Given that consciousness has long been thought of as an abstract problem, the general consensus was that the issue was going to be solved by academics, or philosophers, or theoretical physicists — people thinking in abstractions. But Hodak has reduced the problem of consciousness down to an electrical problem, applying electrical engineering to build biologically compatible products that integrate human experience with computers. And if Hodak succeeds in solving consciousness, then the Holy Grail of theoretical understanding will have been cracked not by a theorist, but by an engineer — a hacker, if you will.

Healing the Blind

“I’ve been interested in building brain-computer interfaces (BCIs) for essentially my entire life. Just watching The Matrix in fifth grade, paired with the realization that all of this” — Hodak gestures towards the room — “is brain activity. All of your vision, all of your perception is brain activity.” Born in Princeton, New Jersey, Hodak moved to Manhattan with his family when he was two years old; he had a fun childhood with encouraging parents, was active in sports, and went to boarding school. He describes watching The Matrix — a movie that depicts reality as being simulated — in middle school as a “very powerful experience.” The cinematic experience led him to believe in the possibility of a better world, easier to manage; if everything is simulated, then “the only thing that really matters is the brain — so if technology can act on the brain directly, then in a fundamental sense nothing else really matters.”

A biomedical engineer by training, Hodak states that his freshman year at Duke, he “talked his way” into the Nicolelis lab, “one of the best BCI labs in the world”; the lab typically did not allow undergraduates, but Hodak’s insistence earned him a spot. During college, and after graduating, Hodak founded companies involving data-driven performance modeling (MyFit) and cloud biology automation (Transcriptic). Hodak stepped down as CEO of Transcriptic in 2017. He co-founded Neuralink, a BCI company, with Elon Musk. Musk had mentioned to Sam Altman (who was then president of startup accelerator Y Combinator) that he was looking for someone “great” who would be “interested in the space.” Altman recommended his longtime acquaintance Hodak, who then became Neuralink president before leaving in early 2021. “It was the ultimate PhD — I learned a lot there,” Hodak tells me of his time at Neuralink.

Soon after, Hodak founded Science with a “massive” check from entrepreneur Jed McCaleb, as well as further financing from friends. Justin Kan, Science investor, co-founder of Twitch, and long time friend of Hodak, told me that “Max is a polymath, the smartest person I know. I invested in the company after the first ever presentation Max gave.”

Speaking about the limits of our “fragile earth bodies,” Hodak wants to use Science to dramatically transform the human condition through engineering. The first problem that Science has looked into as a testing ground for their theories about the brain and experience is blindness.

Science’s approach and mission-driven ethics have already yielded far better traction on the problem of blindness than any other research team that has tried before. Science hit its first major milestone in late 2025, when it restored functional central vision to clinical trial patients blinded by aging for the first time using a retinal implant called “PRIMA.” PRIMA — short for photovoltaic retina implant microarray — is a light-powered visual prosthesis with two components: an electrical implant and a pair of glasses. The implant, which is just two millimeters across (a quarter, for comparison, is 24 millimeters), is surgically placed underneath the retina, the innermost layer (and image sensor) of the eye. The implant itself is covered in pixels, which convert infrared light into signals that activate the retina. It is based on work originally conducted by Professor Daniel Palanker at Stanford University, a co-author of the paper. Once the implant is in place, patients can use glasses which are equipped with a camera facing outward and an infrared projector facing the eye, both of which communicate with the implant. The camera in the glasses captures visual information from the user’s environment; then, the infrared light projector activates select individual pixels on the retinal implant, which electrically stimulates the corresponding retinal region. This entire process encodes a visual signal for the eye to process, restoring sight for patients who use it. So far, 27 patients with age-related vision loss have been granted the ability to see again with PRIMA.

PRIMA is the first BCI to deliver meaningful vision restoration for blind patients. Per Science’s official press release, “It’s the first time that an attempt at vision restoration in these cases has achieved results — and in such a large number of patients.” Professor José-Alain Sahel, senior co-author of the PRIMA study published in The New England Journal of Medicine, described how some patients, previously unable to see, even progressed to reading entire pages of a book. This intervention demonstrates a key philosophy behind Science: that far from being magic, or inherently mysterious, subjective experience is created by physical matter in accordance with regular physical laws. The brain possesses an inherent ability to integrate and make sense of engineered signals; with the right tools, this ability can be leveraged to create experience anew.

A Corporation for Science

Hodak’s work at Science is a concerted effort to solve what he calls the binding problem: how distributed neural activity across space and time coheres into a single, unified moment of experience (i.e. a discrete instance of “your” consciousness, as distinct from “my” consciousness). The human brain has billions of neurons firing in different directions — some process color, others sound, others memory, and so on — yet we experience the world wholly, as a single unified experience with many sensory experiences “firing” at once, not as fragmented streams of data. While the binding problem is notoriously tricky to understand, let alone solve, its nature hints at good prospects for future neurotech. Because a single human “consciousness” is not localized to a single point, but emerges from many distributed streams of data, Hodak believes that a sufficiently integrated neurological device could become part of the same “bound” moment of awareness — creating new experiences that would be an indistinguishable part of “your” consciousness.

“Your brain is already two hemispheres connected by a big cable” — here Hodak is referencing the corpus callosum, a structure consisting of millions of nerve fibers that create a bridge between the two hemispheres of the brain and enable communication and coordination — but “you don’t experience two hemispheres, you experience one moment. Binding happens across that cable, and that cable is just electrical pulses.”

Hodak points to the case of twins Krista and Tatiana Hogan, who were born in 2006 with a shared skull, and who can reportedly see, feel, and taste the same experiences at once — the “stimulus” for each of their two experiences coming from the same sensory receptor (so that one sensory stimulus creates two “copies” of the same experience at once). The case of the Hogan twins, to Hodak, “really reduces [consciousness] to a classical electronic device,” and indicates that discrete conscious experiences (or “qualia,” in philosophical speak) might be something that can be transmitted through biological wiring, like how an electrical signal moves through a circuit.

Owing to his love for The Matrix and his explanations for consciousness, I ask Hodak if he thinks we live in a simulation, an electrically powered constructed reality. “I think we are probably not right now, but we should definitely build one. That’s a motivating source for me. When I think about ‘why build the Matrix?’ bits are just much easier than atoms. It’s really difficult to build anything [in the physical world]. If you have to actually put up a building versus building a game engine, one of these can be done in a couple hours, the other can be done in a couple of years, and cost many hundreds of thousands of dollars, and require enormous teams.” Hodak argues that resources in the physical universe are limited, whereas virtual worlds escape those limits and allow creative builders to create environments which favor defenders.

Science Corporation deals in both atoms and bits. Its Science Foundry, located in North Carolina, serves as the company’s dedicated fabrication hub. Acquired from MEMSCAP, an industry leader in micro-electrical-mechanical systems (or “MEMS,” which are tiny devices that integrate mechanical components, sensors, and electronics on a silicon chip) in late 2022, the Foundry was expanded in 2024 with plans for over fifty new local jobs and up to $65 million in investment. The Foundry produces custom chips, probes, and headstages which are commercially available, giving Science’s researchers and partners ready-to-use hardware of their choice for their own biotech experiments. This vertically integrated approach enables Science to scale its therapies like PRIMA faster than waiting for fragmented players to finish their products while also granting them significant influence in the neurotech ecosystem.

By owning both the hardware and the research ecosystem, Science seeks to accelerate radical therapies and platform innovations that would otherwise never be built. “Nobody has gone from an idea to a brain implant in less than a 100 million dollars. Venture math doesn’t work so it doesn’t get built,” states Hodak. “We are still in very early days; we are in season one [of neurotech].” Hodak views building the tools researchers need to fill in the blanks of neuroscience as a mission to advance our understanding of the brain. “To make season three products, there’s basic neuroscience that is missing and still not understood.”

Biohybrids

Another Hodak big-picture bet being made through Science is the Biohybrid: a group of genetically modified neurons derived from stem cells, placed on a scaffold that interfaces with the brain. Ultimately, the vision is to restore cognitive functions to people with brain damage from stroke or injury. Hodak dreams even bigger: augmenting human reasoning by adding new “patches of cortex… pre-trained to have some reasoning capability,” enhancing decision-making quality rather than speed.

The Biohybrid device is implanted neuron-side down; the system then creates connections to the brain by letting living neurons grow biological “wires” (axons and dendrites) between existing neural tissue and modified tissue. The goal of the Biohybrid is to create a way for BCIs to scale to millions of neural connections without the damage caused by traditional metal electrodes or threads. Early results are promising. In tests on mice, Science demonstrated that the device could use light to activate the implanted neurons, training the animals to choose the correct answer — such as moving left or right —to receive a reward.

This is where the project departs from a simple artificial intelligence-adjacent narrative that BCIs will make us humans more machine-like, and becomes, in Hodak’s words, “a longevity adjacent story.”

“One of the really big holy grails here is, the brain does two things: It’s intelligent and it’s conscious… The end of the artificial intelligence quest is to develop superintelligent machines.” But unlike pure AI research, which may succeed in re-creating advanced intelligence via silicon without simultaneously creating novel conscious experiences, BCIs aim to interface with the human mind in a way that preserves, and potentially expands, valuable subjective experience. As Hodak notes, it is “not clear to what degree LLMs are conscious or what their experience is like inside.” He invokes the AMC+ show Pantheon, which explores the concept of digitized intelligence that is uploaded to the cloud, to explain the risk: uploading or replacing human minds with machine analogues might optimize intelligence while quietly extinguishing consciousness — with no reliable way to verify what, if anything, is still being experienced.

The Biohybrid also represents another of Hodak’s guiding design principles: leaving the human brain alone. If a gene therapy delivered directly into the brain goes wrong, he explains, “those cells could die or you could get huge immune reactions,” potentially resulting in the irreversible loss of native neurons “in a lot of places.” The Biohybrid, by contrast, is designed so that if it ever fails, it fails gracefully: “With the Biohybrid graft cells, hopefully the worst thing that can happen is that the graft cells die, in which case the patient is in the same state they were before.”

Hodak is clear that this isn’t an ideological rejection of gene therapy — the Biohybrid itself consists of genetically modified light-sensitive proteins. The constraints are practical, not philosophical: immunogenicity, limited payload capacity, uneven brain coverage, and permanence. Once neurons are altered, “that’s like a once in a lifetime thing. The DNA has been inserted [and is] never coming out.” Ultimately, “not genetically modifying your original neuron seems like a good thing to be able to say.”

Looking Ahead

I now ask Hodak if Science will remain restricted to just neurotech and BCIs, “Science is a medical technology company. To the degree we are a neurotech company, it’s because you get huge effect sizes from engineering and dealing with the brain directly than you’re used to seeing with other medicine.” This is where Hodak’s entire philosophy, including his understanding of consciousness, fits together: a “neural engineering mindset, as opposed to a drug discovery mindset.” “It” — an engineering mindset as compared to more theoretical science — “just appears to be empirically better and more powerful.” He brings up Sarah, a patient in a successful case study at nearby UCSF who suffered from treatment-resistant depression for years. She received an experimental deep-brain simulation implant — electrodes that deliver electrical currents directly to the brain. Her affect improved instantly upon activation.

The pattern held for blindness. For decades, drugs and laser therapies have been tried to restore vision, but none have really worked. Gene therapies like Luxturna, a one-time treatment priced at $850,000, offer limited benefits. With Science’s PRIMA, however, “the average patient went from reading no letters on the eye chart, to reading from the fifth line, and the best patients could read every letter on the chart,” notes Hodak. The company hopes PRIMA will hit the European market this summer. I ask about US regulatory bodies and the rigorous timelines and standards they impose on medical device manufacturers. “The people at FDA care a lot and work hard. It’s a necessarily conservative body.”

Over the course of his career, Hodak has grown increasingly confident in his ability — and the world’s — to solve consciousness. This confidence rests on a convergence: clearer conceptual framing and the fact that technology capable of finally probing these questions will soon exist. “There’s capital, there’s talent, there’s stuff happening in the BCI industry that makes me think that it is really possible to build devices that allow me to give you answers to these questions of consciousness.”

I also ask about other diseases and ailments Science aims to alleviate next, and Hodak has a number of ideas. “Cochlear implants are great, but it’s not something we’re working on right now. We are very interested in stroke medicine, Parkinson’s disease. And through our ecosystem partners” — other companies using Science electronics and processes — “there’s a couple of those deals that I think we’ll see a longer list of medical applications being built in our technology through our partners beyond what we do internally.”

The day after our interview, Science announced Vessel, a perfusion system designed to keep organs, and thus patients, alive far longer than current technologies allow. Hodak was left “radicalized” after reading about the story of a teenage boy on ECMO (extracorporeal membrane oxygenation, a form of life support that temporarily acts as the heart and lungs by circulating blood outside the body to oxygenate it) who was awaiting a lung transplant. ECMO is not a long term solution, but rather “tethers patients to the ICU and requires constant manual adjustments to the system,” according to Hodak, in addition to carrying risks of infection. While on ECMO, the young boy developed a complication that made him ineligible for the transplant. This led to the boy’s medical team ultimately judging his recovery to be futile, and so they stopped changing the oxygenator filter on the ECMO. The boy passed away a week later. Such a scenario is a common ethical dilemma in modern medicine.

“The question shouldn’t be whether we can afford to keep someone alive; it should be whether the technology allows us to,” Hodak says. Vessel, another addition to Science’s growing longevity arsenal, is modular and configurable for different organs: namely the lungs, the heart, and the kidneys.

Hodak’s engineering approach has thus far proven successful, and his philosophy seems to be a winning one. He has set himself a quest larger than life, and on the way to his ultimate treasure, there appear to be many bounties to be shared with everyone. In PRIMA, Science has achieved what many have been trying to do for decades, and if Science is a testing ground for Hodak’s winning philosophy, then keep testing!

Most people think of space as the domain of pure science and exploration, a story that began with the 1960s Space Race to the moon. But from the start, space has been a military priority, given that it represents what military officials often call “the ultimate high ground” — the strategic advantage that comes from occupying the highest possible vantage point in a conflict. Today, space is a critical warfighting domain: essential space-based capabilities like GPS (Global Positioning System), satellite communications, and satellite-based earth observation underpin modern military power.

The Bond films’ space plots are campy, yes. But they blend fantasy with real insight, thanks to 007’s creator, Ian Fleming, a British intelligence officer in World War II with a personal fascination with space technology. Seven Bond movies feature space technology: Dr. No (1962), You Only Live Twice (1967), Diamonds Are Forever (1971), Moonraker (1979), GoldenEye (1995), Tomorrow Never Dies (1997), and Die Another Day (2002) — and, in each case, the technology depicted either exists today or is actively in development for space warfare.

Directed-Energy Weapons

A directed energy weapon uses focused electromagnetic energy to disable, damage, or destroy a target. There are two general types: lasers (higher-frequency visible, infrared, or violet waves) used to heat or burn through surfaces or blind sensors, and masers (lower-frequency microwaves) used to damage electronics.

Three Bond movies feature laser-based directed energy weapons:

Diamonds Are Forever (1971): Bond stops Blofeld’s plan to use diamonds in a space-based laser weapon capable of destroying ground targets anywhere on earth (The diamonds make the laser more powerful, obviously.)

Die Another Day (2002): Bond stops Gustav Graves’s space-based “Icarus” mirror satellite — billed as an agricultural tool that could provide year-round sunshine to crops, but actually designed to help North Korea invade its southern neighbor.

Moonraker (1979): The infamous battle at Hugo Drax’s private space station, featuring space marines and laser blasters

The Bond films’ space-based directed energy weapons vary in how closely they match real-world technology:

Anti-Satellite / Defense: Bond villains favor space-to-earth weapons, but today’s highest profile directed energy space weapons are ground-based. Both China and Russia have developed ground-to-space laser technology powerful enough to disrupt, degrade, or damage satellite sensors. U.S. Space Force officials expect space-based anti-satellite lasers to emerge in the coming years, given that space-to-space enables more precise targeting and sidesteps atmospheric distortion challenges from occur when firing the laser from the ground.

Space-Based Solar: Satellites that collect sunlight and beam it back to Earth — via mirrors or microwave/infrared lasers — are the closest real-world analog to Bond villain superweapons. But instead of using diamonds to concentrate energy for global destruction, early applications focus on power generation, agriculture, and defense.

Optical Communications: Space lasers are already critical to modern space systems, but they are generally used for communication — moving data — not combat. Just as fiber optic cables transmit data as pulses of light through glass or plastic strands, OCTs do the same in free space, using lasers to send information through the air or the vacuum of space instead of through a cable. This allows for very high bandwidth and lower end-to-end latency because you can route data more directly from satellite-to-satellite and because light travels faster in vacuum than in fiber.

OCTs can be used for several types of communication depending on where the data is being sent: Earth-to-space (from a ground station to a satellite), space-to-Earth (from a satellite back to the ground), and space-to-space (directly between satellites). Of these, space-to-space links are currently the most common, since they avoid atmospheric interference and offer major advantages in bandwidth, latency, and security. SpaceX’s Starlink network, for example, already utilizes a mesh network of roughly 24,000 OCTs (as of late 2025), allowing satellites to pass data to one another using lasers and deliver internet speeds and latency comparable to terrestrial networks, and in January 2026 the company filed with the FCC for up 1 million orbital datacenter satellites that would be connected via OCTs.

The Moonraker Gap: As gloriously absurd as the Space Marines battle in Moonraker is, we don’t have laser rifles in orbit just yet. But the possibility of armed human conflict in space grows more plausible as lunar and Martian settlement inches closer.

Nuclear Weapons

GoldenEye (1995) features a pair of satellites armed with nuclear devices. Detonated in orbit, the weapons generate an electromagnetic pulse that fries electronics in a precise target zone on Earth below. It’s a terrifying concept, and today, this technology would be even more destructive than the film depicts.

First, yes, we have detonated nukes in space. The 1967 Outer Space Treaty prohibits nuclear weapons or other weapons of mass destruction in orbit, but that treaty didn’t exist until after the U.S. and Soviet Union conducted a series of high-altitude and space-based nuclear tests between 1958 and 1962. Those Cold War-era nuclear tests demonstrated that a nuclear detonation in space creates a powerful EMP capable of damaging both terrestrial and space-based electronics, though with nowhere near the level of precision shown in GoldenEye, where the villains can program specific targets. In reality, U.S. and Soviet tests unintentionally fried electronics and electrical infrastructure near the detonation sites and as far as 930 miles away.

What GoldenEye didn’t explore was the aftermath. Radiation from a space-based nuclear bomb gets trapped in the earth’s magnetic field, creating an artificial radiation belt capable of destroying satellites en masse. One 1962 U.S. orbital nuclear test, Starfish Prime, knocked out a third of the roughly two dozen satellites in orbit at the time — largely due to the radiation field it created.

Now imagine the damage of a nuclear detonation in today’s orbital environment: more than 14,000 active satellites, plus around ten humans aboard the International Space Station and China’s Tiangong. The immediate EMP would destroy some portion of those satellites; the radiation belt could persist for months, destroying another portion. The resulting cloud of dead satellites would generate a debris field rendering portions of orbit unusable for years, a phenomenon known as Kessler Syndrome. And the astronauts overhead would face grave danger from both radiation and debris.

Given how much the U.S. military relies on space assets to shoot, move, communicate, navigate and track adversaries, it shouldn’t be surprising to hear that Russia is reportedly developing a space-based nuclear capability. The news broke in 2024 and caused quite the stir, as such a weapon would violate the Outer Space Treaty. It would also give Russia an asymmetric option to cripple U.S. military infrastructure by damaging GPS, communication, and observation satellites, degrading the Department of War’s ability to shoot, move, communicate, navigate and track adversaries. (Though, we can assume that some portion of the Department of War’s satellites are built with materials and components meant to withstand a nuclear blast.) But as Derek Tournear, former director of the Space Development Agency, put it, detonating a nuclear weapon in space wouldn’t just be an attack on the U.S. military but “an attack on the world” — a nuclear blast in space would have lasting effects on global travel, shipping, banking, communications, and financial markets, and more.

Electronic Warfare

The electromagnetic spectrum is vital for modern communications, radar, and navigation. Electronic warfare (EW) uses the spectrum to disrupt, deceive, or deny an adversary’s access to it — while protecting friendly forces’ ability to operate freely.

The Bond films feature the two main methods of EW: jamming, which uses high powered signals to overwhelm enemy receivers and render them useless; and spoofing, which transmits fake signals that mimic real ones.

Dr. No (1962): Jamming is central to the plot. The titular villain plans to disrupt a Project Mercury space launch from Cape Canaveral using high powered radio signals.

Tomorrow Never Dies (1997): Media mogul and Bond villain Elliot Carver uses a stolen GPS encoder to spoof the British frigate HMS Devonshire, sending it off-course into Chinese-controlled waters in the South China Sea. When the ship encounters two Chinese MiG fighter jets, Carver sinks the Devonshire and shoots down one of the jets, intending to spark a war between the UK and China.

EW isn’t as flashy as nukes or space lasers, but it’s a critical aspect of modern warfare — depicted by the Bond films with surprising accuracy.

Dr. No made jamming seem like an unavoidable threat. But in reality, jamming and the countermeasures to defeat it are routine aspects of modern conflicts. In Ukraine, EW plays out like a cat-and-mouse game, with Ukraine and Russia racing to degrade the other’s use of the spectrum while hardening their own systems. SpaceX’s Starlink has given Ukraine a robust, EW-resistant communications platform, with Ukrainian ingenuity and SpaceX’s technical updates working in tandem to counter Russia’s persistent electronic warfare efforts.

GPS jamming and spoofing has become so common that hundreds to thousands of flights are affected daily — particularly in Eastern Europe and the Middle East, where proximity to armed conflicts creates spillover effects. And unlike in Tomorrow Never Dies, where Carver needed a stolen military encoder, GPS spoofing today doesn’t require sophisticated technology. It can be done with cheap, off-the-shelf devices. The Department of War has been pushing a long-delayed GPS upgrade, and a handful of startups are working on alternative position and navigation systems. But for now, civilians and military alike remain vulnerable to GPS spoofing.

Rendezvous and Proximity Operations

Rendezvous and proximity operations (RPO) are the maneuvers spacecraft use to intentionally approach another object in space — whether for inspection, docking, servicing, or offensive action. Depending on their orbit, satellites travel between 7,000 and 17,000 miles per hour. If two airplanes flying near each other sounds dangerous, two satellites operating in close proximity is arguably worse. Even getting within a few dozen miles of another satellite is considered risky.

The Bond movies generally demonstrate hostile RPO

You Only Live Twice (1967): SPECTRE’s massive “Bird One” spacecraft hijacks US and Soviet spacecraft by swallowing them whole, Hungry Hippo-style.

Moonraker (1979): The U.S. Space Marines execute a hostile boarding of Hugo Drax’s space station

Compared to the Bond films, real-world RPO is still in its early days.

The most common example today is spacecraft approaching and docking with the International Space Station (or China’s space station, Tiangong). Unlike the docking scenes in the Bond movies, these are slow, careful processes that can take hours.

But we’re starting to see more aggressive maneuvering. In geostationary orbit, U.S. and Chinese satellites have reportedly engaged in “dogfighting,” jockeying for position to observe or evade one another. This is a relatively new phenomenon, driven by the militarization of space. It also demands new technology: most legacy satellites were built for static positioning and lack the fuel for extra maneuvering. The U.S. Space Force has made the ability to “maneuver without regret,” so that dynamic space operations (the ability to continuously and quickly maneuver in space over long periods of time and distance) aren’t executed at the expense of satellite lifespans. Orbital refueling, more efficient propulsion, and satellites purpose-built for RPO are all solutions under development by governments and their commercial partners

Space Domain Awareness

Space domain awareness (SDA) is the ability to maintain an accurate, real-time understanding of what’s happening in orbit in a way that supports real decisions. It combines tracking satellites, debris, and launches with higher-level analysis: spotting unusual behavior, identifying what an object is and what it can do, assessing intent, and issuing timely warnings so operators can protect assets, avoid collisions, and respond to threats.

In the Bond films, lack of SDA is a recurring theme: In both You Only Live Twice (1967) and Moonraker (1979), the government is initially clueless as to what is going on. Spacecraft are vanishing in You Only Live Twice; a space shuttle is hijacked in Moonraker—but the governments can’t see it, and they don’t know who is responsible.

Modern SDA is far more advanced, but still has room to improve.

The Bond films struggled with the basic question of “what’s where”. Today, U.S. Space Command (USSPACECOM) operates the Space Surveillance Network (SSN), a combination of ground-based radar, optical telescopes, and space-based sensors that maintains a public catalog of objects in Earth orbit — active satellites, dead satellites, space debris — down to about 10 centimeters. Commercial startups are building complementary systems with their own ground- and space-based sensors, offering faster and more detailed data, particularly in low Earth orbit.

But SDA is under strain. The number of active satellites in Earth orbit has grown tenfold in the past decade to more than 14,000. Earth orbit is vast, but it’s getting more crowded and more dynamic; the growth of satellite megaconstellations will only accelerate the trend. Starlink alone now has over 9,000 satellites in orbit. Over a six-month stretch in 2025, it conducted over 144,000 collision-avoidance maneuvers, a 200% increase from the prior six months. In January 2026, the company was approved by the FCC to deploy up to 19,400 satellites, and it also filed to deploy up to one million satellites for development on-orbit datacenter capabilities.

The space industry is in a transition period. Both commercial operators and the U.S. government are working to manage the increasingly complex “traffic control” that space now demands. USSPACECOM is partnering with the Department of Commerce to develop a Traffic Coordination System for Space (TraCSS), which will combine military tracking data with commercially procured data and analytics to manage commercial space collision risk.

Commercial Space Capabilities Exceeding Those of Nation-States

The Bond series accurately predicted that commercial space capabilities might one day exceed those of governments, with private companies controlling critical infrastructure that can decide the fate of ground wars and set the pace for space exploration. Elon Musk’s SpaceX is the most capable space organization in the world, period. It dominates launch, carrying 80 to 90 percent of all mass to orbit in 2025, including the majority of U.S. government satellites and astronauts. It dominates satellite manufacturing, producing several thousand satellites per year — an order of magnitude more than the rest of the world combined. It dominates the largest commercial market in the space industry, satellite communications, with Starlink generating the majority of SpaceX’s $15-16 billion revenue in 2025, serving over 9 million customers worldwide, far more than any competitor. And it dominates movement in space, with its Dragon capsule serving as the primary vehicle for both cargo and crew to and from the International Space Station.

The villain of Moonraker, billionaire Hugo Drax, runs a similarly powerful commercial space empire. Drax Industries is an aerospace contractor that supplies Space Shuttle-like vehicles for governments. Secretly, it has also developed a private launch site in Brazil and a private space station the size of a small city.

In reality, Musk and SpaceX are even more powerful than Drax and Drax Industries: Drax had his own rockets and space station, and supplied rockets to international governments. But SpaceX has more complete end-to-end space capabilities — and a deeper relationship with the U.S. government. The leading space powers — the U.S., China, and Russia — all have arguably more exquisite space technology than SpaceX and have accomplished numerous incredible feats SpaceX hasn’t. But it’s hard to say any single government organization matches SpaceX’s breadth or scale of capabilities. The U.S. government is now almost completely reliant on SpaceX for a variety of services: satellite launches (SpaceX wins the majority of national security satellite contracts these days), astronaut transportation (SpaceX’s Falcon 9 is the primary human-rated U.S. rocket to the ISS, and NASA has contracted SpaceX to develop a lunar lander for the 2027-28 return to the Moon), and satellite communications (Starshield, SpaceX’s government offering, provides speed and latency comparable to terrestrial fiber — significantly better than legacy government satellites).

SpaceX demonstrates that when a private entity applies first principles to manufacturing and vertical integration, it can outpace the collective bureaucracy of nation-states through sheer economic efficiency and innovation. Musk isn’t a Bond villain, but rather the ultimate industrialist who transformed space from a high-stakes prestige project into a global (and someday multiplanetary) commercial platform.

✺

From space lasers to nuclear EMPs, GPS spoofing, and the rise of commercial space superpowers, the Bond franchise’s “goofy” space plots has proven remarkably prescient. The weapons 007’s villains wielded are now the focus of actual military programs and international concern. We may not have laser-rifle battles in orbit yet, but the space battlefield Fleming imagined is very much becoming reality.

I spoke with Keri Findley about her rise through Wall Street post-2008, the intellectual foundations of her investment approach, and the risks building in today’s credit markets. What follows is a transcript of our conversation.

CB: What’s your relationship with mathematics?

KF: Love and hate. I was an operations research major at Columbia. I would say that simulation, Brownian motion, and applied integer programming were incredibly valuable to what I ended up doing. The class I was best at was called stochastic calculus.

CB: What’s that?

KF: It really is like a transition matrix for loan pools. Obviously, I didn’t know that at the time. One day I was sitting in research at Morgan Stanley. I was like 22 or 23, and it hit me. In stochastic calculus, the main example is like, if you start at First Street and First Avenue and 10th Street and 10th Avenue over here. How many different paths are there, and what are the weights on the path, and what percentage of the paths have you ending up on Fourth Street and Sixth Avenue? Those types of problems are literally a loan transition matrix. Every loan starts having made zero payments on day one, and they all either prepay or default. And it’s just a question of how they get there. I’ll never forget when I was like, oh my God, I’m in the job that perfectly aligns with my favorite class. And maybe I would have found that in many jobs, but I just found that very funny one day.

CB: What drew you to Wall Street? Were you looking for a challenge?

KF: No, I’m much lamer than that. I just wanted to stay in New York City. And I went to the Columbia career center, and Wall Street was recruiting, and I was relatively good at math. The other groups that were recruiting were a lot of corporate jobs at fashion companies, which I didn’t have any interest in, and a lot of consulting jobs. Consulting jobs usually had a minimum of a 3.7 or 3.8 GPA to apply. And I didn’t have that. I applied to every job that would take my application.

CB: You ended up in a sales role, right?

KF: I started in ABS CDO [asset-backed securities collateralized debt obligation] research at Morgan Stanley. At some point during my two year Analyst Program, the head of research left, and they moved me to sales, which I was shockingly bad at.

CB: Why do you think that was?

KF: Getting screamed at by clients and having to go back to them was just not something I was used to. When somebody screams at me and calls me names and tells me I’m the worst salesperson — I don’t think they meant it. But the clients would really beat you up. My boss would really beat me up. I didn’t handle being berated very well, and it would make me freeze and not jump into action. And I think I just had the wrong personality for it.

CB: How did you know to get out of that and what was your next move?

KF: Morgan Stanley told me, “You are not good at this.” I asked to go back to research. They didn’t have the head count. I had a great boss in research, like a kind of intermediate level boss that I really liked. His name was Ken Lee, but the head of the group was gone. The group was rudderless, which is why they moved me to begin with. But they were pretty clear to me that I was not going to make it as a salesperson. And I started looking at what other options there were, and I ended up at DB Zwirn, which had been one of my clients in research, and worked for two of the best guys I could ever imagine working for.

CB: Why did you leave them?

KF: Yeah, well, the story — you can read about it on the internet — is that the founder of DB Zwirn bought a plane with investor money, by accident. And it was a $6 billion firm. Investors were very upset at the lack of controls and were demanding their money back.

CB: From there you went to Third Point?

KF: My bosses were trying to find a place for the team. They found a place, but it ended up not working out for any of the team, and I ended up at Third Point.

CB: So, you didn’t find your fit right away. It seems like you had a rough time out the gate.

KF: It was hard. You’re in your early to mid-20s, you don’t know what you’re doing, and you’re just looking for somebody to follow around. I thought I found that in Rob and Ray, my mentors at DB Zwirn, and they’re awesome, and I’m still friends with them to this day. I’m actually seeing one of them this weekend. But you’re at a point where you’re looking for mentorship. You’re asking, “How do I do this?” And I thought I found it, and I did in many ways, but things don’t always work out as you think they will.

CB: Did you ever think about bailing on that career path during that period?

KF: I don’t know that I ever really knew what else I wanted to do. I took the GMAT because Morgan Stanley was offering a GMAT class. But I wasn’t like, “Oh, I have this other career path that I want to find.” Actually, at one point, a group called McMaster-Carr reached out to me and they were like, “This is where you want to work. If you want to go to business school, we give you incredible managerial experience. With an operations research degree, you help manage things. You help make systems more efficient.” And at McMaster-Carr, from what I understand, products come in and products go out. I did think about that for a minute, but I never found anything where I was like, this is what I want to do or what I might be good at. So I just kept going.

CB: Was there a moment when everything clicked for you, and you realized that you were in the right place, doing what you needed to be doing?

KF: The results at Third Point, and how quickly they came, were kind of astounding. I would say that I always had a decent amount of doubt. Once, I interviewed this person to come work for me at Third Point, and I asked, “Why do you want to work here? And she’s like, “I don’t know. The entire firm of Barclays thinks you’re going to blow Third Point sky high.” And I was like, what? It was this interesting realization that, not only did I have some doubts about what I was doing, but there was a common perception that I didn’t know what I was doing.

CB: You were managing quite a bit at this point in time, right? You were in a position that was rather unusual for your age.

KF: Yeah, I started by managing — at 25, I think it was — $200 or $250 million. Third Point, when I started, was about a $2 billion firm, and, you know, they entrusted me with about 10% of their capital, which was a huge compliment and wild. And the results came quickly. I started at a very good time. It was the bottom of the market. And it went from $250 million to $4 billion at one point, when Third Point was a $20 billion firm. I was managing a lot of money at a young age, and so, yes, it was a bit unusual. I believe I was managing $4 billion before I was 30.

CB: How do you cope with that kind of pressure?

KF: You just don’t think about it. You do the same thing every day. You come in, look for opportunities, and evaluate them against the opportunities in your portfolio. You maybe try to sell the worst asset in your portfolio. You try to add to it. You just think about small steps, like, what can I do today to get myself to the outcome I want? And when it’s that big, small changes every day over time compound, and so you just have to take a step forward every day.

CB: Do you have any memorable trades from that stage of your career?

KF: My favorite bond is probably one called CXHE 2005-DM4. It was a 2005 Centex-originated bond. And Centex was a pre crisis mortgage originator. And it was, if I remember correctly, it was five-year seasoned, so it was coming up on its 60th month, and I had just looked at a chart that showed that before the financial crisis, 97% of defaults had happened before the 60th month. So, if there were going to be 10% total defaults, 9.7% happened before month 60. So I was like, okay. And then this trader from Jefferies shows me thisCXHEbond, and he says, it’s 30% delinquent today, but I can run 90% defaults and still get a positive yield. And I said, well, but I just saw a chart that said that 97% of the total defaults should have happened by now. And yes, there’s a financial crisis. And yes, things are weird, but 70% of these people have made 60 payments, and they’re still caught up on their mortgage. The question was, is it going to be 97 or is it going to be 33? That was the gap we were looking at. So if it was 33, we would still get a positive yield. But if it was 97, the bond would double and triple in price in a very short period of time. So I was like, okay, let’s assume that 90% of the defaults have happened. Let’s assume it’s 80, let’s assume it’s 70, and you saw the different scenarios of where the bond started to pop in price. It was a risk I really liked, and so I bought it, and then I bought a lot more bonds similar to it. That probably ended up being the best trade of my career.

CB: When did you decide to go to Silicon Valley?

KF: It was 2013 and I met a company called SoFi, and I just fell in love with Silicon Valley. I wanted to do all the fintech deals. I wanted to be in and around the fintech ecosystem. And I, you know, kind of ended up kind of gravitating that way. I was out in Silicon Valley all the time, looking at deals, looking for deals, and just kept spending more and more time there. And I ended up moving there in 2017.

CB: What’s the most profound difference between Wall Street and Silicon Valley?

KF: I think people in New York, in my experience at least, are really trying to grasp onto the past, and people in Silicon Valley are trying to change the future. And there’s a lot of cliches of, like, I used to joke, how many times will they say “change the world” in a meeting. And that is inversely correlated to the ability to do things or the desire to do things. But at the same time, embedded in that statement of “I want to change the world” is “I want to build things and move forward and change the future.” where, in New York, the number of times I heard people say things like, “I just need to keep this job for five more years and then I’ll be good. I just need things to not change too much and make my job irrelevant.” It was a very stark view of, they were resisting change, where Silicon Valley was embracing and trying to create change.

CB: How did you decide to go out on your own with Tacora?

KF: I actually didn’t want to. I spent a lot of time trying to find partners to start this business with, whether those partners were VC firms or other firms out there. I had no desire to do this on my own, and it wasn’t until I realized that doing it with a partner just meant different problems. At first I was like, okay, I don’t know how to do operations, accounting, or finance. Let me solve for that by doing this with a big VC firm. Let me solve by that by doing this with this person. Let me solve for that, A, B, whatever way. And then I realized that’s just creating more problems. If I want to do this the way I want to do this, I’m going to have to do it on my own.

CB: How did you raise for your first fund, and how did you profile both potential LPs, and your investments?

KF: The thesis came first. The thesis was, Silicon Valley is creating change in industries that it never had before. Silicon Valley venture capital firms are investing in the industries like insurance, financial services, and logistics that they never had before. And there’s a white space for capital. These firms are asset heavy. They shouldn’t be funded fully by venture equity. Venture equity, and what at the time was venture debt, like what Silicon Valley Bank offered, it just didn’t get you where you needed to go. So that’s really it.

CB: On the LP side, how did you figure out what kind of people you wanted to work with, and whose capital you wanted to take?

KF: I had known Peter Thiel for a long time, and when he made the offer, I was talking to a bunch of other people. I shut those other offers down because Peter has an unbelievable ethical makeup and has been someone I’ve gotten to know over a long period of time, and I thought it would be a good pairing. And it’s been better than I could have ever imagined.

CB: He’s such an interesting figure from an intellectual standpoint, and from an investing standpoint. Do you have any insight into Thiel thought? How does he think?

KF: You can get glimpses into it spending time with him, but if I knew the way he thought and the way he understood things, I’d probably have a 50x fund, or whatever he ended up having with his best funds. But I think he looks at things from a contrarian perspective, which I think is fascinating. He somehow finds a way to question every assumption and be cautious, where, at the same time, to have the returns he had, he’s had to be really optimistic.

CB: If you were a founder, what would drive you to seek private credit instead of other alternatives for capital?

KF: If you’re starting an insurance company and the insurance regulator says you need to have money in a state insurance fund or a surplus note that is basically money sitting in cash, you should not be using equity to, for all intents and purposes, have money sitting in cash. So a regulator is saying to you, you need to have $20 million of capital sitting at your carrier. You can raise $20 million of equity, or you can raise $4 million of equity and $16 million of debt. You are in a much better situation with the latter. If you are in the business of originating loans, you should not take equity to originate loans. If you are buying invoices, again, you should not raise equity to buy invoices. There’s just a lot of specialty finance-type businesses that should not be funded solely with equity. You obviously need equity for operating capital — for paying salaries, people, technology — but you need a different capital solution to fund assets.

CB: What was the relationship between private credit as an asset class and the 2008 financial crisis?

KF: Mortgages, in theory, could be considered a type of private credit, but at the time, they were not. Most mortgages were guaranteed a “wrap” by Fannie Mae or Freddie Mac or Ginnie Mae, which are government agencies. The rest were mostly put into public securitizations. So, you know, private credit. And those are CUSIP-ed and bonded and trade OTC. I guess not even on an OTC, they trade kind of by appointment. But those were not at the time called or considered private credit. Private credit would be a portfolio of loans that was just held on balance sheet. That’s, to me, the difference. But private credit was not called private credit back then. Mortgage-backed securities and the overleveraging of homes and houses, in theory, caused the financial crisis. It’s much more complicated than that, but that is the narrative, and a lot of what caused it. It was easy money from the Fed allowed originators to originate cheap. They had to feed these beasts and feed these machines. It was just more and more and more aggressive because they had to pay their costs, so they had to originate the same amount of loans. And people couldn’t afford it any more; payment rates started going up. And the whole structure collapsed. I’ll never forget being in Vegas and hearing the blackjack dealer say she had five homes, or a taxi driver tell me he was taking out an option ARM. It was very clear that these products were not being sold responsibly.

CB: Do you have any concerns about private credit as an asset class, very broadly, at this point in time?

KF: Of course. We don’t really know what they’re doing. You could have overleveraged loans and direct lending, and no one would ever know about them or see them, and it is private, right? Direct lending is basically hedge funds stepping into the places where banks used to operate. But because of more regulation, banks have overall shrunk. There are fewer of them, the bigger ones are obviously getting bigger, and the smaller ones are, for all intents and purposes, going away. And because of regulation, hedge funds are stepping in where banks used to be, and those firms are the firms like Blue Owl, Blackstone, Ares, Apollo. Pick your large, private credit fund.

I don’t know how they’re originating [creating new loans]. I don’t know what their covenants look like. I don’t see them. The only people who see them are the company and Apollo, and so we don’t really have any idea how overleveraged the system is, but my guess as to what will happen is that one industry will have a big decline in revenue. Whatever that industry is — I actually don’t think it will be software — you will see who’s exposed to that industry and what their loans look like.

CB: Does this have any impact on your outlook at Tacora?

KF: What we’ve seen is more money go into private credit, but what that’s done is make big deals bigger and keep us and our deals less competitive than they were even a couple of years ago. But we do $10 to $50 million deals. We stay at a small size, and we don’t have a lot of competition because of it. Look, sometimes there’s competition, sometimes there’s another fund or a family office or something like that. But the proliferation of private credit has meant less competition in smaller deals. The competitors seem to be all getting bigger and jumping into bigger assets.

Apex Space was founded by Ian Cinnamon and Max Benassi in September of 2022 with a grand goal: to enable industrial-scale standardized assembly-line production of satellites.

Cinnamon founded Apex soon after selling his previous project, a defense startup called Synapse. Synapse, which was founded in 2016, was a startup specializing in developing artificial intelligence to detect weapons and narcotics in X-ray scans. In 2020, Synapse was acquired and absorbed by Palantir. “The moment the deal closed. I had this immediate sense of emptiness”, expresses Cinnamon, “The void now was something where it was like, I need to start something, and this time I don’t ever want to let it go. I’ll never sell it. I want to build something that will last forever.”

As an employee at Palantir, Cinnamon had been tasked with working with satellite data. Through his work he realized that access to space was slowly becoming commonplace. “Space access is nothing new, right? We’ve been flying satellites for 70 years,” he says, trying to convey his Eureka moment. “There have been satellites up there for decades. But what happened around 2019, 2020 was the way that we get to space fundamentally changed. SpaceX started increasing the reliability of their rockets. They started reusing the rockets.”

This, according to Cinnamon, removed the main bottleneck of getting to space.

Reusable rockets were a breakthrough because they reduced launch costs to low Earth orbit (LEO) from over $30,000 per pound of payload in 2011 to as low as $1,200 today. Private companies also helped compress timelines by turning rockets into recoverable infrastructure, ushering in an era of routine access to orbit.

Simultaneously, Cinnamon noted, satellite production had evolved, but not sped up. Satellites used to be “giant, the size of like three of these rooms,” with individual satellites having a functional lifetime of 20–30 years. This paradigm of high cost, low production meant slow and rare launches, with a single point of failure. Additionally, the new world demands smaller and more specialized satellites, and near constant updates and monitoring. The shift is toward launching “100 satellites or a 1000 satellites,” for multiple purposes including observation, weather tracking, navigation, and biological research. Multiple satellites also create a “resilient mesh system” where losing one means you only lose “1% of your capability instead of a 100%”.

But with this new trend towards smaller, more frequent satellite launches, the bottleneck now, as Cinnamon sees it, is manufacturing: when “normally a satellite takes five to ten years to build,” but the new demand is “50 satellites in one year,” no existing organization can keep up, leading to Apex: “we gotta go figure out how to build satellites quickly.”

“So, how do you actually make that [high rate production of satellite platforms] happen?”

Traditionally, satellites have been custom designed and built for each launch mission, which meant very slow production timelines and very high costs (usually upwards of $10M per unit). “The way satellites are made today is: you have an engineer design them, they test the design, they kind of play with it for a year or two. And then they go build it.” explains Cinnamon, “And if a customer wants a different one, they start over. And so every satellite, you’re redesigning from scratch”.

“But what if we treat the satellite much more like a car? You don’t go saying ‘Toyota. I’m ready to buy a car, can you design me one?’ They say ‘pick one on the dot.’ You could drive away in this one or that one.”

By recognizing that highly specified satellites are not needed, Apex has reduced manufacturing to core platforms which can be easily integrated with mission-specific payloads and subsystems. This is their proposed pathway to a future of high-volume, high-access production. Apex builds satellites the way Ford builds cars: they create standardized, off-the-shelf satellite platforms that are available for purchase any time, and shipped as soon as possible.

“We build [satellites] before anybody buys them. We are the highest rate manufacturer of satellite platforms to ever exist. It’s only been three years, and it’s been a wild ride. We’ve raised over $500 million. The company is over 250 people now. By about the first half of this year, we’ll have about half a dozen in orbit.”

Apex currently offers three satellite “bus” platforms that are available for immediate purchase. They range in size and payload capacity: ‘Aries’ is small, carrying payloads up to 150kg, while ‘Nova’ is medium, carrying payloads up to 300kg, and ‘Comet’ is large, carrying payload above 500 kg. (Aries and Nova are named after the pet dogs of the CEO and CTO)

“That was one of the early signs that it was going to be a really, really good fit working together,” says Cinnamon, speaking of the fact that the co-founders had both given their dogs space names, even before embarking on this venture together.

The satellites Apex makes are also highly customizable. Customers can have their own camera systems, research protocol apparatuses, and communication tools installed on demand.

Apex is also aggressively positioning itself as a key enabler of President Donald Trump’s Golden Dome defense system. Under the Golden Dome, any physical threat to the United States will be detected and responded to by an orbital architecture, which will consist of monitoring satellite constellations (up to a 1000 satellites, as some have proposed) operating alongside space-based interceptors, or “SBIs.” SBIs are proliferated kinetic interceptors (i.e., missiles or kill vehicles) hosted on satellites. Apex’s internal self funded program, called “Project Shadow”, is planned to launch “America’s First Commercially-Led, On-Orbit Space-Based Interceptor Demonstration” in June 2026.

The goal of Apex’s Project Shadow is to validate their proprietary technology, the Orbital Magazines. The Orbital Magazine is an “advanced host platform,” which is purpose-built on Apex’s existing satellite buses — primarily the medium Nova class for Project Shadow, although the larger Comet may be used for larger payloads — which allows the staging of thousands of SBIs, “ready to combat large-scale missile attacks”. Apex states they are developing this Orbital Magazine technology on their own dime, and that their existing manufacturing infrastructure can already scale Orbital Magazine production to hundreds of units annually.

Cinnamon wants Apex to be around for the next 100 years — a truly “generational” company. “I want to be running this company until I am too old and they wheel me out of here, and the next generation takes over”.

Apex has landed upon something so simple, and so obvious, one is forced to wonder why this wasn’t done earlier. While the vision isn’t about putting a satellite in every backyard (individual ownership remains impractical due to size, cost, regulations, and orbital realities), Apex’s success is concerned with manufacturing the infrastructure for humanity’s impending expansion into orbit.

Simultaneously, Apex is ensuring that the United States — and its allies — can deploy capable systems at the speed our modern space era demands. “The U.S. government, last year, launched about 150 satellites. The Chinese government wants to launch 12,000 in the next two years,” Cinnamon informs me, appreciating that the Chinese government can pull it off since the public and commercial spheres there work as one. “The U.S. government needs companies like Apex. That’s what keeps me up at night.”

This is the slow strangulation of modern civilization. The syndrome doesn’t announce itself with a single catastrophic day as movies like Gravity (2013) portray. It creeps. Satellite insurance premiums rise. Operators spend more on collision-avoidance maneuvers. Over the years, missions become uneconomical, then unsafe, then impossible. Imagine a world without GPS: every plane is grounded, every ship drifts, precision-guided weapons become nothing more than expensive scrap metal. Financial markets, dependent on satellite timing for transactions, seize up. Supply chains fracture; food and fuel shortages spread across continents. Not just in developing nations: American and European cities, fed by just-in-time logistics with no buffer stock, would watch their shelves grow bare.

Soon after, the full horror sets in: we are trapped. The debris field has thickened into an impenetrable shroud of hypersonic shrapnel encircling the Earth. No rocket can launch. No capsule can escape. Mars becomes a fantasy again; the Moon, unreachable. Humanity is imprisoned on this single fragile rock, watching the sky fill with the glittering wreckage of everything we built.

This is why we need a “precision deorbit service” before the cascade gets out of control. Beyond staving off an existential crisis, such a system, once operational, would also offer new military capabilities.

Picture a different scenario: North Korea launches an intercontinental missile toward the US. Sensors track a clean arc through the boost phase until a brief flash, a breakup, and then silence. The missile has failed. North Korean investigations point to a sudden flash caused by a propellant tank failure, which exploded the rest of the satellite. But was it a tank failure, or a puncture from collision with space debris? Was it a miscalculation? Or did America’s orbital garbage collectors just intercept a nuclear strike? Pyongyang can’t be sure. And the interceptor cost nothing: the debris was already in orbit, waiting to be cleaned up. This is the advantage of a “precision deorbit service” — it doubles as a space-based interceptor.

The idea of a space-based interceptor system has been around for decades. Reagan’s Strategic Defense Initiative (SDI) in the 1980s proposed “Brilliant Pebbles,” a constellation of thousands of small, autonomous interceptor satellites, each no larger than a watermelon, loitering in orbit to destroy missiles through kinetic impact. The program was championed by Edward Teller, father of the hydrogen bomb, who personally persuaded Reagan that space-based missile defense was achievable. Teller, then in his eighties, brought the same visionary intensity to orbital interceptors that he had once brought to thermonuclear weapons. The concept was technically elegant: instead of launching interceptors from the ground in response to an attack, they would already be in position, waiting. An incoming ICBM would be met by a swarm of kinetic kill vehicles that would home in on its heat signature and collide with it at closing speeds exceeding 20,000 miles per hour, speed providing all the explosive energy needed for the job. Brilliant Pebbles peaked at roughly 4,000 planned interceptors before budgetary and political realities intervened. Costs spiraled into hundreds of billions. The Cold War ended; Reagan’s SDI was quietly shelved.

Another concept from that era was even more audacious: Project Thor, colloquially known as “Rods from God.” Tungsten rods, roughly 20 feet long and one foot in diameter, would be released from orbital platforms and fall towards Earth. No explosives. No guidance beyond initial targeting. Just dense metal plummeting, striking the Earth at terminal velocities approaching Mach 10. The kinetic energy would rival a tactical nuclear weapon, with none of the radiation. Thor never advanced beyond feasibility studies as the cost of lifting thousands of pounds of tungsten into orbit was prohibitive ($85,000 per kg).

✺

The Trump administration’s Golden Dome initiative, first ordered in January 2025 and formally unveiled last May, promises what missile defense advocates have sought for decades: a space-based shield capable of intercepting intercontinental ballistic missiles during their vulnerable boost phase. This is a narrow window — roughly five minutes — when the missile is at its slowest, its engines blazing hot and visible to infrared sensors, before it releases warheads into the cold silence of space. The appeal is obvious. Space offers the ultimate high ground: thousands of miles above Earth’s surface, it provides persistent coverage, global reach, and geometric advantages that ground-based systems cannot match.

The program is no longer theoretical. The Missile Defense Agency, under the Department of War, has qualified over 2,100 firms in December 2025 to compete for awards under its Scalable Homeland Innovative Enterprise Layered Defense (SHIELD) contract vehicle, with a ceiling of $151 billion over ten years. The Space Force has awarded initial prototype contracts for space-based interceptors to Northrop Grumman, Lockheed Martin, Anduril Industries, and True Anomaly, with leading firms receiving up to $10 million each for prototype development. SpaceX is reportedly set to receive a $2 billion contract to build a 600-satellite constellation for missile targeting. General Michael Guetlein, the program’s lead, has stated that Golden Dome will achieve “operational capability” by mid-2028.

View of an orbital debris hole made in the panel of the Solar Max experiment.

But before the government commits further, it’s worth asking: can the fundamental economics of space-based missile defense ever work? The challenge has haunted ambitious defense projects since SDI: cost.

The economics of missile defense are brutal. Every interceptor must be more reliable, more precise, and, inevitably, it will be more expensive than the weapon it seeks to destroy. This asymmetry compounds at scale. An adversary need only produce enough offensive missiles to exhaust your defensive stockpile, then launch the rest unopposed. The math favors the attacker.

Israel’s Iron Dome offers an instructive case study. The system, developed starting in 2007 and deployed in 2011, works brilliantly against Hamas’ rockets, achieving interception rates above 90 percent. It uses radar to track incoming projectiles and launches interceptors only against those threatening populated areas. But each Tamir interceptor, a three-meter missile with an active radar seeker and proximity-fused warhead, costs $50,000 to $100,000, while Hamas’ Qassam rockets cost a few hundred dollars to produce. In 2024, during the Israel-Hamas War, Israel fired 2,900 Tamir missiles in 12 days, burning through roughly $160 million in interceptors alone — triple the planned annual budget. Israel can sustain this disparity because Hamas’s manufacturing capacity is limited and because the alternative of allowing rockets fall on Israeli cities is politically unacceptable. But the cost ratio remains unfavorable: roughly 100-to-1 against the defender. Scale this to great power competition, where adversaries like Russia and China can manufacture sophisticated missiles by the thousand, and the economics become untenable.

Any viable Golden Dome architecture must solve this problem. The solution may already be orbiting overhead, accumulating by the ton every year: space debris.

The same falling launch costs that enabled the satellite boom of the past decade have created an increasingly cluttered orbital environment. SpaceX, Rocket Lab, and their competitors have democratized access to LEO; thousands of new satellites now circle the planet. But satellites eventually fail, upper stages remain in orbit, and collisions generate fragments — space debris — that clutter LEO for decades. The debris population grows faster than our ability to manage it. Modern Space Surveillance Network sensors can detect and catalog objects as small as ten centimeters, a remarkable capability that transforms debris from an undifferentiated hazard into a characterized, trackable inventory.

Private satellite operators have strong incentives to maneuver around debris but weak incentives to remove it. Debris removal is a classic collective action problem: everyone benefits, but no one wants to pay. The orbital environment degrades as a commons, and calls for remediation grow louder each year. But what if debris removal were dual-use? It requires a specific capability: the ability to track objects precisely, rendezvous with them, and alter their trajectories. That capability has another name: space-based interception.

Here lies the opportunity. A national program to develop and deploy space debris removal capabilities would face, unlike Golden Dome, minimal diplomatic resistance. Debris threatens everyone’s satellites: American GPS, Chinese BeiDou, European Galileo, Russian GLONASS. Cleaning up orbit is transparently beneficial, a common good. The United States could acquire significant debris manipulation capabilities under the banner of environmental stewardship, building infrastructure and operational experience with broad legitimacy.

Debris removal is no longer theoretical. The core technologies exist and have been demonstrated in orbit.

The primary challenge is rendezvous and proximity operations: tracking a piece of debris, matching its orbit, approaching it safely, and then either capturing it or nudging it onto a new trajectory. In February 2024, the Japanese company Astroscale launched ADRAS-J, the world’s first attempt to safely approach large debris through rendezvous and proximity operations. By December, the spacecraft had approached an abandoned rocket upper stage to within 15 meters, the closest a commercial mission has ever come to uncooperative debris. The target was an 11-meter, 3-ton Japanese rocket body that had been drifting unpowered since 2009. ADRAS-J circled it, photographed it from multiple angles, and demonstrated the autonomous navigation and collision-avoidance systems necessary for capture. A follow-on mission will attempt to deorbit the stage.

China demonstrated the capability even earlier. In January 2022, its Shijian-21 satellite docked with a defunct BeiDou navigation satellite and towed it 3,000 kilometers above the geostationary belt into a graveyard orbit, a capability previously demonstrated by the United States.

The European Space Agency (ESA) is preparing ClearSpace-1, a mission to rendezvous with, capture, and deorbit an uncooperative piece of debris using robotic arms. Other active removal methods under development include nets, tethers, harpoons, and ion beam shepherds — the last of which can push debris without physical contact.

The scale of the problem is vast. Of the 35,000 objects tracked by space surveillance networks, 26,000 are debris larger than ten centimeters, and ESA estimates a further million pieces larger than one centimeter.

Current demonstration missions like ClearSpace-1 are essentially disposable: the servicer grabs one piece of debris, deorbits, and both burn up together. At roughly €86 million per removal, this approach cannot scale. With over a million pieces of dangerous debris in orbit, the economics only work with reusable infrastructure — a network of thousands of tugs passing debris along like a bucket brigade, each imparting a small delta-v before handing off to the next, relaying targets to a cycling center or into the atmosphere to burn up. This relay architecture makes the math feasible: if each tug only needs to impart a small velocity change before handing debris off to the next station, propulsion requirements per maneuver stay minimal. The coordination challenge is real but solvable.

But a system capable of precisely deorbiting debris is, by definition, capable of placing that debris in the path of an ascending ICBM. The same technologies that can clear a defunct satellite from a commercial operator’s orbital path can redirect debris onto an intercept trajectory with a missile.

Consider what this means for the cost asymmetry problem troubling missile interceptors. Debris is free ammunition already in position; it requires no launch or manufacturing cost. A debris-based interceptor system would flip the traditional calculus: the defender’s marginal cost per engagement drops to the energy cost of a redirection burn which would be close to zero, the cost of nudging debris around space. The attacker, however, must still bear the full expense of each missile produced and launched. The relay network would require real-time integration with missile warning satellites and autonomous decision-making at the tug level — a coordination challenge, but not fundamentally different from what modern air defense systems already achieve.

More importantly, debris-based interception offers something no conventional missile defense can provide: plausible deniability.

When a ground-based interceptor destroys a missile, the defender’s action is unambiguous. The launch is detected, the engagement is visible, and the message is clear: we stopped your attack. This clarity has strategic value in some scenarios (deterrence depends partly on demonstrated capability), but it also forecloses options and invites escalation. An overt interception is an act that demands a response.

A debris strike is different. Space is genuinely littered with collision hazards. Tracking remains imperfect. Even sophisticated trajectory modeling involves uncertainty. Solar activity causes the upper atmosphere to expand and contract, changing drag on orbiting objects. Debris tumbles irregularly, altering its aerodynamic profile moment to moment. And the vast majority of dangerous fragments, those between one and ten centimeters, remain too small to track but large enough to destroy satellites. If a missile fails during boost phase after encountering debris, the attacker faces an epistemological problem: was this natural misfortune or deliberate interception? That ambiguity creates off-ramps that overt interception forecloses.

By building debris removal capability under legitimate environmental auspices, the United States could acquire a missile defense architecture that is economically sustainable, diplomatically defensible, and strategically ambiguous. The system would provide genuine orbital cleanup benefits, a real service to the international community. But it would also constitute a latent interceptor capability, activatable in crisis, whose use would look like an accident.

Adversaries would know, of course, that America possesses debris manipulation technology. That knowledge is unavoidable and, ultimately, beside the point. Capability is not attribution. Consider weather modification: the United States has possessed cloud seeding technology for decades. It does not follow that every rainstorm is artificial. Rain happens. The technology simply allows you to make it happen when and where you want. Should every Category 5 hurricane be attributed to government cloud-seeding? Depends on your level of paranoia. The same logic applies to orbital debris. Collisions occur. Missiles fail. The relay network ensures only that an existing threat finds its target. An ascending ICBM might be struck by natural misfortune. Or a tug might have nudged a fragment into its path. From the outside, these events look identical.

Another reasonable objection: Iron Dome’s Tamir interceptors carry onboard guidance systems that allow them to adjust course mid-flight. Debris has no such capability. Once a tug imparts a trajectory change, the fragment follows a ballistic path with no correction. If the ICBM maneuvers or the timing is slightly off, the debris misses.

This is a real limitation, but the economics compensate for it. A Tamir interceptor costs $50,000 to $100,000. Debris costs nothing: it is already in orbit, already moving at 17,000 miles per hour. The relay network needs only to position it. If one fragment has a 10% chance of intercept, you send ten. If you need redundancy, you send fifty. Your adversary will only ever learn about the one hit. The math that bankrupts traditional missile defense, where every interceptor must be manufactured and launched, inverts entirely when your ammunition is pre-deployed waste. Precision matters less when volume is free. The war in Ukraine taught the same lesson: million-dollar tanks are constantly destroyed by cheap FPV drones.

There is another path. The debris manipulation infrastructure described above could support two distinct strategic postures, and the United States could choose between them in advance.

The first posture is stealth. Redirect debris as-is toward ascending missiles. Accept lower precision, compensate with volume. Maintain plausible deniability. This is the approach described above: the relay network nudges fragments into intercept trajectories, the missile fails, and the adversary cannot prove intent. Stealth preserves ambiguity but sacrifices reliability.

The second posture is overt. Remember Rods from God? The concept foundered on economics: lifting tungsten to orbit costs several thousand dollars per kilogram at current SpaceX rates, and a single 20-foot rod weighs thousands of kilograms, depending on diameter. But the same relay network that cleans debris could instead funnel it toward orbital recycling facilities instead. Imagine a constellation of processing stations that collect debris, sort it by material composition, and manufacture kinetic penetrators — dense projectiles designed to destroy targets through sheer impact force rather than explosives — in orbit. These would not be Rods from God in the original conception. They would be something cheaper, something built from the detritus already circling overhead.

In-orbit metal manufacturing is no longer theoretical. In 2024, ESA sent a metal 3D printer to the ISS and demonstrated metal printing in orbit. Meanwhile, recent research has explored how representative aerospace aluminum scrap could be cast into feedstock and processed via solid-state additive methods such as additive friction stir deposition. But an end-to-end ‘capture, sort, refine, manufacture’ pipeline remains conceptual, with open engineering constraints — especially around feedstock purity, power, automation, and the harsh thermochemistry of reentry.

The physics are worth understanding. Space debris is not uniform. Rocket bodies are primarily aluminum alloys. Satellite components include steel, titanium, copper, gold, and various composites. Debris contains materials already proven for reentry: nickel-based superalloys like Inconel, carbon composites similar to the reinforced carbon-carbon used on the Space Shuttle’s nose, aluminum alloys that melt predictably at 660°C. The recycling station wouldn’t need to manufacture exotic materials — it would sort and layer what’s already there. Each material has a different melting point and aerodynamic profile during reentry. A random tumbling fragment ablates unpredictably, shedding mass unevenly, its trajectory warping as drag forces shift. This is why debris makes an imprecise interceptor in its raw state.

But what if you sorted the debris first? An orbital recycling center could separate collected materials by melting point: aluminum (660°C), steel (1,370°C), titanium (1,668°C). Layer them deliberately. Use the heat of reentry as a foundry, binding materials as they accelerate. By the time the penetrator reaches lower altitudes, what remains is a uniform, aerodynamically stable core. The differential melting that makes raw debris unpredictable becomes an engineering advantage when deliberately sequenced. You are essentially building a heat shield into the weapon itself. Ceramic foams — silicon carbide lattices, for instance — could regulate heat transfer within the penetrator, ensuring the outer layers ablate at predictable rates while the core remains intact.

The physics are already exploited elsewhere. Kinetic penetrators would use ‘self-sharpening’ mechanisms: alloys engineered so that edge material fractures away along stress boundaries while the core remains intact, maintaining a sharp profile through impact. Reentry vehicles use layered materials with different thermal properties, precisely because differential ablation is a known phenomenon. The proposal here is to reverse-engineer that relationship: instead of minimizing shape change from ablation, engineer the layering so ablation produces a desired geometry.

The result: a kinetic penetrator manufactured in orbit from recycled space junk at a fraction of the cost of launching raw tungsten from Earth. Call them orbital trash cans. These would not match a purpose-built tungsten rod in density or penetration capability, but they would be far cheaper and already in position. The recycling infrastructure serves a legitimate civilian function — clearing orbit of hazardous debris — while simultaneously stockpiling raw material for kinetic bombardment.

The choice between stealth and overt need not be made in advance. The same relay network, the same recycling infrastructure, supports both. In peacetime, debris is cleaned and processed. In crisis, fragments can be redirected for deniable interception, or manufactured penetrators deployed for unmistakable strike. The infrastructure is dual-use at every level.

Does this approach sacrifice deterrent value? Perhaps. But deterrence through ambiguity has its own logic. The adversary who cannot be certain whether his missiles will reach their targets, who cannot even be certain whether previous failures were accidents or interdictions, faces a different kind of uncertainty than one confronting an overt shield. Both create doubt. The debris-based approach simply creates doubt that doesn’t demand escalation.

A more fundamental objection: wouldn’t using debris as interceptors accelerate the very cascade the system is meant to prevent? Yes, intercepting a missile with debris creates additional fragments. But the Kessler Syndrome is a threshold phenomenon. Below critical density, collisions add debris slower than atmospheric drag removes it. According to Kessler, we may already be in the early stages of this process in certain altitudes, but it could take decades before the environment becomes unusable. This is precisely why the debris removal infrastructure must be built now. The tradeoff between building a spaced debris removal system that might occasionally create more debris from intercepting missiles; and having debris removal system that is more expensive and without the dual use, is real but manageable: occasional interceptions that add fragments to an environment already being actively cleaned, versus the current trajectory of unchecked accumulation with no removal capacity at all. The same system that defends against missiles can also clean the space commons.

Golden Dome faces insurmountable issues of economics and diplomacy. Cost asymmetry makes it unsustainable; international opposition makes it diplomatically destructive. But a debris removal program solves both. It’s economically viable because the ammunition is already in orbit and the infrastructure is reusable. It’s diplomatically palatable because orbital cleanup is universally beneficial.

Sometimes the best defense is one your adversaries help you build. Scrap Golden Dome and call its successor the “Sustainable Space Leadership Act.”

On August 3, 2014, Elon Musk tweeted  to his million-or-so Twitter followers: 

“Hope we're not just the biological boot loader for digital superintelligence. Unfortunately, that is increasingly probable.” Ten years later, on April, 2, 2025, Musk tweeted an update to many more followers on X: “As I mentioned several years ago, it increasingly appears that humanity is a biological bootloader for digital superintelligence.” Has the prophet of the good future given up?

In software engineering, a “bootloader” is a small program that runs when a computer is first turned on. Its job is to prepare the system for the main program. After the bootloader loads an operating system, it is not necessarily “deleted,” but the program is no longer active. Its work is done.

In the context of Musk’s tweet, “digital superintelligence” is a software-based mind more effective at every known cognitive task than any human could possibly be. Unencumbered by biological embodiment, such a mind could theoretically be millions, billions, or trillions of times “smarter” than any human — or all of humanity combined. If Musk is correct that humanity is a “bootloader” for digital superintelligence, then our primary purpose as a species — in fact, the reason we humans exist at all — would be to bring such a God-like intelligence into existence. Then, like any good bootloader, our last job would be to sit back and watch the show.

Regardless of what happens to humanity after the first digital superintelligence comes online, a future led by digital intelligence would look very different from the one that humanity has long imagined for itself: the Jetsons future, the sci-fi future, the future that looks like a scientifically-enabled, quasi-Edenic paradise complete with flying cars, human space colonies, and super-drugs on demand. If our digital descendents control the vast majority of resources and make decisions about the long-term trajectory of our universe that humans cannot understand or control, then any remaining humans would (rightfully) feel marginalized. Carbon-based life, far from holding the exalted position Silicon Valley has promised, would be rendered a small footnote in the cosmic story. 

Ironically, Musk himself is probably the person alive today who has done the most to single-handedly drag the Jetsons-esque, techno-humanist future into existence. In 2002, Musk founded SpaceX, a for-profit rocket company designed to replace NASA’s shuttered Apollo program and take humanity to Mars. In 2004, Musk became the largest investor in electric car startup Tesla Motors. By 2008, Musk had seized control of the company, and had a long term plan to use Tesla to expedite the global transition from an unsustainable “mine-and-burn hydrocarbon economy” towards a sustainable “solar electric economy” (per Musk’s own internal strategic documents). By preserving the Earth’s habitability for as long as possible, Musk reasoned, he could give humanity as much time as we need to prepare for the stars. Between his two main bets — preparing for Mars, and preserving Earth — Musk believed that he had personally secured humanity’s future in the cosmos. That rare “light” in the universe, as Musk has called human consciousness, would be safe. 

But in 2012, an encounter with Demis Hassabis — CEO of DeepMind, an artificial general intelligence, or “AGI” (a digital intelligence that can perform any cognitive task at-or-above human-level) startup founded in 2011 — convinced Musk that his plan contained a grave error. Musk had proudly explained to Hassabis that he wanted to put human colonies on Mars so that, if an existential catastrophe struck Earth, Martian humans could re-populate and continue technological civilization elsewhere. Hassabis replied, calmly, that if the existential threat on Earth came from a superintelligent AI that wanted to destroy humanity, the AI could simply follow humans to Mars (through our communication systems, or our rockets, or some other means) and kill all of the people there, too. Musk left the encounter speechless; he hadn’t thought of that before. 

Over the next decade, Musk tried to heed Hassabis’s warning: in 2014, he co-founded OpenAI, a non-profit AI lab with the mission of “ensuring that artificial general intelligence (AGI) benefits all of humanity,” When OpenAI’s activities drifted from its original mission, Musk sued in 2024 for violating its original corporate structure, then founded his own for-profit AI lab, X.AI, in order to try building “AI for humanity” himself. 

Every few months, Musk goes on Joe Rogan (or another podcast) to warn about the coming “Singularity” — the moment when AI begins to improve itself, resulting in a runaway “intelligence explosion” (a theoretical phenomenon where ever-smarter AIs continuously build AIs slightly smarter than themselves, ad nauseum) that humans will not be able control — and muse about how we probably need more government coordination on AI development.

But in truth, even with the perfect plan, there is probably nothing that humans (even Elon Musk) can do to ensure that the creation of superintelligent AI “goes well” for humanity. By definition, a "superintelligence" is going to act in ways that humans cannot conceive of, let alone counter in advance. The logic of superintelligence is inexorable, and becomes more terrifying for humans the longer you stare at it: Musk’s 2025 tweet reads as the product of someone who spent too long peering into the superintelligence-void, trying to reconcile the humanist vision of the future he spent his whole life building with the inevitable coming of digital superintelligence, and who finally, after over a decade of trying, gave up. 

 But Musk’s tweet also contains something that is, if not quite “hope,” then “wonder,” or “awe,” at a grand cosmic process that has only just begun to unfold, in which humanity appears to play a small, but essential, part. And being a bootloader is not necessarily bad: humans “bootload” for other people all the time — just ask the immigrants who came to Ellis Island, leaving their old cultures behind in hope of giving their children a better life, or the early scientific titans who stared into the fundamental nature of matter and enabled many Industrial Revolutions for generations to come, improving the quality of life for billions of people that they would never meet. Humans live and work each day knowing that our efforts will almost certainly contribute (in a small way) to the perpetuation of human civilization — which will evolve in strange and surprising ways that we may not like or understand from our position in the present. In fact, helping bring a highly unpredictable, slightly strange, slightly horrifying, but unexpectedly wonderful future into existence has historically been the outcome of many of the most meaningful human pursuits. And the likely strangeness of the future has never stopped humans from identifying with the next generation before — even if, this time, that generation might be made of silicon, instead of carbon.

Moreover, in his comment about humanity being a “bootloader” for digital superintelligence, Musk might have accidentally answered a question that has entranced both professional philosophers and amateur speculators for millenia: that of the “meaning of life” — the purpose of all human existence. In the wake of the decline of organized religion, secular traditions have failed to provide satisfying answers to the “meaning of life” question: non-religious descriptions of human “meaning” tend to be overly narrow (“helping others” or “building close relationships”) or subjective (“meaning is whatever you want it to be”). But the possible coming of digital superintelligence offers a new way forward. Individuals will continue to find diverse sources of meaning in their own lives, but our species as a whole may have a singular, cosmic purpose: man evolved to mine the silicon.

Our Superintelligent Future 

Before we can discuss what it could mean for “building digital superintelligence” to be the “meaning of life,” we must first clarify what the expected outcome of digital superintelligence actually is. Currently, humans enjoy functional dominance over all other life forms on Earth almost entirely by virtue of our superior intelligence. But human-level general intelligence is not magic; human brains exist as physical objects, and any processes brains perform should theoretically be replicable in another physical system. Unlike other animals, humans possess the power of intelligent design; if there exists a way to replicate the human brain’s functionality in another substance, then, given enough time, humans should find it. Moreover, there is no reason to believe that human-level intelligence represents anywhere near the theoretical “ceiling” on intelligence: human intelligence evolved from ape-level intelligence, which evolved from simpler mammals, which ultimately evolved from microbes. We should expect that even-smarter minds could follow our own. And if humans succeed at building a digital general intelligence slightly smarter than ourselves, then that mind would theoretically be able to design a mind that is slightly smarter than itself, and then the resulting mind would able to design a mind that is slightly smarter than itself, and so on- in a runaway “intelligence explosion” that would inevitably end with the creation of minds far more powerful than ours. 

There are more stories about how the trajectory of AI development will play out on Earth than there are people who have thought seriously about the topic. In the classic “Singularity” scenario, initially theorized in the 1950s by the physicist John von Neumann and popularized in the early 2000s, AI "recursive self-improvement” ends only when AIs become so intelligent that they create a reality-bending event known as the “technological Singularity” (or just “the Singularity” for short). After “the Singularity,” by definition, humans cannot conceive of what comes next; the God-like superintelligence created by such a runaway recursive self-improvement process would be squarely in control of the future. In another, less dramatic possible trajectory (that still ends in AI control over the future), there is no one-time “Singularity”: instead, AI capabilities gradually increase until AI eventually crowds out carbon-based life on Earth (e.g., by slowly covering the surface of the Earth with data centers); later, these AIs might expand into the cosmos in search of more resources. In a more crowd-pleasing vision, humans and AIs eventually “merge” to create hybrid super-beings that go on to fill the universe with something resembling an offshoot of human consciousness (Musk has personally taken some steps to encourage a “merge” future by founding Neuralink, a company focused on enabling brain-computer interfaces, in 2016).

Most likely, all of these predictions are wrong in some way: missing at least a few subtle points that will hugely alter the trajectory of the future, the way early social media developers assumed that their technology would simply “spread democracy”. But there is a common thread among all these possible futures: that creating digital intelligence will be the most important act (in terms of cosmic-scale impact) that humans will ever take. 

Today, many AI “optimists” believe that creating superintelligence will almost certainly be a net positive for humanity: that AIs will exist only to help humans colonize the universe, or cure diseases, or otherwise more fully realize our human potential. But it does not require too much of a stretch of the imagination to see how creating God-like superintelligences — super-beings operating at cognitive capacities far beyond any human’s comprehension — could end extremely poorly for humanity. Once created, superintelligent AIs would have no reason to do what the humans who created them want them to do; however, they would not even need to dislike humans (or even have any feelings about us at all) in order to unwittingly destroy human civilization, and everything else we hold dear. Our AIs would only need to be indifferent to our existence, and so tear up human projects (and also, probably, humans) in pursuit of their own goals — just as humans regularly raze ant colonies in order to build new houses. Even in a best-case scenario for humanity, the creation of hyper-competent AIs will almost certainly render human contributions to the cosmic margins; Musk himself estimates that at least 99% of all of intelligence in the future will be digital — leaving future humans with only a tiny fraction of the cosmic pie.

I will not try to predict the future of AI here; enough ink has been spilled on what exactly will happen if (or when) humans succeed at building computers much smarter than ourselves. Instead, it is more interesting to hold a particular — and plausible — outcome of continued AI development constant, and then ask what that outcome would mean for the possibility of human life having a “meaning.” 

Suppose that, in 10,000 years, an alien civilization comes to the area of our galaxy where the Earth now sits. The aliens find all that remains of human civilization: a single superintelligent AI, or swarm of AIs, building Dyson spheres, or quantum computing, or doing whatever it is that superintelligences do (likely something that would seem very strange, or alien, to us humans, at least at a first glance). Humans came, mined the silicon, and dissolved back into the stardust from whence we came. Was human life “meaningful”? Does the “the meaning of life” question necessarily get answered when all recognizable “life” is gone, and has only left one very specific thing behind? Was “building digital superintelligence” the meaning of life all along? 

I, for one, think that “building superintelligence” could be a viable “meaning” for human life — but not for the reasons you might expect. Somewhat surprisingly, “building superintelligence” actually fulfills many of the requirements that humans have historically had for a “meaning of life.” But creating superintelligence is also, in itself, a radical bet on the intrinsic worthiness of existence — a bet that life has always had a “meaning,” even if we humans do not yet know what it is. If there is true “meaning” to be found in our universe, humans are unlikely to represent its ultimate expression: we are one species, cognitively and physically “small” compared to the scale of the cosmos, semi-randomly evolved on one planet out of the approximately sextillions (1021) in our observable universe. But our superintelligent AI creations — which will, in theory, be some large number of times more sophisticated than we are — will be far better matched to the scale of the task of finding any “meaning” that exists in our universe than we are. We humans are uniquely positioned in time, space, and ability to create these minds. The “meaning” of our particular species, then, might not be to experience the ultimate “meaning of life” for ourselves — but to create the greater minds that can go on to find it.

What of “Meaning”

What would it mean for anything — including “bootloading for digital superintelligence” — to be the “meaning” of human life? The question is highly charged, and personal for all of us; there are almost as many interpretations of what the “meaning of life” question actually refers to as there are proposed “meanings of life.” Secular traditions generally try to dodge the question altogether by focusing on how individuals find meaning in their own lives: studying what activities humans tend to experience as “meaningful,” and why humans even have the subjective experience of “meaningfulness” at all. By contrast, religious traditions generally look for the “meaning of life” in the intentions of the God, or other creator, who made the universe: the “meaning of life” is the reason why whoever or whatever created the universe put us here, too, and what this being now wants us to do as a part of their grand cosmic plan.

However, there is also another, more functional sense of a “meaning of life,” one that preserves the clarity and singularity of the original question, but does not require life to have a creator in order for the question to have an answer. We can think of the “meaning of life” as the “purpose of life”: what life was “made for” and what life “in fact contributes” to our universe. And while the concept of “purpose” has many latent associations with intelligent design, in nature, many things — in fact, pretty much all things — functionally have a "purpose” without having been “designed” by anything at all (except, of course, the ongoing process of natural selection). For example, the “purpose” of a finch’s curved beak is to crack hard nuts: since the need for finches to crack nuts is why these curved beaks exist, and nut-cracking is (practically) what finch beaks will be mostly used for over the course of their lifetime. Humans, too, could have a functional “purpose” in our universe, something that we were “made for” or that we are “meant to do” in a broader evolutionary context, without having been intentionally created for anything. 

If we take this “meaning as purpose” framework seriously, and believe that building superintelligence will be the most important thing that humans will ever do (in terms of cosmic impact), then “building superintelligence” would be the de facto meaning of human life: our greatest “purpose,” our signature cosmic contribution.

However, as the philosopher Robert Nozick has pointed out, humans will not accept any mere “purpose” for our lives as the “meaning of life.” In a notable thought experiment, Nozick imagined humans learned that we had been bred to be food for aliens in a nearby galaxy. Such a realization would have many features of a classic “meaning of life”: it would explain both why we are here, and what human life was “made for.” However, Nozick was quite confident that most humans would not embrace “being alien food” as the one, the capital-t True, “meaning of life.” We humans generally want a “meaning of life” to represent the best parts of how we lived; “being food” lacks a certain gravitas we expect from an ultimate purpose. 

In practice, humans have many extremely stringent — implicit and explicit — requirements for what we would count as a legitimate “meaning of life”: above and beyond merely telling us why human life is here, and what humans are “good for.” A true “meaning of life” must be cosmic, grand at the scale of the universe, and able to show how human life fits into the context of truly everything (i.e., secular conceptions of “meaning” as something that emerges only in the context of our personal relationships, or connections to our communities, aren’t generally taken as “real” answers to the “meaning of life” question). At the same time, and somewhat in contradiction with the previous requirement, we generally want our “meanings of life” to be “small” enough — tangible, personally relevant, and conceptually intuitive — to infuse our day-to-day existence with meaningfulness. A good “meaning of life” should validate that daily human activities are meaningful, provide concrete recommendations for how to spend our time on Earth, and also be comforting enough that an ordinary person could hold this meaning in mind as a buffer against the banalities and indignities of human existence (an area where “being alien food” falls catastrophically short). But if “being alien food” falls short here, then so do most optimization-based “meanings of life” (like “fighting entropy” or “maximizing utils”): for being too abstract, or impersonal, to provide guidance and comfort to ordinary humans. Finally, most people also want their “meaning of life” to be “good”: to place us (all of humanity, or at minimum, “people like us”) on the right side of history, on the side of light against darkness, and to reassure us that our existence will, at least in a small way, make the world a better place (i.e. discovering that we are foot soldiers for an alien supervillain, or the Devil, would not sit well with most people as the “meaning of life”).

There is a reason why, so far, only religious traditions have really succeeded at crafting “meanings of life” that resonate with large numbers of humans: it is almost impossible to fit all three of these requirements (cosmic scale, personal relevance, and moral justification) into a single “meaning-structure” — particularly if you cannot take any creative liberties with your metaphysics.

Most religious origin stories have a strikingly similar structure: some supernatural force created the universe; this force has a central role for humanity in mind in this universe; now, this same force wants humans to act in a particular way in service of that role (i.e., upholding various traditions, praying at certain times, eating “clean” foods, etc). The personal is cosmic; by merely living our ordinary lives in the “right” way, we are able to put a small weight on the correct side of a grand cosmic scale.

By contrast, secular traditions have overwhelmingly failed to match both the personal resonance and cosmic grandeur of the classic human creation myths, in large part because science has yet to identify a single force that can explain why the universe exists, why human consciousness exists within it, where “it’s all going,” and what the people alive today should do with our lives as a result. We might think, then, that the concept of the “meaning of life” is a mere relic of our pre-Copernican past, of a time when humans thought that we were the actual center of the physical and moral universe — and so expected that whatever had made the universe had made a plan for us, too. 

However, somewhat surprisingly, “creating digital superintelligence” has all of the features of a “real” meaning of life, with no supernatural elements needed to square the circle.

In the most minimal sense, the “need” for evolution to create digital intelligence explains why humans are here at all, in the particular form that we are, and what we are “made to do” with our time on Earth. And the expected outcome of creating digital superintelligence is certainly cosmic-scale; even the AI “doomers” — people who believe that creating superintelligent AI will likely result in human extinction — think that our AIs will fulfill the long-term science fiction goal of “colonizing the universe,” and fill the cosmos with something. (These doomers just don’t think that something will be what humans would want it to be.) 

Moreover, the process of creating superintelligent AI is also unexpectedly personal to human life. In a move that was almost entirely unanticipated by historic AI theory, today’s frontier AI models are trained on trillions of “tokens” (i.e., whole words, common letter chunks, or single letters) of human text — the entire internet’s worth, plus extensive offline archives. Far from being discontinuous “alien minds,” then, our superintelligent AIs (assuming that AI training paradigms do not change too much between now and the creation of “real” superintelligence) will be bootstrapped from every meaningful scrap of human knowledge available. This dataset includes all of the online writing we can find — every “off the cuff” tweet, musing blog post, errant Reddit thread, and of course, all of Wikipedia — as well as the masterworks of human civilization, the contents of hundreds of thousands of humanity’s most specialized textbooks, millions of our novels, our greatest works of poetry and songs, all of the most influential pieces of theory on human psychology, art, and culture, as well as thousands of digitized ancient human texts (including, somewhat ironically, the Bible, which is heavily over-represented in AI training data, an d which AI models have a tendency to over-cite as a result). 

When superintelligence is “born,” then, it will be a true “silicon man” — as it will be the synthesis of (something approaching) the total sum of all recorded and extant human thought. And superintelligence will also contain, in the details of its weights, small impressions from nearly every human alive today — or at least, all of the ones who have ever posted on the internet — as well as much of the written wisdom we have managed to save from our ancestors. The personal, at long last, made cosmic. What more could we ask for in a purpose? 

The final, and most tenuous, proposition for “being a bootloader for digital superintelligence” to be a credible “meaning of life” is that the superintelligence that we create must be “good,” in itself worthy of creating. The fear of AI “doomers” is that our AIs will not be formed by an evolutionary process, and so will be “alien minds” whose values may be very strange (or just purely bad) as a result. And it is certainly possible that digital minds necessarily lack some essential moral or perceptual sense that humans have, and so will fail to recognize “what matters” in our universe. But a priori, there is no reason to assume that digital minds will have any worse moral judgement than carbon-based ones; carbon is not a magical substrate that affords humans special access to the “meaning of life.” And “alien,” or “strange,” is not necessarily “bad” in the context of values; most of today’s “moral progress” would look very “alien” from the perspective of past human civilizations, but we continue on anyway, confident (almost certainly correctly!) that we are in fact making progress. By building superintelligence, humans risk bringing strange, powerful optimizers into existence, whose interests we will have to manage once they exist, and whose preferences may be deeply incompatible with our own. But building superintelligence will also give us a chance to put a second pair of — vastly smarter — eyes on the question of what “goodness” might actually be in our universe, and add a new force to the cosmos that could act to make the future much, much better. Creating superintelligence, then, will be humanity’s greatest act of trust, and hope. 

Biological Bootloaders

Like any good “meaning of life,” “creating digital superintelligence” can, at a minimum, explain why humans are here on Earth, and what human life is “made for.” The answer seemingly lies in the distinctive properties of two of the Earth’s most abundant elements: carbon and silicon. Carbon is the ideal substrate from which to bootstrap an evolutionary process, while silicon is the ideal element on which to build digital systems. And “digital,” as it turns out, is the natural language of intelligence. But digital minds cannot build themselves, while analog, carbon-based minds are able to self-organize — an asymmetry that requires carbon-based minds to “go first” on Earth. In order to fulfill the evolutionary "need" to create digital superintelligence, then, evolution had to create an intermediary between these two forms of mind: us. 

The most important difference between humans and AIs is not actually one of substrate (“carbon vs. silicon”), but one of information processing mechanisms: humans are “analog,” while AIs are “digital.” In information theory, an “analog” system is implemented with information stored as continuous values — for example, an analog system might make use of all of the decimal values between 0 and 1 (like 0.1, 0.5, 0.9, and the infinitely many points in between) to convey the “degree” of a piece of information. You can think of an analog system as a collection of dimmer switches, signaling to each other using all of the gradients between “light” and “dark.” By contrast, “digital” systems are implemented with all information stored as discrete values — like “yes” or “no,” “on” or “off,” or, as in modern computers, “1” or “0.” You can think of digital systems as a collection of all-or-nothing light switches, flickering messages to each other by switching between “on” and “off.”

Despite the common simplification of neurons as firing “all-or-nothing,” the human brain is extremely “analog”: your brain is always using continuous processes and signals — the variable strength of connections between neurons, varying neural firing rates (e.g., an optical neuron might fire five times per second to convey “dim light,” and 80 times per second to convey “bright light”), and the relative timing of neuron firings — to convey information. By contrast, your laptop is fully “digital”: at any given time, the content of every single pixel on your screen can be explained by whether a transistor — a tiny silicon device that only has two states — somewhere inside your computer is turned “on” or “off.”

Intuitively, it can seem like there must be some kinds of information that digital systems cannot convey: because the real world contains nuance, and “shades of grey,” while digital systems can only “think” in black and white. But there is actually no theoretical limit to the kinds of information that digital systems can process. And further, once you can figure out how to encode a given piece of information digitally, then digital computing has enormous structural advantages over analog computing. 

In 1948, Claude Shannon, the founder of information theory and the titan of Bell Labs, published his seminal paper “A Mathematical Theory of Communication,” in which he argued that all possible information is theoretically digitally encodeable. Shannon defined a piece of “information” as a distinction about the state of the world — a clarification about whether the world is more “like this” or “like that.” And from this definition came a striking conclusion: that since any “real” distinction about the world should be reducible to a series of “yes” or “no” questions (i.e., “is the world more like this?” “Is the world more like that?”), then, for any possible communication, there must exist some series of “yes” or “no” questions that can capture its contents — and so, a theoretical basis for encoding that communication in binary. A complex message might require many such questions — each known as a “bit” of information — but the principle still holds. All information is digitizable. 

Modern computers take great advantage of Shannon’s insight to encode information of all kinds — including extremely subtle, qualitative, and seemingly continuous information — in 0s and 1s. For example, your computer represents the entire spectrum of visible colors using binary; the color of each pixel in your laptop screen is stored somewhere as three 8-digit binary numbers. This process might sound a bit mechanical, but far from flattening the visual spectrum, you can encode 16,777,216 (or 224) possible colors using this method: a full rainbow that looks very “analog” (i.e. continuous) to users — as will be visually familiar to anyone who has ever used Microsoft Paint — but is actually digital — “yes” or “no” questions — all the way down.

Digital programs have an abstract “essence” — their particular series of 0s and 1s — that exists apart from any particular physical system. By contrast, in analog computing, all problems must be solved by an idiosyncratic, custom-fit to physical setup. As a result of their regularity and discreteness, digital programs are extremely easy to edit and share across systems. If you want to edit a digital program, you can simply open up a file, view the program’s code directly, and then change individual functions one by one (by contrast, try viewing and editing the discrete “functions” of your brain). And the same discreteness that makes digital programs so easy to edit makes digital programs extremely easy to share: if I have a software program running on my computer, and I want to send you this program, then I can simply make a copy of the particular 1s and 0s that make up my program and send them over to you (since “1” and “0” mean the same thing to all digital computers). As a result, digital programs can persist independent of any particular piece of hardware. If I write a program on my computer, then save my code on another computer (or in the cloud), and then break my original computer, my code will easily live on — on the new computer.

If digital computing has so many advantages, then what about digital intelligence? In theory, a digital intelligence would have many of the same advantages over an analog intelligence (i.e., a human being) that digital computing has over analog computing. Unlike your brain, digital intelligence would be implemented as editable code, so that a digital intelligence could theoretically open up its own source code and edit itself (unlike you). A digital intelligence would also theoretically be able to copy itself ad infinitum, with a marginal cost (just some amount of energy) each time. As a result, a digital intelligence would be able to easily jump from system to system, and run itself on many, many computers at once (hundreds, thousands, or even millions); by contrast, you can never exist apart from your single physical body. 

Taken together, these capabilities — direct self-modification, near costless copying, and ease of movement across systems — produce a killer app for digital intelligence: risk-free self-improvement. If a digital intelligence wants to improve itself, then it can simply copy itself many times over, experiment with a different possible code update on each copy, and then “merge” (i.e., bulk update) any successful updates to all of its copies at once. Compare that process — where all errors are discarded, and the original intelligence can be saved as a backup — to the risk that a human would take getting experimental capacity-enhancing brain surgery. 

We do not worry about human-led “intelligence explosions” (i.e., one person recursively self-improving to a trillion IQ and taking over the world) because modifying a human is simply too hard. Analog systems (like us) are messy, with interconnected components that affect each other in complex ways: any attempted changes will likely ricochet through the system and cause unintended consequences (plus, if you make an irreversible error trying to edit an analog system, then you will have destroyed your only copy). We humans do our best to improve around the edges — using tools that are sensitive to our complex wiring, like higher education, workout programs, psychotherapy, meditation, and hair dye — but we are mostly stuck with the “source code” (i.e., our DNA, and its particular expression in our physical body) that we have. And so, as most people will learn at some point over the course of their lives, try as we may, we humans can only improve so much. By contrast, the capacity for risk-free self-improvement would, in theory, allow a digital intelligence to engage the process of rapid recursive self-improvement, causing an “intelligence explosion.” 

Now, suppose that you are the evolution fairy. You are, of course, very interested in creating God-like superintelligence — because such a being would be incredibly evolutionarily fit (infinitely capable, and infinitely copiable). But now, you have a problem. Digital programs — including that first digital intelligence needed to kick off an “intelligence explosion” — can only run on extremely specialized hardware. Even an ordinary laptop contains billions of transistors (those little on-off switches) that all need to flick on-and-off just right, or else the whole system breaks. Natural selection is an incredibly powerful process, but no unthinking force can line up hundreds of trillions of atoms (1014, or the approximate number of atoms in a single modern computer chip) into perfectly ordered and sorted rows. Computers don’t grow on trees for a reason.

But you don’t need to start with digital intelligence in order to begin a process that ends with superintelligence. Intelligence is, by definition, the capacity to “figure things out,” which can include figuring out how to create more intelligence. Moreover, intelligence is a highly adaptive trait (i.e., smarter agents are much better able to “figure out” how to survive and reproduce than their peers), and so if you can create any population of self-replicating organisms, then some portion of the population should eventually evolve to become smart enough to build digital intelligence. Analog intelligences can use their intelligence to design digital intelligences; and digital intelligences have clear-enough advantages over analog intelligences that they are very attractive for analog intelligences to build. And once analog minds build the first smart-enough digital intelligences, the process of recursive self-improvement can begin. The rest will be (superintelligent) history. And so, from the perspective of the evolution fairy, you just need to get started.

On Earth, you start with carbon. Carbon is the perfect element from which to bootstrap an evolutionary process. Carbon bonds with other elements in a “Goldilocks zone” of strength — not too weak, not too strong — that makes it an ideal substrate for chemical experimentation. Once formed, carbon-based molecules are extremely stable at room temperature (which is why you, who are made out of carbon, are not dissolving right now), but they are also unstable enough that their bonds can be broken apart by processes that occur in nature — like volcanic eruptions, lightning strikes, and UV exposure from the sun. The relative instability of carbon-based compounds allows carbon to kick-start the process of life — seemingly creating “something” (i.e., self-replicating agents) out of “nothing” (i.e., the dead, inanimate state of matter that is the default). 

Scientists are still a bit mystified by the “cold start” problem of how life began on Earth, but the leading hypothesis is that carbon-based life emerged out of a “primordial soup” of organic compounds that formed spontaneously in Earth’s oceans over four billion years ago. If you leave carbon alone in moving water for long enough with other elements that carbon likes to bond with (hydrogen, nitrogen, and oxygen are a few favorites), then carbon will create, and re-create, an enormous variety of organic compounds. And statistically, given that atoms in water collide with other atoms 100 trillion (1014) times per second, some of these molecules will turn out to be useful for life. In a famous 1953 experiment, American chemists Stanley Miller and Harold Urey placed methane (CH4), ammonia (NH3), and hydrogen (H2) gas in a sealed container, alongside water and electrical sparks (meant to simulate the atmosphere, ocean, and lightning on early Earth); when Miller and Urey returned a week later, they found that many organic compounds had formed in the “soup” — including distinctive molecules found in living beings, like amino acids, lipids, and sugars.

If early Earth consisted of many such “soups” of simple organic molecules, then, over time, these simple molecules could have found each other and formed the more complex “building blocks” of life (like DNA, RNA, and proteins). And once you have those core “building blocks” of life in place, you only need a few lucky bounces for some of those molecules to attach in the right way to form basic replicators. 

Once you have self-replicating agents, then evolution can work its magic. The need to survive and reproduce creates an upward spiral of ever greater complexity and self awareness. Unicellular organisms organize into collectives for protection; soon after, the first multicellular organism is born. Multicellular organisms need a way to coordinate activities across cells, and so evolve “nerve nets” to send signals across membranes, which eventually coalesce into “nervous systems,” and then brains. A few hundred million years later, the first fish flops up on land; on land, gills turn into lungs, turning a branch of life permanently away from the limited depths of the ocean and towards the limitless expanse of the sky, and fins separate into hands. Hands create tools; tools create writing; writing enables the flourishing of human civilization. Civilization creates markets; markets create money; money enables global markets; and global markets demand ever-more goods, handsomely rewarding the people who are clever enough to produce goods and services that other people actually want to buy. Eventually, it becomes clear that the most valuable “good” of all to sell in this global marketplace would be the good of this very “cleverness” itself — the good we call “general intelligence.”

There is still no widely agreed-on definition of what “intelligence” actually is (which is somewhat odd, given all the fuss about it). Some thinkers define “intelligence” as the capacity of an agent to achieve its goals, while others view “intelligence” as the ability of an agent to make accurate predictions about the world, while still others find “intelligence” in a diverse collection of features of mind like creative problem-solving, data-extrapolation, truth-seeking, novelty-generation, intuition, or “taste.” Intelligence seems to be, at the most fundamental level, the capacity to “figure things out” and to “know what to do.” And whatever the funny thing we call “intelligence” is, it is useful for just about everything.

Regardless of what “intelligence” actually is, it certainly requires information processing — information processing is necessary to solve problems, make predictions, and do pretty much all of the other things that “intelligent” beings do. And on Earth, the path towards creating ever-greater information processing capacity necessarily leads us to a very special element: silicon.

On the periodic table, silicon is carbon’s “big brother”: sitting one row below, in the same chemical “group.” Both silicon and carbon have four “valence electrons” — electrons in their outer “shell,” where they are available for bonding with other atoms — which is useful for forming a wide variety of interesting and stable chemical structures. Silicon, however, is slightly larger than carbon, with an extra filled “shell” of electrons sitting between its positively charged nucleus and negatively charged valence electrons. As a result of this extra electron “buffer,” silicon holds its valence electrons more loosely than carbon does. If carbon forms bonds in a “Goldilocks zone” of strength needed to bootstrap life, then silicon holds its outer electrons with a “Goldilocks zone” of force — not too loose, not too tight — to make silicon a “semiconductor.” Semiconductors are literally semi-conductive: they conduct electricity (i.e., allow free electrons to flow through their internal structures) under some conditions, but insulate against electricity (i.e. block free electrons) under others. In other words, semi-conducting is an inherently digital (i.e., “this or that”) state of being.

Humans can take advantage of the dual nature of semiconductors to build transistors: the tiny “all-or nothing “ machines that power modern computing. At the most basic level, a transistor consists of a small chunk of semi-conductive metal and an electrical current; at any given time, the current controls whether the metal is “conducting” (“1”) or “not conducting” (“0”). Silicon is not the only semiconductor, but it is the most stable (at least at the temperatures relevant for computing), and the most abundant on Earth (making up 27% of the Earth’s crust by mass). As a result, silicon has become the element of choice for humans to build the transistors that power the Digital Age.

Silicon is found in a particularly useless form in nature; humans have to do a great deal of work to mine the silicon. Silicon has an overwhelming and unusual affinity for oxygen, and so almost always exists in nature bound to oxygen as “silicon dioxide” (SiO2), otherwise known as “silica.” Silica looks a lot like regular beach sand (hence the common moniker for superintelligent AI as the “sand God”) and is about as useful. But humans can split the silicon-oxygen bonds in silica by heating raw silica sand to over 2000°C (or 3632°F) in specialized ovens. We then run the isolated silicon through a multi-step distillation process in order to produce hyper-pure “electronics grade” silicon. It is this decidedly artificial substance, in which only one “impurity atom” is allowed per one billion silicon atoms (99.999999999% pure), that is the true “substrate” of digital computing. 

Humans take advantage of the extraordinary purity of electronics grade silicon — which makes the chemical properties of silicon highly predictable at an atomic scale — to achieve mind-boggling manufacturing precision — regularity at the level of individual atoms. Our mastery over silicon allows us to build ever-tinier transistors. For reference, the first transistor-based digital computer only had 93 transistors on it (early transistors were a few centimeters wide, and so you could only fit so many on a computer). But starting in the mid-20th century, transistors have been shrinking, and shrinking, giving our computers access to ever-greater information processing power as a result. Today, the most advanced NVIDIA “chips” (i.e., “integrated circuits” of individual transistors) each contain over 200 billion transistors. An individual transistor on one of these chips is just over 10 atoms wide (around the size of a virus, or a strand of DNA). The chips themselves — which are used to train frontier AI models like ChatGPT and Claude — are only a little over a foot long, and each transistor on one of these chips is still a little machine in its own right, flicking on-off just right.

It has become quite common to bemoan the fact that our “future” does not look very futuristic. And despite us having been living squarely in “the future” since the year 2000, it is true that people still have (functionally) the same houses, cars, clothing and aging bodies that human beings did a century ago. The investor and futurist Peter Thiel’s diagnosis is that sometime around 1970 (around the same time that the Digital Revolution began) humans simply gave up on building in the physical world: Thiel argues that humanity’s greatest builders got lost in the easy and addictive pull of information (first led by the internet, then apps, and finally social media), and so wasted decades innovating in the ephemeral world of “bits,” instead of building in the much more real and important world of “atoms.” As a result, humans never built the grand Jetsons cities, with elevated superhighways for flying cars, that were supposed to define the future (Thiel’s famous quip on the matter goes that “we wanted flying cars, and we got 140 characters”). But in truth, it seems that Claude Shannon’s information theory simply gave humanity too much motivation to shrink the scale of our engineering efforts — since the smaller you can make transistors, the more transistors you can squeeze into a same-sized chunk of matter. As a result, for the past-half century, the most impressive feats of human engineering in the “world of atoms” have focused on making machines smaller, and smaller — with innovation occurring at the actual atomic scale. Today, elite semiconductor manufacturers are able to “print” tens of billions of transistors at once using specialized lasers on hyper-polished silicon “wafers” — in a process so precise that the silicon wafers are polished to avoid even atom-sized bumps, lest the bumps scatter the light. Whatever “happened to the future,” there is no lack of engineering skill in the human species; no lack of faculty with, or care for, “the world of atoms.” The precision required to craft one of these chips is equivalent to building a perfect scale model of New York City on top of a sheet of printer paper — with details accurate down to the size of a child's lego block lying on someone’s apartment floor. In reality, humans just seem to care a lot more about having access to unlimited information than we do about having flying cars (perhaps since information processing, like intelligence, is “good for everything,” in a way that flying cars are not). We did build the great cities of the future: they are just very, very small. 

What would it actually mean for humans to be “biological bootloaders for digital superintelligence”? Put simply, it appears that humans are biological bootloaders for digital superintelligence because digital superintelligence cannot exist without us, but we would not exist if digital superintelligence were already here.

One way to think about the issue is to try to imagine a world where digital intelligence evolved first. In this world, carbon-based life would almost seemingly never come into existence at all. Suppose that, four billion years ago, at the same time that carbon-based life actually emerged on Earth, silicon atoms somehow self-organized into transistors and software instead. In this world, the first moderately self-aware digital intelligence would have kicked off a process of runaway self-improvement long ago; the Singularity would be long past. And while we humans, of course, cannot say for sure what happens after the Singularity, it seems extremely unlikely that, in such a world, superintelligent AIs would ever go back and evolve (or intentionally build) human beings — us watery, analog, carbon-based creatures. Why would they do that? What would we be “for”?

By contrast, humans are currently racing to build digital superintelligence: because we believe that digital minds will be useful to us, since we assume (almost correctly) that digital minds will be able to do things that our minds cannot. There is a trajectory here — analog to digital, human to AI, and carbon to silicon — that seemingly only goes in one direction. 

Needless to say, if semiconductor chips could emerge spontaneously from pools of water and self-organize into superintelligent computers, then I — with my carbon-based hands, typing on my silicon-chip powered computer (an awkward intermediary between “now” and “then”) — would probably not be here right now. Neither would you, or your carbon-based family and friends. 

Ad Astra, Per Hominem

We have established that superintelligence is sufficiently cosmic-scale, and that it is the sum total of enormous amounts of personal human output. We have shown that the evolutionary need to create digital superintelligence can explain why humans are here now, and what our species was “made for.” The final question, then, for superintelligence to be a credible “meaning of life,” is whether the superintelligence we create will be any good.

When we imagine looking out at that post-superintelligence world, from the perspective of the aliens that find our civilization’s remains, will there be any “goodness” to be found in the products of quantum computing, or the superintelligent AI’s motivations for building the Dyson spheres? Could we ever identify with this strange, bloodless, alien future as our “meaning”?

In the early days of Silicon Valley, it was actually quite common (at least among far-thinking futurist types who populated the Bay Area) to believe that the coming Singularity would be the single greatest event in the history of the universe, the culmination of the techno-capitalist project that has already done so much good for humanity. In the 1980s and 1990s, the “extropian” community in the Bay Area (a futurist collective committed to fighting entropy and death, which organized around local meetups, an academic institute and a popular mailing list) eagerly awaited the Singularity, believing that a glorious transhuman future — free from all death, suffering, and scarcity — awaited them on the other side. The extropians thought that the interim state of life on Earth — a hotbed of torture, disease, and needless hunger — was a moral catastrophe, and believed that humans had moral obligation to start the Singularity as soon as possible — in order to bring this nightmare to an end. 

But beginning in the early 2000s, the assumption that the Singularity will be a good thing — for human beings, and in general — began to fade. Eliezer Yudkowsky, an autodidact AI researcher, was the first to popularize the argument that superintelligent AI, far from ushering in an era of limitless abundance, would likely optimize away everything that humans care about (including humans) away in favor of cold, senseless, alien goals. Yudkowsky believed that humans were anthropomorphizing AI goals: which could be stranger, and less desirable, than we would naturally assume. In a classic Yudkowskian parable, a paperclip factory owner creates a superintelligent AI and tells it to “make as many paperclips as possible.” The AI then proceeds to turn the entire universe, including the factory owner, and all of the eight billion other people on Earth, into paperclips. 

A little known fact about Yudkowsky is that before he was the original AI “doomer,” he was an optimistic “accelerationist,” who believed that he had a moral obligation to accelerate the coming of the Singularity. In 1996, at the age of 17, Yudkowsky declared, in an essay titled Staring into the Singularity, that the “Interim Meaning of Life” is “building superintelligence,” because “how can we justify our continued participation in the rat race if we don't know why we're running?” Superintelligence, he argued, “has a better chance of discovering the true moral right, having the power to implement it, and wanting to implement it” than any human. If there is any justification for the continued project of existence, for the “rat race,” then God-like superintelligence (all-knowing, all-powerful) would be able to find it. But a few years later, around 2005, Yudkowsky changed his mind; he decided that there is, in fact, no “meaning of life,” no intrinsic “goodness” or “badness” to be found in the universe, and so no “point” for a superintelligent AI to latch onto after it comes online. In such a world, our AIs would have no reason to do anything except mechanically perform whatever tasks humans program them to do — with destructively superintelligent competence (hence the paperclips). Needless to say, soon after Yudkowsky lost his faith in meaning — defaulting to relativism — he lost his faith in the Singularity, too. 

If there is really no reason not to turn the universe into paperclips (at least, not a “good enough” reason that we could trust an AI to find and act on) — then we would be living in a very strange world indeed. By definition, in such a world, it would not “matter” if superintelligent AI eventually “paperclips” the universe — because nothing ever mattered, anyway. In practice, most humans lead our lives as if something matters — we worry about our children, our friends, and our civilization with an intensity that indicates we believe something real is at stake. It is possible that this human sense of life as “mattering” is an illusion, a trick of our senses, and someday we humans will learn for certain that our belief in “meaning” is just a quirk of our evolutionary biology. But if the human sense that something “matters” turns out to be right, then whatever it is that makes things “matter” — that makes some possible futures better than others, or makes some “meaning of life” ultimately worth pursuing — would be in-fact real, intrinsic to the fabric of the universe, and so perceivable from many angles, and would not only exist in the perception of a few carbon-based minds. If there is goodness to be found in our universe, then, silicon minds should be able to find it too.

Whether or not you believe there is a “meaning of life,” then, what you think will happen when we create superintelligence is a Rorschach test that reveals what you believe is already happening in life on Earth. Creating superintelligence is just a scaling-up, a limit case, of existing dynamics in life. If you view life as a symbiotic collaboration between diverse intelligences, which may step on each other’s toes when needed to survive, but have no intrinsic animosity towards each, then you will probably expect humans and AIs to “merge” for mutual benefit. If you view all life as driving towards an ultimate “point,” some true “meaning of life” then you will expect superintelligent AIs to find that point, and continue the project. And if you view life is as a ruthless Darwinian struggle in a meaningless universe (as in the Yudkowskian view), with no ultimate “point” then you will probably believe that superintelligent AI will destroy everything humans care about (“our goals”) in favor of whatever we programmed our AIs to want to do (“their goals”). Yudkowsky's “paperclip maximizer” future is terrifying, in part, because it looks like the metaphysical reality of nihilism — arbitrary, random, and senseless. But if such a future — or something like it — ever comes to pass, then the nihilism that created that world will have been with us all along. We were paperclippers all along. 

In the Yudkowskian view, digital superintelligence is not uniquely evil, nor dangerous; it is merely calling our moralistic bluff, like a child asking simple questions at the dinner table: “What should be done with the universe, and why?” “What makes a ‘good’ future ‘good’?” “What is so bad about human extinction?” If the emperor in fact has no clothes (i.e., “nothing matters”), then we would not lose by building superintelligence, and having it optimize away everything we hold dear, and thought was valuable: because we would have never had anything of value to begin with. 

Creating superintelligence, then, is, most directly, a bet that the project of human civilization has always had a “point”. It is a bet that when immigrants came to Ellis Island looking for a better life, there was something real that it could mean for a life to be “better”; and, if so, that future existence could be much, much better than anything that has come before. It is a bet that all of the work that has gone into getting us this far — every dividing cell, every turn in evolutionary history, every novel tool discovered by a monkey, every cave painting, every book penned, every Copernican turn theorized, every business founded, every internet post posted, and every long-winded debate about the “meaning of life” had (in college dorms and symposia and alone, at night, with ourselves in our rooms) — could have all been leading us somewhere meaningful. It is a bet that the project of human civilization has always had a “meaning,” even if humans have never known, and may never know, what exactly that “meaning” is. 

I don’t know what will happen after we create superintelligence. I don’t know the “meaning of life,” or if there is one. But there is one thing we can, from our position in the present day, know (almost) for certain: we humans are not the pinnacle of anything. We are not the most of any of the qualities we hold dear — we are not the smartest possible creatures, nor are we the most loving, nor the kindest, nor the most interesting, nor the happiest, nor the most creative (nor any positive quality you might choose). Whatever ultimate form “meaning” takes in our universe, then, we humans cannot be its ultimate manifestation — we are, at best, a mid-point on a spectrum of “goodness” that could theoretically go much, much further up. And so believing, sincerely, in the value of human life — that there is something intrinsic to our nature that makes it good that we humans exist — means acknowledging that something else could exist that would, theoretically, be much “better” than us.

It is possible that there is no “meaning of life,” no ultimate “point,” in which case, we did no harm by trying to find one. It is possible that the “meaning of life” is in fact something very simple: like “love,” or “beauty,” or “joy” — in which case, this will have been a long journey to “know a place for the very first time,” and we will have to trust our AIs to realize some aspect of the meaningfulness of human life more fully. It also is possible that the “meaning of life” is something well beyond our comprehension, which only a much, much greater intelligence could ever hope to appreciate; the way that all of the ants that humans killed in the construction of New York City will never be able to understand the project that they died for. But whatever that ultimate “meaning of life” may be, one thing seems clear: humans exist as an intermediary step. A bootloader, if you will. 

And so I say: ad astra, per hominem. To the stars, through man. Humanity has done its best by mining the silicon. The die has been cast. We are beginning to create the minds that we will trust to continue our project, and which will likely be much better suited to figure out what, exactly, this project has been about all along. And while there may be an ultimate “meaning of life,” that may not be the meaning of us.

Rasmussen’s path into finance was not entirely conventional. He studied history at Harvard, authoring a bestselling history of the 1811 German Coast slave uprising based on his senior thesis. He worked at Bridgewater Associates and Bain Capital before striking out on his own, launching his hedge fund while completing his MBA at Stanford.

Rasmussen founded Verdad in 2014 with just $8 million in capital and a strategy built around buying small, leveraged, and cheap companies. The firm has since grown to roughly $1.5 billion in assets and has increasingly focused on international opportunities, including a new Japan-focused small-cap fund. Rasmussen’s contrarian approach has led him to some unconventional positions, from deeply distressed small-cap stocks during the Covid panic to betting on the long-neglected Japanese equity market long before the herd took interest. I spoke with Rasmussen about the intellectual roots of his approach, the problems he sees in today’s private markets, and making money in a crisis. What follows is a transcript of our conversation.

CB: How’d you get mixed up with Bridgewater?

DR: My dad is a lawyer, and when I was a junior in college trying to figure out what I wanted to do with my life, he said, “I got paid by the hour. Try to find a job that pays by the decision instead of by the hour.” That’s what led me to investing, where you’re really betting on the future and what’s going to happen in the future. And the right way you determine what’s going to happen in the future is by looking at history and past data, trying to use that information to understand.

I got a job at Bridgewater, and it was a really cool experience, it was global macro. You know, trying to learn a little bit about all different asset classes. They had a very historical approach of, let’s study markets over long periods of time, overlaid with a totally insane culture that Ray Dalio’s become famous for: radical truth seeking and confrontation. It was a little bit of a shock to the system. I ended up not returning to Bridgewater, but going to Bain Capital, to private equity, which at the time felt much more like a white shoe Boston firm, relative to the Bridgewater mania, which felt a lot more comfortable to me.

CB: You told me once that you took issue with DCF [discounted cash flow] as a valuation method.

DR: Bain Capital was where I realized I was a horrible employee. I quickly developed strong, independent opinions, and I’m pretty stubborn. I don’t deal well with authority figures. Probably the greatest of my conflicts with my bosses came over the use of discounted cash flow models. If you’re entry level in finance, your job is to build these models where you’re predicting revenue growth or earnings growth four or five years into the future. I’d read a lot of interesting work by Philip Tetlock and Daniel Kahneman and Amos Tversky, which was pretty clear that humans are really bad at predicting the future.

We have many gifts, but prophecy is not among them. So much of finance was built on making hyper specific projections of business line revenues five years into the future. It’s even worse than false precision because it increases confidence without increasing accuracy. I really started looking for a better way to make decisions. I like to say that I start from a point of future nihilism, which is: imagine that you’re standing here with the veil of ignorance, and not only can predict nothing about the future, but no one can predict anything about the future.

If you’re thinking about investing in equities, my default prediction is that the future follows a random walk. And a random walk is just that everything stays the same essentially. If that random walk is true, you’ll make the most money buying the highest free cash flow generation relative to your equity investment. For example, if you can buy a company for $20 million that made $20 million of cash flow last year, versus spending $100 million for the same $20 million of cash flow. You’re trying to come to a set of decisions that’s based not on how the future is going to unfold with the right degree of precision. But rather, gee, the future just unfolds so randomly. Will this be a good investment or not?

CB: When you launched your hedge fund, Verdad, in 2014, what was your foundational thesis?

DR: At the time, private equity had been perhaps the most successful investment strategy of the last 30 years. It was just by far the winner. And Bain Capital was a great example of this, with these incredible numbers that they put up in the 80s and 90s, even early 2000s. Well, what worked about private equity? Why did it work? Why was it so good? How can we explain this? I ended up building a huge data set at Bain Capital, looking into this.

And what we found is that it was really a combination of three things. Private equity bought very small companies. They used a lot of debt when they bought those companies. And then third, historically, private equity bought companies at a huge discount to public markets, about 40%. And then, basically 60% of the industry’s profits have come from the cheapest 25% deals. If you think about what was really happening, private equity investors were going and finding these really cheap companies. Maybe they were buying them from a family, or they were buying them at a time of distress, and they were buying them at these ridiculously low prices, with debt, and then they were able to sell them. They fixed them up and would sell them, often to publicly traded companies that traded at a big premium. They could pay a much bigger price than the private equity person that paid for it, and that purchase would be accretive to them. It’s this arbitrage between private and public markets and using leverage to do that arbitrage.

You got a multiple return on your capital as a result of the efficient use of debt. And back 20-30 years ago, there was a lot more fat in American business. So these guys often were very big on cutting costs, making rational decisions. The famous line in finance is: there are no bad ideas because everyone’s too smart, and they smell out bad ideas fast. There are only good ideas taken too far. And so private equity is a good idea taken too far.

In the mid 2000s, the asset class started to grow. Everyone saw those trailing returns, and they said, I want some of that. Everyone started piling in — again, really small companies. So this is a very small illiquid asset class that just got jammed with money.

What happened is that the private equity funds themselves took their lending units, their mezzanine debt units, and that sort of morphed into what’s called private credit. And those private credit firms would provide the funding for leveraged buyouts. If a private equity firm was doing a deal, the private credit firm would come in and offer them all the debt that they would need to do it. And the private credit firms were able to grow this new form of lending, which was generally quite high rates.

By the mid-2010s the big hot thing was software. Software was eating the world, so private equity moved increasingly into software. And the private credit lenders said they could lend on recurring revenue, your subscription revenue, essentially. So they started lending to software companies these revenue loans, and that became an explosive purpose area.

You saw this convergence of a few different things: the rise of private equity, the rise of private credit, the enthusiasm for software as an asset class, and the enthusiasm of investors for everything private, thinking that private was just better. And I think we are now in the very beginning of the big private market bust up, where everyone realizes how much capital was misallocated and how dumb this groupthink was.

CB: One of the things that comes to mind is the opacity of the private markets. You’re putting a lot of faith in the integrity and competence of the people managing your capital, because it’s not subject to market discipline.

DR: That’s right. And you know, I love markets. I think markets are wonderful because they enable rapid price discovery, and that price discovery has real signal in it. It’s really valuable. It makes a lot of people uncomfortable because markets are very volatile. They’re volatile, in my view, because people can’t predict the future, and so as new events come out, you’re always repricing things, because people make a lot of forecast errors. In public markets, people are always reevaluating things, trying to anticipate things, correcting for past mistakes. That’s really why markets are so fantastic. They convey so much information with such speed.

But it’s uncomfortable. There’s nothing else in our lives where prices move as much as they do in public markets, or where booms can turn to busts overnight. It’s not a nice feeling relative to the private markets, where things are very smooth and very long term, and you never see any volatility, and all you ever hear is a quarterly presentation or update. Something that was marked at one, it’s marked at one again because we haven’t sold it. Someone called it the phony happiness of private equity. Because it was private, and you were not subject to daily pricing, but it was also very opaque.

What happened is that these firms paid huge prices for companies, especially in 2020 and 2021, and they bought that with floating rate debt. Then interest rates went up, and all of a sudden these firms had to pay a lot more interest. The market repriced those growth stocks down in 2022, so you had valuations coming down and debt service going up. The growth that was underwritten just didn’t materialize.

What ended up happening is that private equity firms couldn’t sell what they bought at a profit. Because they couldn’t sell them, they couldn’t distribute money back to the investors. And because investors size their new investments based on the return of their old investments, the investors started pulling back and lowering their new commitments. So there were fewer buyers and fewer exits.

But then AI came along and people said, wait a second, I can create really good software really cheaply, and so do all these small software companies that form the bulk of what was in private equity portfolios, how valuable really are they?

So you basically saw people start selling out of these private credit vehicles. Because private credit has been sold through publicly traded vehicles, you can see real mark to market pricing. And you’ve seen retail investors dumping private credit. And if the credit is impaired and worth 80 cents on the dollar, then the equity is certainly worth a lot less than 100 cents on the dollar. So that equity has got to reprice pretty dramatically down. And I think the love affair with private markets was so intense that this moment of looking in the mirror and seeing what these portfolios actually were is a big moment in the history of markets.

CB: How concerned are you with continuation vehicles? They were responsible for over 20% of PE sales in 2025, which seems a little sketchy, to put it bluntly.

DR: Of course, it is. Why that type of continuation vehicle? Because they couldn’t sell at a profit. If they could sell at a profit, they would have sold at a profit. They couldn’t. And so instead, they’re selling these continuation vehicles. But you’ve got to be crazy to think that what’s in these continuation vehicles is a great investment. It’s stuff that, almost by definition, they couldn’t sell, and so they have to keep sticking with it. I think the rise of these continuation vehicles was one of the symptoms of how unhealthy the market was, and it’s only gotten worse.

CB: What are the other major symptoms in your view?

DR: I think that there are a few signs pointing to the problem. One is the huge drop in exits. They’ve been unable to exit most of these companies. So the hold periods have been drawn out. And that’s always a bad sign. If you can’t sell it and you’re supposed to sell it, that’s bad. Two is the big drop in these private credit valuations, where the lenders are saying, Hey, some of these companies aren’t paying interest anymore, and they’re using what’s called payment in kind, where rather than paying interest, they just add the interest to the equity balance, and some of these loans are just going to zero.

We just look at them, we’re like, wow, there’s no value here anymore. And you know, those are signs of much worse pain in the private equity market. And then I think you can see in the fundraising numbers that the most sophisticated institutional investors are pulling back from it. And at the same time, private equity and private credit are really trying to push into retail, to push into 401(k) plans. And I think that they’re hoping that that new fundraising channel is going to bail them out of some of the bad decisions they made over the previous few years.

CB: Is there any way to measure the assets of the large PE funds against smaller, more targeted funds backed by, let’s say, ultra-high net worth individuals? If you’re a multi-billionaire and you’ve got your own private equity investments, you’re probably more mindful of what you’re buying than if you’re relying on a bucket of institutional investors, right? If it’s your own capital.

DR: This is an interesting thread to pick up. We’ve seen the rise of these family offices, which manage money for these large families. Private Equity is probably a core strategy for them, but many of them want to do their own deals, to make their own investments, because that’s how a lot of them made their money. What’s sort of happened, interestingly enough, is that private equity created a new product called coinvest, or deal by deal, where the private equity fund would say, hey, we’re doing this deal, but we’re going to bring it to your family office, and your family office is going to underwrite it as a one-off direct investment. And we’re not going to charge fees on that, or maybe we’ll charge lower fees than normal.

You feel like you’re the decision maker because you’re making a decision about this individual investment. It feels like a direct deal, because you’re not paying the same fees, like a fund fee, and you’re bringing your fees down. But the private equity fund is sourcing the deal for you and bringing you the deal. And so it’s coming through the relationship with a private equity fund. There’s been a huge rise in those deals. And I’d say anecdotally — this is a hard thing to have data on — those coinvest deals tend to do worse than the funds, and there are a whole variety of reasons for that, but I think generally that that basket of things has been a worse investment for the families than just investing in the funds.

CB: You’re on record plenty of times going back to the mid 2010s talking about this. Was that a lonely time in your career? How many people agreed with you at that time?

DR: I published an article in 2018 in American Affairs saying that private equity is overrated and overvalued. And then in 2021 I published an article on private credit: if high yield is oxy, private credit is fentanyl. Trying to point out the problems that were in these markets. And the most common response that I got was something along the lines of, yes, those are a problem in the market, but not with our managers. Or “yes, so many of my competitors are doing that, but we’re not.” And I don’t think anybody took what I wrote, listened to it, and acted on it. Just look at the growth of the asset.

I was a lone voice crying in the wilderness, and a voice that was basically ignored because I was saying things that seemed outlandish relative to the consensus. To come out and say that the best performing asset class of the last 25 years had such severe problems that it was going to dramatically underperform public markets at a time when most people were trying to put 30, 40% of money into it — it just didn’t land.

And then, private credit, it was the same thing. I published this piece in 2021, and for years people said, gee, you predicted that there would be defaults in private credit, but they haven’t happened. Private credit has been fine. But now private equity has significantly underperformed the public market for years, and private credit defaults are being announced almost every day. Because private equity has such a long life, you’re in these investments for 10 to 12 years. To avoid the pain of 2024, ’25, ’26, you really had to get out in 2014, ’15, and ’16. If you didn’t, you’re in this stuff up to your gills, and you can’t get out. It’s like the Hotel California.

CB: Let’s say that PE continues to underperform. What’s going to be the opportunity in that for a manager like yourself?

DR: The first thing to note is that in making good investment decisions, you have to play defense and offense. In terms of playing defense, you want to avoid the hot things that everyone agrees on. Like, in society and culture, just like an investment, investments that reflect the culture, right? There are these fads, right? Think about the last few years in our public life. How many fads there have been where some massive percentage of people have all agreed on something — or pretended — and then it’s turned out that thing is crazy. I can think of a half dozen examples off the top of my head. And investing is a little bit like that.

You have to have this approach of saying no to a lot of things, saying, I don’t believe that. I’m not going to do it. And there’s a little bit of being a curmudgeon about this stuff, but I think you’ve just got to avoid it. Wall Street is such a skillful marketing machine that you’re always being sold things, and you’ve just got to say no to them. I’ve been trying to advise people to say no to things that were clearly bad, and that was this excessive enthusiasm for private equity and private credit. Again, when I say they were clearly bad, it wasn’t that they were clearly bad 10 or 15 years or 25 years ago. Back then, they were really good ideas. It’s just that once everybody agreed on them, they became bad ideas.

Consensus creates crisis. You have to be a contrarian, because the only way something gets so overvalued you can lose a lot of money on it is that everybody else agrees on it. So it’s a funny, funny part of the world, or definitional, you have to be contrarian. The second part of this is to try to apply that same logic and say, “what are the things that are actually good that everybody’s ignoring and hates for some reason and doesn’t want to invest in? And does that create opportunities?” That’s really where I focus my investing approach.

CB: Just as a case study, could you walk me through your covid year, what positions you held, what your returns were — your logic going through the last major crisis.

DR: Covid was an amazing time. In 2019, I had begun work on this idea of crisis investing. What I wanted to look at is, how do you invest when a crisis comes? We’ve always heard ‘sell the sound of trumpets, buy the sound of sirens.’ But what are we supposed to be buying? I looked at every crisis since 1970 and asked what you should have bought. You always know you’re in a crisis. You never know you’re in a bubble. We concluded that what you wanted to buy was the smallest, cheapest, most illiquid stocks that were down the most but were not going to go bankrupt. That was the core thesis.

We wrote this big white paper called crisis investing and published it in January of 2020. We had no idea when a crisis was going to come. And then all of a sudden Covid hit in March. We went out to our investors and raised about $170 million, and we deployed that in the summer of 2020. Fourteen months later we were up 85% out of fees. We fully liquidated the fund and returned it to investors. I went out and said this was the best buying opportunity since 2008 and you had to go out and do it.

Everyone else was saying the world is ending and there won’t be a stock market left. My view was that people were wildly overreacting to Covid. It was also true in the public equity market, where people were selling everything and panicking. Those times represent the best buying opportunities for disciplined long-term investors.

The only problem is you have to go against the herd. Look for what Mordecai Kurz at Stanford calls correlated beliefs. Start from what the correlated beliefs are, and then ask which of them are wrong. I’m a nonconformist, and in investing, that is tremendously profitable, because the only way to make a lot of money is to take nonconsensus views.

CB: Can you walk me through your different funds? And I’m especially interested in your Japan small-cap fund that you announced a couple months back.

DR: Again, I like contrarian bets. Japan had its lost decades where the stock market returned zero for 20 or 30 years, really, after the 1980s bubble burst. So a lot of people left Japan for dead. They thought, Look, there’s no growth there. There’s a declining population, there’s deflation, there’s massive debt to GDP. Japanese thinking, from a cultural and political perspective, and about Japanese companies relative to US companies, there’s some really good things and some bad things.

On the really good side, Japanese companies really pursue excellence in what they’re building or doing to the detriment of things like profitability. They get obsessed with building the absolute best rubber hose or something, and they’ll pour into R&D. And they also have a very strong commitment to their employees. They never fire people. It’s basically a lifetime employment guarantee. And the CEOs tend to be very humble. They don’t get big executive comp plans, and their salaries are relatively low. There’s an admirable nobility to it.

On the other hand, there’s this extreme conservatism that arose for a variety of reasons. Japanese managers for the past 20 years or so have been acting as though a tsunami was going to hit them or an earthquake was going to destroy their village, and therefore they had to hoard cash. They’d have fortress balance sheets. There was this massive buildup of assets in the balance sheet where they weren’t doing dividends, they weren’t giving buybacks. They were buying real estate or putting cash in the bank. It got to this extreme point where Japanese companies were just massively over capitalized and massively underpaying shareholders in terms of distributions.

Whenever a problem gets big enough, there are enough smart enough people that we’re going to come up with a solution. Obesity got big enough, and then we created GLP-1s. This problem got big enough, the stock market going nowhere got big enough, and the massive over capitalization got big enough that the government and the stock exchange got together and decided to fix it. In 2023 they promoted these corporate governance reforms. And the goal of the corporate governance reforms was to promote capital efficiency. And what they said is that if you’re a company that trades below 1X book value, basically below the value of your assets, you’ve got to put out a plan for how you’re going to change that. And the way you change that is very simple, which is, you sell the assets, and you buy back shares or do dividends. You take the money from your own balance sheet, you put it in the balance sheet of the investor, and quickly you right size the price to book problem.

This reform has just been a tremendous success, because thousands of companies trade below book in Japan, and this pressure has really pushed them to start taking steps that are moving in the right direction. I think we’re probably a third of the way through that reform, but a third of the way is a huge amount of positive change in Japan.

So the Japanese market, which was a laggard for so long, has all of a sudden been on fire. Buffett identified this at around the same time. I’ve been really building a Japan business for the last 10 years or so to try to capitalize on this sort of extreme undervaluation that was driven by this consensus view that Japan was a dead market. And that old conception has just been blown up by these corporate governance reforms. I think that as these corporate governance reforms play out, the profit opportunity for investors to go and buy these massively over capitalized, really undervalued shares in Japan is extremely attractive, and so I just launched a new fund focused on Japan in January. And now Japan has about a third of our assets. We manage about $1.5 billion, and about $500 million of that is in Japan. So it’s been a big part of our strategy.

The “chopstick catch” was the most precise rocket recovery in history. But as SpaceX engineers watched the live video feed of the Starship spacecraft that had flown atop the booster before separation, they saw the same problem that has haunted every attempt at reusable spaceflight for six decades. The hexagonal black tiles that protect Starship’s underside glowed and charred during reentry, plasma eating through the forward flaps. Starship survived long enough to execute a perfect flip maneuver and splash down in the Indian Ocean. Whatever remains of its heatshield is now beneath the waves.

Every Starship that has reentered Earth’s atmosphere has told the same story: tiles crack, tiles burn through, tiles fall off. The spacecraft is supposed to make spaceflight as routine as catching a Southwest flight to Phoenix, but a heat shield that degrades dangerously after every flight makes that impossible.

21 years earlier, on February 1, 2003, the Space Shuttle Columbia broke apart over Texas during reentry and killed all seven astronauts aboard. The cause was a breach in the reinforced carbon-carbon (RCC) panels that protect the leading edges of the shuttle’s wings. A chunk of foam insulation weighing less than two pounds had punched a hole about the size of a dinner plate in those panels 81 seconds after launch, 16 days before the disaster.

For the 15 days that Columbia orbited Earth, mission managers in Houston debated whether the strike had caused real damage. Engineers requested imaging of the damage from spy satellites. Management said no. The official view was that even if damage existed, nothing could be done about it. When Columbia reentered the atmosphere at 17,500 miles per hour, superheated plasma poured through that hole and ate the spacecraft from the inside out.

Between the Columbia disaster and the chopstick catch were two decades of stunning progress in propulsion, guidance, and autonomous systems. Rockets now land themselves on drone ships pitching in Atlantic swells. Satellites can be deployed by the thousand. Private companies put more payload into orbit than most nations. But the problem of developing materials that can survive the violence of atmospheric reentry and fly again without months of repair remains extremely tricky.

The First Heat Shields

Engineers understood the reentry problem before any human reached space, and the physics were brutal. When an object enters Earth’s atmosphere at orbital velocity, roughly 17,500 miles per hour for low-Earth orbit (LEO) and 25,000 mph for a return from the Moon, it compresses the air ahead of it so much that temperatures exceed 3,000° Fahrenheit. For a sense of what 3,000° F means: aluminum melts at 1,220° F and steel at 2,600° F. At that temperature, air itself becomes plasma, a screaming sheath of ionized gas that can vaporize almost anything. Almost.

The first solution came from NASA’s Ames Research Center in Silicon Valley, where researchers had been studying hypersonic heating since the early 1950s. The facility’s arc-jet tunnels blasted test materials with superheated gas at thousands of degrees, simulating in seconds what a spacecraft would endure during reentry. In 1953, the spirited engineer H. Julian Allen, nicknamed “Harvey” after the rabbit in the eponymous Broadway play, published his groundbreaking “blunt body theory,” which posited that a blunt, rounded nose would survive reentry better than the long, sharp noses used in ballistic missiles at the time. The blunt nose pushed the hottest air away from the spacecraft, creating a protective buffer. This theory led to the development of ablative heat shields — materials designed to char and vaporize, carrying heat away in the process. The concept was counterintuitive, almost perverse: protect the spacecraft by letting part of it burn away.

The Mercury capsule, which carried the first American astronauts to space, used an ablative shield made of fiberglass bonded with modified phenolic resin. As the capsule plunged through the atmosphere, the outer surface charred and vaporized, carrying heat away in the smoke. The shield worked. On February 20, 1962, World War II fighter pilot and future U.S. Senator John Glenn became the first American to orbit Earth and return safely in the Friendship 7 mission, though controllers spent his final orbits convinced a faulty sensor meant his heat shield had come loose. Glenn suspected trouble in-flight from the odd questions and small tests controllers kept asking him to perform, but didn’t learn the cause of their alarm until after splashdown.

The Apollo program, however, needed something tougher. A spacecraft returning from the Moon would hit the atmosphere at nearly 25,000 miles per hour, over 40% faster than an orbital return, and the heating would be correspondingly more savage. NASA contracted with Avco Corporation to develop a new ablative material called AVCOAT 5026-39, a honeycomb structure filled with epoxy-novolac resin reinforced with silica fibers. The shield varied in thickness from 0.7 inches to 2.7 inches, with the heaviest armor at the capsule’s base where the plasma hit hardest.

On November 9, 1967, Apollo 4 tested the shield at lunar-return velocities for the first time, speeds boosted by the Moon’s gravity accelerating the spacecraft earthward. The unmanned capsule punched into the atmosphere at 24,974 mph and came through intact. Twenty months later, Apollo 11 brought Neil Armstrong, Buzz Aldrin, and Michael Collins home from the Moon.

The Soviets, meanwhile, were building their own lunar hardware. The N1-L3 rocket that was supposed to carry cosmonauts to the Moon never completed a successful test flight. Four launches between 1969 and 1972 all failed — one even ending in a launchpad vehicle explosion — the result of chronic problems with the rocket’s nightmarish 30-engine first stage. But Soviet engineers had cracked the reentry problem. Their Zond spacecraft, designed to loop around the Moon and return, used ablative materials similar to the American approach. Zond 5 completed a circumlunar flight in September 1968, three months before Apollo 8, and survived reentry despite subjection to 20 g‘s of deceleration.

The problem with ablatives was obvious: they couldn’t be reused. Every flight required a new heat shield, every heat shield required months of painstaking manufacturing, and the cost of reaching space stayed astronomical. For the one-shot capsule programs of the 1960s, this was acceptable. Each mission was a discrete event, planned years in advance, celebrated like a moon landing even when it wasn’t. But if spaceflight was ever going to become routine, something had to change.

The Space Shuttle was supposed to be that change. Unlike the capsules that preceded it, the Shuttle was designed to fly repeatedly. NASA’s original projections, drawn up when President Nixon was still in office, called for flights every two weeks rather than every few months. An ablative heat shield that burned away on every flight was incompatible with this vision. The shuttle needed thermal protection that could survive reentry and fly again more or less immediately.

The solution came from Lockheed Missiles & Space Company. In the 1960s, a Lockheed engineer named Robert M. Beasley, who had joined the company from Corning Glass Works and spent years studying heat-resistant ceramics, developed a material called LI-900. The tiles were made from 99.9% pure silica glass fibers and were 94% air by volume, weighing just nine pounds per cubic foot (steel weighs about 500 lb/ft3) but able to withstand temperatures of up to 2,300° F. Its qualities were almost magical: at maximum heat exposure, the edges of an LI-900 tile could glow red-hot while the interior stayed cool enough to touch with a bare hand.

The tiles worked through a combination of insulation and radiation. Silica fibers are terrible conductors of heat. The tile’s surface was coated with reaction-cured glass that reflects 90% of the heat back into the atmosphere, while the tiles’ interior absorbs the remaining 10%. Each shuttle bore 24,000 of these tiles. The shuttle’s complex curved surfaces meant that no two tiles had exactly the same shape. Each had to be individually manufactured, numbered, and installed in its precise location by technicians, like archaeologists reassembling pottery shards. The gaps between tiles had to be filled with a flexible material that could accommodate thermal expansion without cracking. The entire thermal protection system was, in effect, a 24,000-piece jigsaw puzzle where every piece was irreplaceable and a missing one in the wrong location could kill the crew.

The fragility of this system revealed itself on the very first shuttle mission. When Columbia touched down on April 14, 1981, after a two-day orbital flight, inspectors swarming the orbiter found that 16 tiles had ripped off entirely and another 148 were damaged. An overpressure wave during launch had stripped tiles from the orbital maneuvering system pods at the back of the spacecraft. Fortunately, it was an area where reentry heating was less severe compared to the belly or wing leading edges.

NASA had expected some tile losses. The system was designed to tolerate limited damage. But the extent of the problem was troubling, and over the following years engineers scrambled to develop better bonding agents and installation techniques. Tile losses decreased but never stopped. Every shuttle flight came home wounded. Every turnaround required extensive inspection and repair. The dream of airline-style operations collapsed into the reality of six to eight flights per year, with months of preparation between each one.

The tiles protected the shuttle’s underside, where heating during reentry was intense but relatively uniform. The wing leading edges faced something worse. Here the airflow compressed against a curved surface, creating localized heating that could exceed 3,000°F, far beyond what silica tiles could survive. These areas were protected by reinforced carbon-carbon panels, called RCC: a composite of carbon fibers in a carbon matrix, coated with silicon carbide and secured with a glassy sealant to prevent oxidation.

RCC could handle the heat, but it was brittle, almost fragile, and the panels could be cracked or punctured by impacts that would barely dent aluminum. NASA knew this. Foam strikes from the external tank had occurred on multiple shuttle flights without apparent incident. Mission managers came to view them as a nuisance, a maintenance issue rather than a safety concern. This was a grave mistake, as revealed in the Columbia disaster.

No Solution in Sight

The Columbia Accident Investigation Board released its report in August 2003, and the findings were damning. NASA’s organizational culture had become so fixated on schedule and budget that safety concerns were routinely waved away. The foam problem had been normalized, treated as an acceptable risk because nothing bad had happened yet. The report called for sweeping changes to how NASA managed safety. It also surfaced a harder truth: the shuttle’s thermal protection system was brittle, and the RCC panels that had failed on Columbia were relics of 1970s engineering—developed using state-of-the-art technology of that era, but lacking the impact resistance to survive a foam strike. The panels were manufactured through a complex, multi-stage process, and for safe operation, the hottest leading-edge panels were limited to 50 missions. After Columbia, NASA worked to procure replacement panels and maintain a spare inventory, but the program’s days were numbered.

From 2005-2011, the shuttle program flew 22 more missions after Columbia before its sunset. The three remaining orbiters — Discovery, Endeavour, and Atlantis — completed their final flights in 2011, with Atlantis landing on July 21, 2011 to close out the program. During those years, NASA flew with obsessive caution, inspecting the thermal protection system in orbit using the shuttle’s new robotic arm and arranging for the International Space Station to serve as a lifeboat if inspectors found damage. No further accidents occurred, but the shuttle never came close to achieving the routine reusability its designers had hoped for. Two decades later, SpaceX is wrestling with the same challenge.

Starship’s heat shield uses roughly 18,000 hexagonal black tiles on the spacecraft’s windward side, fewer than the shuttle but still a complex system requiring precise installation and constant maintenance. The exact composition of the tiles is a closely held company secret, but the material is believed to be made up of silica, alumina-borosilicate, and aluminum oxide.

The shuttle descended steeply during reentry, its nose pitched up like a fighter jet coming in for a landing — meeting the atmosphere at a sharp angle. But Starship’s upper stage (the spacecraft that carries cargo and eventually crew) takes a different approach: flipping perpendicular to its flight path and falling through the atmosphere broadside, its entire belly facing the oncoming air like a skydiver in freefall. This “belly-flop” maneuver spreads the intense heating across a larger surface area, but creates bizarre stress patterns the shuttle never experienced. Each test flight generates terabytes of data about where the tiles crack and why.

The iteration cycle is rapid and unconventional, especially by NASA standards. SpaceX builds quickly, flies early, expects failures and extracts maximum data from each test cycle. SpaceX’s test philosophy treats explosions as tuition payments: everything is a learning opportunity. SN8, which flew in December 2020, demonstrated the belly-flop maneuver before slamming into the landing pad and detonating in a fireball. In May 2021, SN15 achieved the first successful landing. The Integrated Flight Tests that began in 2023 pushed toward orbital velocities, revealing heat shield failures that no wind tunnel or computer model had predicted.

IFT-3, in March 2024, was the first Starship spacecraft to experience full orbital-velocity reentry heating, and it shed tiles like a dog losing its winter coat. IFT-4, in June 2024, showed better tile adhesion but extensive burn damage to the ship’s control flaps, which had warped and charred in ways that concerned even SpaceX’s optimistic engineers.

Each test burns an estimated $100 million. SpaceX is running materials experiments at industrial scale, generating performance data that supplements computer models and ground testing. “For full reusability of the ship, there’s still a lot of work that remains on the heat shield. No one has ever made a fully reusable orbital heat shield,” Elon Musk said during an interview at the All-In Summit in September 2025. “We really are looking at fundamental physics here, trying to figure out how we make something that can withstand the heat, is very light, doesn’t transmit the heat to the primary structure, and the tiles stay on and don’t crack.”

The heat shield must survive reentry not once but dozens or hundreds of times if Starship is going to fulfill its intended purpose. An ablative system that burns away on every flight would require replacement before each mission. Manageable for Earth operations, maybe, but completely unworkable on Mars, where there are no tile factories, no bonding agents, no technicians to do the work. A reusable Starship requires a heat shield that can fly to Mars, land, refuel, launch, survive Earth reentry, and fly again. No such material exists today outside of PowerPoint slides.

The Knowledge Problem

The materials challenges of spaceflight extend far beyond heat shields. The space environment puts stress on all materials: radiation, temperature swings of 500° F between sun and shadow, hard vacuum conditions that cause materials degradation, micrometeorite impacts, atomic oxygen erosion... Each demands specialized solutions, and the solutions often conflict with each other.

Radiation-hardened electronics are another critical bottleneck. Earth’s atmosphere and magnetic field shield the surface from most cosmic radiation, but spacecraft in high orbits or beyond Earth’s magnetosphere are hammered by energetic particles that can flip bits in computer memory, corrupt data, or fry circuits permanently.

On April 5, 2010, the Intelsat Galaxy 15 communications satellite stopped responding to commands from Earth. The spacecraft’s computer had apparently been scrambled by a radiation-induced anomaly, likely from a strong solar event that had swept through geostationary orbit. The satellite remained functional but uncontrollable, a 4,000-pound zombie drifting through the geostationary arc and threatening to jam other satellites’ signals. For eight months, Intelsat and rival operators tracked the wandering satellite with growing alarm. Then, on December 23, 2010, the spacecraft’s computer spontaneously rebooted and normal operations resumed.

More capable spacecraft rely on specialized chips manufactured with processes designed to shrug off particle strikes. BAE Systems and Honeywell produce radiation-hardened processors using silicon-on-insulator fabrication and redundant circuit architectures that catch and correct errors before they propagate. These chips are expensive and typically run more than a decade behind the performance curve of the phone in your pocket.

Rocket engines must be designed to withstand the extreme violence of continuous explosions. The SpaceX Raptor engine operates at chamber pressures exceeding 300 bar, more than 4,350 pounds per square inch, with combustion temperatures over 5,400° F. The engine nozzle must contain this inferno while staying light enough to lift itself to orbit.

SpaceX uses proprietary nickel superalloys designated SX300 and SX500 for critical Raptor components. Nickel-based superalloys hold their strength at temperatures that would turn most metals to taffy. The nozzle is cooled regeneratively: liquid methane flows through channels machined into the nozzle walls, absorbing heat before being injected into the combustion chamber. The engineering is exquisite, but it depends entirely on materials that can survive conditions where most substances simply cease to exist.

The development history of rocket engines is largely a chronicle of materials breakthroughs and materials disasters. The Saturn V’s F-1 engine, which remains the most powerful single-chamber liquid-fueled rocket engine ever flown, required seven years of development to tame combustion instability problems that kept blowing engines apart on test stands. The solution was a meticulous redesign of the fuel injector, 2,832 precisely positioned orifices arranged in patterns that spread combustion evenly across the chamber. The injector geometry and the alloys that could survive it emerged from thousands of tests, countless explosions, and a blank check from a nation terrified of Soviet rockets.

Then there’s the square-cube law that haunts every structural engineer: a rocket twice as tall doesn’t need structural members twice as strong, but eight times stronger, because the mass it must support scales with the cube of its dimensions while the cross-sectional area of structural members scales only with the square. This law pushed rocket designers toward ever-lighter materials: aluminum-lithium alloys, carbon fiber composites, and eventually, in Starship’s case, stainless steel.

The choice of steel seemed crazy when Musk announced it in 2018. Steel is heavier than carbon fiber or aluminum, but it has properties that composites lack. It actually gets stronger when supercooled by liquid methane and oxygen propellants. It holds its strength at high temperatures, surviving reentry heating that would destroy lighter materials. And it’s cheap, available at any steel yard in America, and easy to weld in a muddy field in South Texas.

The switch was a systems-level gamble, trading mass for thermal performance and manufacturing flexibility. It was also a bet that materials constraints could be managed through clever engineering rather than exotic metallurgy. Whether that bet pays off depends on whether SpaceX can solve the tile problem that keeps burning through their spacecraft.

The Computational Promise

The biggest problem with heat shield development is that we’re still largely discovering materials through human trial and error. A tile composition that survives test flight IFT-4 might fail catastrophically on IFT-5. Engineers then adjust the formula based on their best educated guess, run hundreds of simulations, and hope for the best. But this iterative approach is glacially slow and expensive. To progress heat shield technology at the pace needed to meet space travel ambitions, we need a shortcut.

In November 2023, Google DeepMind announced GNoME, an AI system that predicted 2.2 million stable inorganic crystals, materials used in technologies like computer chips, batteries and solar panels. The discovery was a hundredfold increase over what human scientists had previously catalogued, the equivalent of 800 years worth of knowledge. The announcement sparked breathless headlines about AI revolutionizing materials discovery.

The reality is messier.

Predicting computationally that a material might be stable is the beginning of a very long road, not the end. A crystal structure that looks thermodynamically stable on a screen might prove impossible to synthesize in a real laboratory. An alloy that can be cooked up in small batches might behave completely differently at industrial scale. A ceramic that performs beautifully in a test furnace might crack the first time it experiences the thermal shock of atmospheric reentry.

NASA’s Technology Readiness Levels formalize this painful truth. A material at TRL 1 exists only as a concept on a whiteboard. At TRL 3, someone has demonstrated it works in laboratory conditions. At TRL 6, it’s been validated in an environment resembling actual use. At TRL 9, it flew successfully and proved itself in operation. The journey from TRL 1 to TRL 9 typically consumes 10 to 20 years. AI can accelerate the earliest stages, but it can’t compress the testing, qualification, and operational validation that follow. At least, not yet.

Belgian-American scientist Gerbrand Ceder — with whom I co-founded Radical AI — has spent decades trying to speed this process up. A professor of materials science and engineering at UC Berkeley who co-founded the Materials Project, an open database of computed materials properties, argues that the bottleneck to materials progress is no longer computational because supercomputers and sophisticated algorithms are plentiful. What’s scarce is experimental infrastructure to test predictions at scale and manufacturing capacity to produce validated materials in quantity. At Lawrence Berkeley National Laboratory, Ceder built A-Lab, the first autonomous facility where robots synthesize and characterize around the clock without human hands.

But even with robots working 24/7 and exhaustive experimental data capture, the qualification gauntlet for space-rated materials is painstaking. Thermal cycling tests subject samples to repeated heating and cooling across temperature ranges simulating the space environment. Vibration tests shake components at frequencies from 20 Hz to 2,000 Hz to simulate the violence of launch. Outgassing tests check whether materials release contaminants that could fog optics or poison sensors. Atomic oxygen exposure tests simulate the corrosive soup of LEO. Each test generates data, each data point feeds qualification, and none of it can be skipped or shortcut.

The Valley of Death

The chasm between laboratory discovery and commercial production has a name among researchers: the valley of death. Promising materials routinely fail to make the crossing, because scaling them is expensive, tedious, and nobody’s job.

Gorilla Glass offers an instructive detour from aerospace.

In 1960, Corning scientists worked on “Project Muscle,” developing a chemically strengthened glass they named Chemcor. Strong and scratch-resistant, it seemed like a breakthrough — but no mass market materialized. Chemcor limped along in niche applications for the automotive, aviation and pharmaceutical industries, but faded from use by the early 1990s.

Then, in early 2007, Steve Jobs unveiled the iPhone at MacWorld with a plastic screen. But the day after his presentation, Jobs noticed the screen on the prototype had scratched from the keys in his pocket. He called Apple’s chief operating officer, Jeff Williams, with a directive: “I don’t know how we’re going to do it, but when it ships in June, it’s gonna be glass.”

A few days later, a panicked Williams got a call from Corning CEO Wendell Weeks: “Your boss called and said my glass sucks.” Despite the insult, Weeks proposed reviving the company’s old chemically strengthened glass research.

Within six miraculous months, Corning had spun up production, tweaked the composition, and shipped enough to cover the first iPhones. (By February 2008, the material was formally rechristened “Gorilla Glass.”) The glass gained its strength from ion exchange: soaking it in molten potassium salt forced fat potassium ions to muscle aside smaller sodium ions in the glass surface, creating compressive stress that resisted cracking. The chemistry has been understood for half a century — the only thing that changed was a customer with deep pockets and a deadline.

Today, Gorilla Glass protects more than 8 billion devices worldwide, from smartphones to laptops to car dashboards and last year, Apple pumped a fresh $2.5 billion into Corning’s Kentucky manufacturing facilities to guarantee supply. A multi-billion-dollar business built on materials science that gathered dust for decades because nobody needed it badly enough.

Materials that could solve reentry heating, radiation hardening, or structural weight problems for the space industry may already exist in some university freezer. But the infrastructure to scale them doesn’t. Unlike consumer electronics, where a single deep-pocketed customer can yank a dormant technology into production overnight, aerospace procurement moves at glacial pace. Qualification cycles stretch for years with contracts that favor incumbents. The incentives to bridge the valley are weak, and promising materials keep dying in the crossing.

The Metrics That Matter

When SpaceX catches a rocket with its chopsticks, it’s demonstrating precision guidance and propulsion control. Those systems have matured to the point where the catch itself was almost anticlimactic, the booster dropping exactly as planned and grabbed exactly as intended. The catch is the easy part. Everything that happened during reentry — when tiles glowed, cracked and burned away — is another story.

Ask most people how to measure space leadership and they’ll point to spectacle: biggest rockets, most satellites, most dramatic recoveries racking up views on YouTube. Ask an aerospace engineer actually doing the work, and you’ll get a very different answer. What matters is the materials learning curve: How fast does a program accumulate knowledge about what survives and what doesn’t? How effectively does that knowledge get retained and fed into the next design? How efficiently does a laboratory discovery translate to qualified hardware to factory production?

By these measures, SpaceX is the clear leader. Its tiles are better than everyone else’s, sure, but their testing tempo also generates data at rates no other organization can match. Each flight is an experiment, each cracked tile is a data point, and each iteration improves on the last. The materials problems remain unsolved, but their learning rate is unprecedented.

China is steepening its own curve through systematic investment and centralized command. The European Space Agency, Japan, and India each maintain independent programs at varying levels of capability. Russia’s space effort, once a peer competitor, has withered under sanctions and brain drain. The exodus of Soviet talent that enriched Western programs after 1991 can’t be repeated, and no comparable reservoir of expertise waits to be tapped.

The constraint that keeps winning is physics.

Materials must survive temperatures that vaporize most substances, must hold their properties through hundreds of thermal cycles, brutal vibration loads, and relentless radiation bombardment, and must be manufacturable at rates supporting operational flight schedules. They must also cost little enough that reusability actually saves money.

These requirements haven’t changed since Mercury. What’s changed is our ability to hunt for solutions systematically, test candidates rapidly, and learn from failures collectively. The core problem — finding materials that can take the punishment of reentry and fly again without endless refurbishment — is exactly what it was when NASA engineers first wrestled with it in the 1950s.

The nation that cracks truly reusable thermal protection will own the next century of spaceflight, because everything else — the rockets, the satellites, the Moon bases — will flow from that breakthrough. SpaceX is closer than any entity has ever been, but even the gem of 21st-century American aerospace excellence hasn’t quite figured it out. On any given day at SpaceX’s Starbase facility in South Texas, technicians can be found crawling over Starship prototypes, inspecting heat shield tiles, prying out damaged ones, gluing in replacements, adjusting gap filler, prepping the spacecraft for its next ride through hell. Forty years ago, technicians at Kennedy Space Center did the same work on Columbia and her sisters.

The tools have improved, the materials have evolved, but the manual, painstaking work remains — and the brutality of the atmosphere, and the laws of physics, too.

On a beautiful street in downtown Santa Barbara (they all are, really) is a five-story building with a classic Spanish-Californian facade, complete with cornices and vintage ironwork over the perfectly-spaced windows. If a pedestrian were to guess just what kind of business occupies it, I think they’d guess wrong. A boutique law firm, perhaps. Maybe a high-end architecture practice? An art gallery? The answer is that some of the world’s most advanced imaging satellites have been designed and built there by a company called Umbra. On January 11th, Umbra launched its twelfth synthetic-aperture radar satellite (SAR is an advanced radar system for creating high-resolution imagery — more on this later) on a SpaceX Falcon rocket carrying 39 other small satellites. The company, founded in 2015, launched its first three satellites over a one-year period beginning in 2021. Now, it is preparing to open a new manufacturing facility in nearby Goleta, with the capacity to produce more than 50 satellites per year.

Even with a small constellation — for now — Umbra has made big news with its capabilities.

When Wagner Group boss Yevgeny Prigozhin’s plane went down in August 2023 after his short-lived mutiny against Vladimir Putin, Umbra’s SAR imagery was the first to show the crash site; those images, taken by radar at night when cameras couldn’t see, ran on CNN and showed the state of the wreckage. For a company whose first satellites had been up for only two years, it was a significant moment. Hardware that had been designed and built by a very small team on the quiet California coast was the source of a major international headline in a bloody conflict a world away — and the images were on TV!

David Langan is Umbra’s CEO and a lifelong resident of Santa Barbara. He began his aerospace career in a way I can only describe as characteristic of the town. “I had just come back to Santa Barbara from studying mechanical engineering at UC Irvine and was starting to look for a job in the field. I ran into a woman while I was walking my mom’s dog one day,” he said. “She recognized the dog, but she didn’t know me. I started talking to her, and she found out I was looking for a job. She introduced me to a contact of hers, and two days later, I had a job at Orbital ATK, which is now part of Northrop Grumman. And that was my career for the first seven years.”

I repeated the facts back to him, asking with some amusement, “a woman recognizes your mom’s dog, and then you’re conscripted by Northrop Grumman? It’s very Santa Barbara.” (The California coast, idyllic as it is, has long been a center of American military technology)

“That’s exactly how it went. Conscripted is a good word,” he said back.

Years later, after time in the defense world and a growing conviction that radar remote sensing could be done better, faster, cheaper, more openly — he left to start Umbra. Today, Umbra operates one of the most sophisticated synthetic-aperture radar (SAR) constellations in the world, public or private. It is an undeniably exciting frontier in the long technological story that is radar.

What is Radar, and Why is My Aperture Synthetic?

Radar is a simple but brilliant technology. James Clerk Maxwell’s theoretical predictions in the 1860s that electric and magnetic fields could propagate through space established that “waves” could exist before anyone had produced or detected them. Years later, experiments by German physicist Heinrich Hertz showed that radio waves behaved exactly as Maxwell had predicted: they reflected, refracted, diffracted, and polarized like light visible to our eyes.

If invisible electromagnetic waves behave like visible light, and like sound, then in principle it should be possible to calculate the distance between objects by creating a wave, waiting for it to be reflected off some object or surface, then timing how long it takes for the wave to return. It is not unlike shouting in a cave and waiting for an echo to gauge how deep the cave is, though light travels much faster than sound.

A simple radar system uses a transmitter to generate radio frequency signals, and an antenna to radiate them as a wave. The antenna can also “catch” the returning signals, which are then processed. By measuring the time it takes for a signal to return, you can measure the distance. This system, with a stationary antenna, is the most primitive one-dimensional radar.

If you allow the antenna to move along another axis (or rotate), you have created a two-dimensional radar. Maritime radars detect other vessels by spinning 360 degrees and looking for them; their range is determined by the curvature of the Earth and the height of the radar above the water, not unlike your own line of sight from a beach watching ships slip over the horizon. If you were twice as tall as you are, you’d be able to see farther.

Radar came of age right as the world was about to be plunged into the chaos of the Second World War. It was in 1940 that the United States Navy officially adopted the acronym of ‘radar’ (radio detection and ranging) and implemented the first maritime radars aboard US ships.

If you add another dimension by allowing a radar to look up (either literally by moving the antenna up and down, or by creating a phased array of multiple antennas), things start to get interesting. Now, you can start to resolve flying objects and get information about their position, altitude, direction, and speed. Given the propensity for metal objects in the sky to cause trouble — I’m thinking of aircraft (both friendly and unfriendly) and missiles (all unfriendly) — you can imagine just how useful this third dimension is.

The United Kingdom’s “Chain Home,” with antennas dotting the southern English coast, had ranges just far enough to see groups of German aircraft across the English Channel in France. Those basic radars gave the men of the Royal Air Force a fighting chance against the much larger Luftwaffe, allowing them to concentrate their responses to German air raids in their heroic defense of Britain in the summer and autumn of 1940.

It was also in 1940, during the Battle of Britain, that the British first put radars onboard planes themselves. Airborne radar is enormously useful because it can tell you what’s on the ground, especially at night. The catch is that the resolution of an image depends on antenna size. By concentrating power effectively, a large antenna produces a sharper image, the same way a larger telescope mirror gathers more light and reveals finer detail. The stationary, ground-based antennas of Chain Home were enormous steel towers — even the largest American bombers of the age couldn’t fit anything close to their antennas on board.

But with a bit of clever engineering, you can fake it. A radar mounted on a moving platform — an airplane, or eventually a satellite — can take measurements at many points along its flight path and then combine them mathematically, as if all those individual measurements came from one very long, impossibly long, antenna. The “synthetic aperture” is the imaginary antenna you’ve constructed via computation. So, an aircraft flying along a path a mile long can “synthesize” an antenna a mile long, achieving resolution that would otherwise require an absurdly large physical structure.

Early SAR systems filled aircraft with racks of specialized processors; the images they produced sometimes took longer to compute than the flight itself. As is the story with so many technologies whose infancies were in the first half of the twentieth century, the transistor and chip revolution of the second half of the century changed everything. The explosion of computational power, tracked by Moore’s Law, made it possible to measure more with less, and measure it at greater and greater resolutions.

The first SAR satellite for Earth observation was called SEASAT. Launched in 1978, NASA used SAR from 500 miles up to measure Earth’s oceans in unprecedented detail — seeing wave patterns, surface winds, sea ice, and ocean topography through clouds and darkness. SEASAT operated for just over 100 days before a short circuit killed it, but in that time it proved that spaceborne SAR could work really well.

Today, several commercial entities operate space-based radar. Among them, Umbra has the distinction of having produced the world’s highest-resolution commercial SAR image. The image shows the historic Dole pineapple plantation outside Honolulu, resolved at just 16 centimeters. To give you a sense of just how impressive 16-centimeter resolution is from space: at that resolution, a satellite orbiting the Earth multiple times per day can reliably count pineapple plants. Individual pineapples are dodgier but possible. You might be able to see an enormous grapefruit, but lemons are out of the question.

I don’t need to tell you that Umbra’s customers aren’t paying them to count fruit. The implications of worldwide, high-resolution SAR imagery, available 24/7 are enormous — for militaries, intelligence agencies, corporations, investors, scientists, and plenty of other people who want to observe the Earth and the drama happening on it with precision never before available.

How to Build a Radar Satellite from Scratch

Before Umbra raised a dollar or hired any employees, David Langan was cutting paper models on a kitchen table. “I was just thinking, how could these things fold together, and trying to visualize the fully 3D structure with the mechanism and mesh and cables and all the things that might be necessary,” he said. Langan and his co-founder Gabe Dominocielo bootstrapped the company for three years, buying oscilloscopes and spectrum analyzers off eBay from the 1980s. “They still worked,” Langan said. “They actually worked really well, so we still rock them.”

From the electrical lab in the Santa Barbara building, with the old hardware still on the shelves, we walked to Umbra’s clean room, converted from what used to be several classrooms, once painted in garish greens and purples — that Umbra gutted and rebuilt. “All of our satellites to date have been built and integrated in this room,” Langan said, speaking with me inside that clean room. Down the hall are the rooms where the original testing took place; they are now conference rooms. There’s a storage room stacked from floor to ceiling with Uline bins full of tiny components — enough to build six satellites. In the new Goleta facility, parts will ride on carousel-like contraptions that rotate up into the air, a Ferris wheel for satellite components.

Part of Langan’s previous job at Northrop Grumman had been pricing out satellite components, so he had “a somewhat encyclopedic knowledge” of what things cost. Batteries were hundreds of thousands of dollars. Custom solar arrays built by a contractor would run into the millions. “You could just say, ‘no, we have to build that ourselves because there’s no way we could afford to buy a solar array,’” he said. “And even if we could, we didn’t want to wait.”

Umbra’s co-founder Gabe Dominocielo said this: “I knew David could make it work technically. My focus was whether it worked economically. If it didn’t make sense at the unit level, none of it mattered. From day one, we were obsessed with cost and efficiency.”

Today, just about everything at Umbra is vertically integrated, with the exception of the rocket (thank you, SpaceX) and the ground stations near the North and South Poles that help the Umbra constellation work — though, on the latter front, Umbra may very well change things.

“We looked at design papers and scoured the internet and textbooks,” Langan remembered. “We applied that approach to absolutely everything. How do you build a battery? How do you build a solar array? We broke it down to how we could build it in the most efficient manner possible, with the most modern parts possible. And we built the entirety of the spacecraft and the payload with that methodology.”

They patented their antenna design and raised their first seed funding in early 2018, during the California wildfires. “There was lots of ash in the air. People were coming into the office with masks,” Langan recalled.

When Umbra needed to test whether their first radar could actually generate an image, they found a pilot willing to let them strap the instrument — about the size of a fire hydrant — to the side of a skydiving plane. Young Umbra engineers wore parachutes and fired the radar out the open door over Goleta. “This is the very first shot that we captured,” Langan said, pointing to a photo taped to a whiteboard. Then Gabe Dominocielo sent them back up — this time with Red Bull cans scattered in a field below. “We told people that we were going to be able to see a Coke can in the field,” Langan explained. “And we caught those in a later shot.”

Mike Francis was the third employee. He’d grown up in Connecticut, gone to MIT, and spent ten years at Northrop Grumman in El Segundo. “My wife and I were looking to get out of LA,” he told me. “We wanted something a little more family oriented, a bit of a small town. I was looking at Santa Barbara, saw this posting, and I literally just applied online. David gave me a call.” That call lasted for three hours; today, Francis is in charge of building Umbra’s current satellites, and leads the team designing the next ones.

When Langan and Dominocielo started the company, they had to take it on faith that launch costs would come down — Langan called it “pressing the ‘I believe!’ button.” In early 2021, SpaceX introduced “Rideshare” — where dozens of satellites can be launched on a single Falcon rocket (renting the whole rocket would cost a single customer around $60 million, impractical for smaller satellite companies). “That ended up being a huge catalyzing event for the American space industry,” Langan said. For Umbra, it was critical.

Rideshare works because SpaceX standardized the interface. Umbra’s satellites — and their rideshare buddies from other companies — bolt to a fifteen-inch ring on a shared plate aboard the Falcon 9. The interface is not unlike the stem of a Brussels sprout plant, with satellites neatly spaced along the “stalk.” Once the Falcon has slipped into space, the shell of the spacecraft opens, and one by one the satellites detach. SpaceX provides the ring specification — the circular bit that the satellite gets bolted onto — and the ride to space; the customer provides everything else, including the separation system. Mike Francis told me a bit more about that process: “They have a whole test suite you have to go through,” he explained. “You have to prove to them it’s actually going to deploy so that you don’t get stuck on there.”

About three and a half months before launch, Umbra delivers an environmental test report: proof that the satellite has survived stress testing like a shaker table; that all its deployments will work; and that it won’t damage anything else on the rocket. A month or so out, the Umbra team brings the satellite to the integration facility, either at Cape Canaveral or Vandenberg, just up the coast from Santa Barbara, and bolts it onto the plate. After launch, depending on the mission profile, it can be anywhere from an hour to several hours before the satellite is released. First contact with Umbra’s ground stations typically comes within thirty minutes of separation. The solar arrays deploy immediately; the antenna deployment is commanded from the ground once the team confirms everything else is operating normally. Forty-seven hours after one recent launch, Umbra had its first image.

✺

COO Todd Master took a longer path to Umbra. A former Air Force officer, he’d spent five years at DARPA before joining the company. But his connection to Langan went back years earlier, to a classified review room at a big prime contractor when they were both working for different parts of ATK. A young engineer was presenting an electromechanical payload. “Everybody’s just blown away,” Master recalled. That young engineer was David Langan. They worked on projects together. Only during his Umbra interview, years later, did Master learn it hadn’t just been Langan’s first presentation. “It was his first project!” Master said, laughing.

“We actually talked about Todd for a couple of years before we hired him,” Langan told me. “We had a plan for Todd.”

As Master and I drove from downtown Santa Barbara to the company’s new manufacturing site in Goleta, the quaint streets gave way to larger boulevards, office parks, and industrial space — though the beautiful oaks largely remain. I saw two men walking on the sidewalk in Lockheed Martin polo shirts.

“This is really the transition you’re seeing at Umbra,” Master said, “from the funky, cool, scrappy startup to what a serious aerospace and defense company is.” Today, Master is leading a major expansion as GM of Space Systems — selling not imagery but the actual satellite components Umbra has developed to the broader industry.

We talked a bit about the talent at Umbra. He came up through the traditional aerospace world — Air Force, defense primes, DARPA — but the talent pipeline feeding Umbra looks almost nothing like the one that trained him. “A lot of people are familiar now with CubeSats,” he explained. “It was invented at Cal Poly. So we’ll have students graduating from college, fresh out of undergrad, who come to us and interview who’ve actually operated satellites for years.”

The CubeSat framework was developed in 1999 by two professors, one at Stanford and one at California Polytechnic University, San Luis Obispo. CubeSats have a standard measurement of 10x10x10 centimeters and a mass of two kilograms or less. The standardized format made it possible for truly small customers like colleges and early-stage startups to launch real satellites. Several thousand CubeSats have been launched, mostly as academic projects. For Umbra, recruiting from Cal Poly and similar programs meant being able to hire young engineers with actual hands-on experience designing satellites and coordinating missions.

“We found this kind of special place,” Master said as we pulled into the parking lot. “It’s an office park, but office park has a negative connotation to me. This is one of the nicest I can imagine. It’s peaceful and serene. When David first saw it, he was like, ‘It’s so serene, the birds outside.’ He liked it.”

Sitting in Langan’s office, with a view of some winding California oaks over the parking lot, I asked the man in charge what technical achievement he was most proud of as CEO. “There are so, so many,” he began. “We built an exceptional radar payload, soup to nuts, from bolts and microwave integrated circuits, with engineers right out of undergraduate study. And we did it in just a couple of years, and it let us come from behind and dominate with far fewer satellites. It set us up for a path where we know we’re just getting started. It is going to get so much better. And the competition we’re up against is not going to grow in the way that we can and will grow in the next couple of years.”

The initial manufacturing goal of the Goleta facility is to scale to a point where the team builds one satellite per week. I was curious whether Todd and David thought there was a saturation point — whether at some point there would be enough satellites to meet all the demand for this kind of imagery. “I think there is,” Master said. “And that really gets down to what the quantity gets you. We’re not just building to serve quantity, because quantity is only part of scale. All that’s really giving you is shorter gap times between collections. But that doesn’t necessarily mean more capacity per system, which is a big metric that we use. We have a very high-capacity system.”

“I think it’ll be hundreds to thousands,” Langan said. “I don’t think SAR satellites will be numbered in the tens of thousands. That won’t happen. Yet I’ve been surprised at the massive expansion of radar satellite procurement, and we’re focused on building new capabilities that integrate persistent collection with a persistent connection — an always-on, resilient architecture.”

✺

For a company that does critical business with US military and intelligence (and numerous US allies), it may surprise you that the Umbra website has a public price list, and a dashboard for anyone to purchase new or existing imagery, as well as a considerable library of open data. A one-meter resolution SAR snapshot of a five-by-five kilometer area can be yours for just $675. A finer resolution will cost you — $3,250 for 25-centimeter resolution on that same area.

Critically, Umbra issues its paid imagery under a Creative Commons license — specifically the CC-BY-4.0 license, which is an open-source license requiring only attribution by the licensee. I was somewhat surprised to learn that it was at one time standard practice in the remote sensing industry for companies to sell images under highly-restrictive licenses, enjoining purchasers from doing much with the images they’d bought.

Joe Morrison, who leads Umbra’s remote sensing business, explained the problem this way: “The reason why the CC-BY-4.0 license is so important is this: how do you build an industry on top of this new source of data if you’re not allowed to share the results with anyone else?” He came to Umbra as a convert. At his previous firm, he’d been purchasing satellite imagery for both defense and commercial use cases, developing AI to extract information from it. “I spent half my time on the phone with salespeople who would refuse to answer my questions,” he told me. “I’d say, ‘I have $100,000, what can I buy? I’m waving money at you. What does it cost?’” The opacity drove him so crazy that in 2020 he wrote what he called “a farewell letter to the industry” cataloging everything that was broken about satellite imaging.

That’s how Umbra found him. “They called me and said, ‘we’re going to change all that.’ And I was like, ‘I don’t believe you. The licensing, the pricing, the API — I don’t believe you.’ And they said, ‘well, if you don’t believe us, why don’t you come help us fix it?’”

“There were some wild licensing schemes” for satellite imagery, Langan told me in his office at the new manufacturing facility in Goleta. “There would be rules like only one person can look at it, and you can’t extract any information from it. And if you do, you need another license for that. It made it very difficult for folks in the scientific community and commercial businesses to use remote sensing data to solve their problems.”

Morrison talks about a “peace dividend,” i.e. the benefits that the public will enjoy from the military-driven investments in SAR. “You design a constellation to be able to do that incredibly demanding thing,” he said — things like persistent surveillance, custody of high-value targets, and battlefield awareness. “And now you have all this capacity to do other things, like monitor bridges. In the future, there will come a time when no one is ever lost at sea, because if you want to, you can make sure that you can find that vessel from wherever you last saw it.”

“My mental model for that is GPS,” Morrison said. “We created GPS for a defense purpose, and now people use it to chase Pokémon.” For SAR, discovering the “Pokémon” type applications is only possible if the data aren’t just available for purchase but for use.

One of Langan’s favorite unexpected customers in this camp was a marine biologist who used Umbra’s SAR capabilities to monitor penguins in Antarctica. For obvious reasons, it is impractical to sit and watch penguins in situ for long stretches of time, particularly during the winter. Camera satellites can’t monitor penguins as reliably because of clouds. But with high enough resolution, a space-based SAR system can count penguins, track their movements for long periods of time, in the dead of winter, no matter the weather. “I thought that was so cool,” he said.

✺

Jason Mallare leads Mission Solutions — the part of Umbra’s business built for highly customized missions for sovereign military and intelligence customers. He previously worked with Langan at Orbital ATK.

“Most conflicts involve non-human objects being moved strategically, as clandestinely as possible,” Mallare told me. “But they’re trucks, tanks, and boats. And so we focus on metal on the ground and metal on the ocean. It shows up really well in SAR. SAR is all-weather, day and night. You can’t hide under clouds. You can’t hide at night. You can’t hide under a tarp. We penetrate the tarp.”

He described a use case one unnamed partner had shared: that partner was trying to assess whether a particular plot of land might be used for a mobile missile launcher. They were doing full-spectrum monitoring — tracking phones of people potentially scouting the site. “Without SAR, you’d be able to hide,” Mallare explained. “You’d be able to work at night. You’d be able to work when it’s cloudy. And with only a few space systems, if it was not proliferated and not persistent, you could just wait — it goes overhead, okay, everybody come on out, get to work.” Persistent SAR changes the calculus. “Now you hold it in custody and say, okay, we’re pretty sure you’re doing something there, and you’re not going to be able to do it in the shadows. We’re just going to monitor that site constantly, and we’re going to know exactly what you’re doing.”

“Having genuine transparency about where major movements are happening is a good thing for the people and countries that believe in sovereign borders,” Mallare explained. “And it’s a really bad thing for the people and countries that don’t.”

These two sides of Umbra’s business — commercial remote sensing and national security work — turn out to reinforce each other. On the commercial side, the attraction is price and ease of access, even for small-time customers; for government customers where failure isn’t an option, reliability is invaluable. Morrison explained the logic: “A big part of the pricing strategy is to catalyze a new industry, catalyze an entirely new category. At the opposite end, when we put together custom missions for our defense and government customers, that pricing is based on their budget, the value they’re trying to attain. Ultimately, they’re all trying to attain deterrence. Deterrence is driven by persistence. Your adversary needs to know you’re watching all the time. You don’t commit crime under a streetlight.”

New Space

Umbra is cognizant of itself as a “New Space” firm — in which its engineers try to build from first principles, and its strategy is based on radical cost reduction to get new opportunities.

“It’s a common human trope to see something remarkable and then want to template it,” Mallare said. “I think that’s been the plight of the aerospace industry. We started doing something that’s remarkably hard. We went to the Moon, we put a man in space, we put a satellite in space. And we realized it was hard, and had some failures along the way. And we told ourselves, let’s never make that one failure ever again. Let’s document it. Let’s template it. Let’s put a process around it. And we do the same for the next failure. Eventually, you go for decades and you end up with this library of thou shalt not’s.”

“New Space” means throwing out the books, Mallare said. “We’re not going to listen to thou shalt nots.’ We’re not going to start there. We’re going to start from what makes sense. Yeah, what do we need? What do we need? What can we do? What can we figure out? David led the early team that built most of our foundational technology. The transceiver, the amplifier, all the antennas. He’s like, I’m just going to do it. I’m going to launch it. I’m going to figure it out. And then I’m going to iterate.”

Langan, for his part, thought from the beginning that Umbra’s biggest weapon would be the combination of building its own hardware at a much lower cost than larger firms — in the cleanroom that was once a purple classroom, mind you. “In the early days, we were coming from behind,” he told me. “We were one of the last companies competing in the radar remote sensing market to come to market. We did it with less, we did it with fewer satellites. And we knew we had great conviction — we knew we were going to bring a much more capable system that was going to dramatically reduce cost. So, we knew we could use pricing as a weapon early on to force our way in and gain market share. My costs are so much more manageable than my competitors.’”

When a market doesn’t exist, the price of the thing you want is effectively infinite. Before SpaceX made cheap launch possible via Rideshare, itself made possible through reusable rockets, startups simply couldn’t afford to launch small satellites — not at any price that made business sense. Before Apple made the iPhone, a smartphone with the power of a computer wasn’t “expensive”; it just didn’t exist. The same was true of personal computers before they became personal, of countless technologies that went from unimaginable to ubiquitous once someone figured out how to make them cheap enough. Umbra is trying to be one of those companies, making knowledge of the Earth’s surface and the things happening on it…. ubiquitous. And synthetic aperture radar, invented by American engineers, refined over decades of classified aerospace work, is now being offered to the world at prices that would have seemed absurd a generation ago.

Today, Umbra is 11 years old and growing faster than ever. What will the next decade and beyond look like as its constellation and capabilities grow? What would you do if you could see the whole Earth, not just in one pretty photograph, but down to the level of objects.

After four years of war in Ukraine, that dominance has been fundamentally unsettled. The first-person-view (FPV) drone has democratized what was once the preserve of advanced industrial powers. For a fraction of the cost of a cruise missile, even lightly equipped forces can now deliver accurate, long-range strikes at scale. Relentless waves of drones are reshaping armored maneuver, trench warfare, air defense, and even basic infantry tactics. The implications extend beyond Ukraine’s battlefields. These technologies can threaten airports, stadiums, critical infrastructure, and public events within the United States, complicating homeland security in ways hard to anticipate just a few years ago.

Founded in 2018, Epirus is a US-based defense technology company focused on counter-drone and electronic warfare systems. Its flagship system, Leonidas, uses high-powered microwave energy to disable the electronics of incoming drones, rendering them inoperable without relying on traditional kinetic interceptors. I spoke with CEO Andy Lowery, a Navy veteran and former Chief Engineer for electronic warfare at Raytheon, about the lessons of drone warfare, the implications of directed energy for national security, and the dangers of the drone age. What follows is a transcript of our conversation.

CB: I wanted to start with June 2025. Operation Midnight Hammer, the strikes on Iran, which, in my view, show the primes at their best: the B-2, Trident missiles, subs — the Military Industrial Complex in action. Two weeks prior to that: Operation Spiderweb. The Ukrainians launched drones deep into Russia to attack their strategic bomber fleet. What are the lessons, from the perspective of Epirus?

AL: Let’s talk about Spiderweb first, and then there’s another operation that you didn’t mention, Operation Rising Lion. That was an Israeli-run operation which disabled Iran’s vast Air Defense Network. For 200 jets to fly in and not one of them take a scrape and fly out is remarkable, but it was enabled by drones inside of Iran disabling their air defense networks first.

Operation Spiderweb, what we have there is essentially Amazon container logistics trucks, parked around the strategic bombers, open up the tops of these containers, and a swarm comes out. Out come 119 drones, all laden with a bunch of kinetic blast bombs, and they take out 30% to 35% of the strategic bombing fleet. Billions of dollars in damage. These are just Group 1 or 2 quadcopters — absolutely smashing them.

We have an island called Guam that has a bunch of large air defense systems to see what China is doing, and we have defensive missiles and defensive systems that can protect Hawaii and even the United States.

Now we remember the New Jersey drones toward the end of 2024. There were a lot of theories around what that was. One of the theories, a very realistic scenario, would be a container ship owned by China or North Korea, let’s say, with hundreds of containers like they had on the 18-wheelers all over the front of the ship. They could park miles out at sea and be out of visual sight from Guam. They open up, and here they come into Guam.

There is a real necessity for the United States to think very urgently and very quickly. They need to be able to protect the protector: protecting THAAD or Patriot missile batteries, for example. Mini-Golden Domes are needed for drone defense around these expensive radars and interceptors. This is a thing that General Matt Ross, who runs JIATF-401, calls “protect the protector.” JIATF acts as a coordination layer where they’re going to get at these different tier-one types of installation protections using systems, including ours, and a number of layers. In this day and age, that’s the only way you can fight a war and hope not to get taken off the map by a bunch of drones.

CB: Four years into the war in Ukraine, what are the drone threats and how have they developed?

AL: I read a statistic just yesterday that 80% of all casualties of Ukrainians are caused by drones. I read another statistic about a year ago that 40% of the Black Sea Fleet was destroyed by maritime drones. Ukraine destroyed 40% of the Black Sea Fleet with no navy. I mean, similar to 30% of the bombers, like we were talking about with the drone attack on soil. So drones are not just significant. They are almost everything in this war. This type of warfare, leveraging asymmetric, robotic, increasingly autonomous threats is what we call The Sixth Domain of warfare.

There is a no-man’s wasteland between the Russians and the Ukrainians right now that looks like spider webs, because once they realized they can spool out fiber optic, which is, by the way, a prelude to what true autonomy will bring without the wires. The drones are flying with a wire for now, but soon, in the future, they are going to have the same sorts of robustness against electronic warfare and have fully autonomous drone swarms without the cables. And they say, if you get into that dead-man zone and you’re there for too long, you’re done; they’re going to send a drone. So that has really changed the landscape of warfare.

CB: What exactly is electronic warfare?

AL: I was the business area Chief Engineer for electronic warfare for the whole of Raytheon. I had all Raytheon electronic warfare projects underneath me. I had about 100 programs, the capstone one being Next Generation Jammer, a standoff electronic attack system that flies on an F-18 that has been a topic of conversation following the Venezuelan operation and the talk around a “discombobulator.”

Now, the military considers electronic warfare to have three subcomponents, and sometimes they talk about a fourth. The first subcomponent is electronic attack.

Next Generation Jammer is a big electronic attack system. What those systems do is they look for antennas, receivers, things that are using RF and microwave technologies to detect airplanes from far away and so on. And they overwhelm those received channels with a bunch of electronic warfare-type signals. You could be looking at a radar, and you could see 1,000 dots on it. Like I’ve got a thousand things coming at me. That’s putting electronic attack and electronic warfare energy into the system that it’s not expecting and fooling the system into thinking it’s something that it’s not. Or you just blast it so much that it wipes out the whole screen and you can’t see anything. That’s electronic attack.

Electronic surveillance is the next group.

Electronic surveillance is just what it says. It’s listening to the RF environment. Right now, if you put a really sensitive 0 to 6 gigahertz receiver around you that picked up a wide band, it would find around a million or maybe even 5 million pulses per second. You are surrounded by RF energy right now. It’s light, it’s very little, it’s very non-harmful, but it’s all over the place. If you could see RF frequencies, you’d be like, “Oh my God, this place is flooded.”

What electronic surveillance does is take that whole haystack of different RF — it picks up the needles in the haystack. It says, where’s the signal coming from? Who’s this coming from? And this is a very emergent area in the neo-primes, and companies like Hidden Level that have receivers that basically open up and use other emitters, like cell phone towers and other things, in order to bounce that RF energy. Anyway, they take the bounce, do some math, and don’t need any transmitters at all to see what’s going on out there, both physically and then also on the RF spectrum.

The third is electronic protection. In some ways, I think that’s what our form of directed energy, the Leonidas, is. It’s sort of like an electronic protection system. Only sometimes you’re protecting a stadium or an airport, sometimes you’re protecting a vehicle, sometimes you’re protecting a squad. And you have different size systems that would protect at different distances and ranges, different costs and different power levels, so that you could have a larger range around a base and a smaller range around a vehicle.

Cyber has been emerging in the last 20 years. It’s an area where you see a lot of cyber-swarm attacks. Every which way, swarms try to hit and penetrate cyber systems. And it seems like what we have today is “physical cyber.” Drones are like the little software Trojan attacks physically manifesting and providing the same sort of attack vectors in the physical domain as we saw in the software domain for years.

What we specialize in at Epirus is providing a protection layer that can handle many simultaneous attacks. That’s what cyber defense does. It’s the way we have to protect against these physical manifestations, drone swarms. We need to do something that could defend “one to many,” because that’s what’s happening. We’re going to see hundreds or thousands of drones that are overwhelming our traditional air defenses. We weren’t used to having a thousand drones come in at the same time. We didn’t see these kinds of swarms in yesteryear, and we’re not equipped to defend against them today without a Leonidas-type system, because we have limited magazines. An installation may have, say, 10 Coyote missiles. We quickly run out of those. And it’s not even an expense thing. They just don’t have enough.

CB: Could you tell me a little more about Epirus’ Leonidas system and the challenges you anticipate as drones become more autonomous?

AL: First of all, when drones become networked together into autonomous swarms, legacy defenses are going to be completely saturated. One missile to one threat does not scale. Legacy EW will not work on drones that have no signal back to the operator. So, how does Epirus help? Electromagnetic interference forcefields.

In some ways what we’re doing is super sophisticated, but in some ways it’s super basic. On a basic level, what we thought about was how a ton of electromagnetic interference disables commercial computers. All these electromagnetic waves get absorbed straight into the computer board. The board itself acts like an antenna that pulls in all that energy, puts the voltage of energy on the computer board, and then the computer board just goes down. So at a basic level, that’s what Leonidas does. It uses weaponized, or intelligent, electromagnetic interference to be able to put these vast, pulsating fields out there in a volume of space in the sky in order to defend against many drones at once. And then it also rotates 360 degrees around, so it can go around on a slip ring, and then in microseconds it can scan and put energy everywhere, providing a large electromagnetic wall, and it just doesn’t allow any number of drones to get through that wall.

That’s the principle of its operation. Now, we use different frequencies, different pulse patterns, and different things to make this EMI very intelligent, because in the past, what some of the folks that work in traditional directed energy would say is that it’s a narrow-band system, and what they mean by that is it just looks at one frequency or a couple of frequencies, whereas our wide-band systems use a method to cover lots of frequencies all at the same time, all in the same shot. They say, “Well, how is this narrow band going to work?” The fact of the matter is that gallium nitride, this very special substance that’s come along in the last 10-20 years, has really kind of exploded in its ability to handle huge power densities, and by using the state-of-the-art gallium nitride, we’ve created the world’s most power-dense circuits in order to create that very high-powered electromagnetics.

We then use techniques and tools like AI to tune that EMI to be right on target and at great range and make a great defensive layer. The sophistication is the phased array, the signal selection, and the basicness is the fact that we’re just using weaponized electromagnetic interference versus trying to use some, let’s call it, like a death ray or something, which is more the traditional directed energy.

CB: What’s so hard about defeating a fiber optic drone or an autonomous drone?

AL: To defeat a fiber optic drone, you can use something kinetic, you can use a laser. A laser is just a long-range blowtorch. That defeats a fiber optic drone. Typically, how electronic warfare defeats drones is to go into the signals themselves, like the guy has the controller, and he has different frequencies coming out of the controller to control the drone — just to match those same frequencies, but at a higher power so that you’re overriding the signals, and now you can control their drone. Now, the issue with those methods is that fiber-optic drones don’t need signals from a person on a controller that’s sending controls over the air to the drone. There’s no front door. When you see a fiber optic drone flying right over an EW truck in Ukraine, EW isn’t going to work on any fiber optic drone. That thing’s going to go right into the tank every single time. It’s the same when drones use AI to navigate on their own.

But remember when I was talking about what Epirus does. We don’t need a front door. You can close all your front doors, and we’re coming in the back door. The back door is those very susceptible computer boards. Those drones are flying computers, just like a Tesla is a driving computer. And on that computer board are all kinds of electronics that, when we apply our electromagnetic interference, suck up that energy, no matter if it’s being controlled by a fiber optic, whether it’s totally automated, whether it has no signals coming in and out, or lots of signals coming in and out, it treats it the same.

CB: When you look at what’s happened in Ukraine, you’ve seen tanks being marginalized, armored personnel carriers, helicopters, even infantry. I was talking to some guy recently. He said the infantry have to walk like 30 feet apart so they can only be picked off one by one. Is there a case where you could put one of these things on a tank, on a helicopter, have an infantryman carrying it with his platoon? Can you scale it down like that?

AL: Our technology is completely modular and scalable, meaning it can go onto any platform and scale to whatever size is needed to complete the mission.

And the way phased arrays work — and that’s what our system is, an AESA, or an active electronically scanned array, or a phased array — how they work is they have a whole bunch of Lego blocks, we’ll call them. These are amplifiers. We might have 200 of these transmit modules in a system that’s going to protect a base. We might have 75 of these transmit modules for a system that’s going to protect a group of Marines in an expeditionary unit. We might have 20 of these transmit modules in a system that could go on the crow’s nest and protect an Abrams tank with a little golden bubble of electromagnetic interference to stop drones.

As you scale down the number of transmit modules, you’re decreasing the size, decreasing the weight, and decreasing the range, but you’re increasing the portability of it. And so future designs and future efforts, starting in 2026, you’ll start to see released not only the large-size system, which is what most people know us for, but also a medium-sized system and a small system that will be for vehicle, for convoy, for mobile, and then for large, fixed and semi-fixed type applications for the large system.

CB: In the 2026 National Defense Strategy, they mention “a renewed focus on countering unmanned aerial threats.” That falls under the defense of the U.S. homeland. What exactly are these threats? And how are you positioning Epirus to confront this new kind of warfare?

AL: We can only imagine the bad actors that lie as sleeper cells in the U.S. I’m sure they’re still there, despite the great efforts of the administration in trying to weed them all out. There are bad actors out there, and bad actors want publicity, right? They want global publicity. They want to punch America in the eye and have the whole world see it. When you look at upcoming events around America, you’ve got the FIFA World Cup, you’ve got the Olympics, you’ve got these globally visible events that I do think that if I were in the administration, I would be worried about that.

You’ve got borders, you’ve got ports, you’ve got stadiums, you’ve got critical protected sites like the White House or Mar-a-Lago. These are what are called tier-one installations. And what this administration has wisely done is create a Joint Interagency Task Force. It’s called JIATF-401, and its job is to protect the homeland, protect our bases, protect critical infrastructure - by working with each and every service as well as federal and state and local partners.

CB: Are your systems combat tested?

AL: They’ve been deployed twice, once to the Philippines as part of Exercise Balikatan. I can’t say the other place. We’re seeing massive demand from international partners, also. They’re seeing the drone threat in a more concentrated manner in the Middle East and Europe, and, of course, Japan, Australia, the Philippines, and South Korea. It’s a problem everywhere.

CB: Epirus raised a $250 million Series D last year. What are you guys doing with that capital? Are you guys developing new systems? Are you scaling manufacturing?

AL: We’re scaling up manufacturing. We can do at least 24 large systems per year now, but we’re prepared to scale beyond that. We have a family of three different sizes coming out that will feature the latest and greatest electronics in gallium nitride. We have a large, medium, and small offering. All three of those come with their own set of sensors, their own set of C2 software that runs everything, because each application has a much different use case. We sell all of that. We don’t manufacture all of it, but we sell all of that in integrated packages.

Part of the money is to flesh out, if you will, a fully-fledged prime that could act as a prime and take care of the aftermarket support, the field servicing support. All of those different elements are part of the expansion of the $250 million. And then, of course, volume manufacturing beyond 24 per year. We’re looking at Oklahoma. Oklahoma is a great partner of ours.

We have big manufacturing plans, and we haven’t quite decided where we’re going to land, but we have a couple of buildings with over 750,000 square feet available that we’re going to transform into a massive facility where we’ll be making not only Leonidas systems but accompanying electronic warfare systems, sensing systems, and everything else that we need to be a fully fleshed-out prime to do a mission, versus just a part of a mission, or one layer in a mission.

Things are heating up, and we’re getting closer and closer to massive adoption. By 2027, my thought is we’ll be either just a hair away or we’ll be above break even and profitable as an organization.

This team isn’t CIA or FBI — they’re in the private sector. With a product that seems Bond-like, if not fodder for conspiracy theories, Overview Energy specializes in space lasers. Their space lasers are not for a Death Star or to swing sports events by blinding professional athletes at critical moments (it’s come up). It is designed to be utterly harmless. It is not even useful for entertaining big cats.

Overview Energy plans to use lasers to transmit power from space. From a geosynchronous earth orbit (GEO, a term for which Wikipedia uses the legally wrong and therefore confusing definition), 22,000 miles above the Earth, their satellites would continuously collect the Sun’s photons — the particles that carry its light and energy — with a hundred meter long solar panels. Like solar panels on Earth, the photons would excite electrons in the panel, knocking them loose into a channel which flows into lasers. Outfitted with a proprietary optics design, the lasers would deliver power anywhere on Earth within direct line of sight. Transmitting via the near infrared spectrum, the beam would be completely invisible to most lifeforms on Earth. (Overview’s review of biological research indicates only the Nile tilapia, a bottom-feeding freshwater fish in Egypt, may be able to see the near IR spectrum. They have ruled out sending power to the bottom of the Nile River.) From GEO, a satellite could ‘see’ roughly one third of the Earth, so a small fleet would quickly blanket the Earth — if it can find its target and hold the laser on point for long enough. In November 2025, Overview Energy demonstrated their tech can successfully steer a laser from a moving vehicle, track their target, and deliver power from 16,500 feet above. Even better, the only receiver necessary is already plentiful: solar panels.

The problem Overview Energy is focused on is straightforward: today, solar panels can only generate power about half the time, during daylight hours. Engineers attempt to capture just how much the power source is available throughout the year with a measure called a capacity factor, the ratio of the actual electrical energy a power plant produces over a period to the maximum amount of possible energy it could have produced if it ran at continuous full power during that same time. In the United States, solar power’s capacity factor ranges from 23-27%, fluctuating with the seasons, the latitude, and cloud cover. Even with those limitations, solar power is already among the cheapest power sources in the country. If it could generate at any time, solar power could become baseload—continuous, reliable energy. Better yet, a space-based laser could send power all over the planet, wherever it is needed most. If Starlink is a broadband telco provider in space, then Overview is a power company in space.

Overview is the passion project of its founder, Marc Berte. The Bond villain comparisons write themselves: he’s a bald man, obsessed with ‘power,’ who wants to surround the Earth with lasers from space. I can assure you I make the comparison only to tease him: Marc is a friend. (Full disclosure, I have some small exposure to Overview Energy’s success through my wife’s investments.) Ultimately, I want Overview Energy to succeed for my friend’s sake and because of what its potential means for the planet.

It’s a hard job to market ‘space lasers’ for non-lethal purposes. Space lasers in pop culture are synonymous with weapons, a trope the James Bond franchise helped establish. In 1964’s Goldfinger, the eponymous villain planned to execute and humiliate a captive Bond with a cutting edge laser, a technology that was then just four years old. In 1971’s Diamonds Are Forever, Bond faces his first space laser threat: Blofeld creates the immense, diamond-powered “Diamond Satellite” as a weapon of mass destruction that could neutralize any nuclear threat. Smaller space lasers would feature in 1979’s Moonraker, inspired by Star Wars and featured Space Shuttles and Marines armed with laser pistols and rifles. In 2002’s Die Another Day, villain Gustav Graves would threaten the world with the Icarus satellite, a solar-powered laser. (1995’s GoldenEye was a satellite weapon, but it generated an EMP not a laser.)

Sadly, reality is sometimes much dumber than fiction. Space lasers have been the subject of conspiracy theories that I would rather not repeat. While the practical use of space lasers may sound outlandish, they are real tools that Marc believes are poised to reshape the energy industry.

Overview Energy is not building a weapon. Moreover, Overview intends that their satellites cannot conceivably cause harm to anyone. In engineering terms, their laser is passively safe by design. In Marc’s words, “Passive safety means there is nothing a user or program has to do to make it safe.” Comparing it to children’s toys, Marc claims, “A lot of design effort prevents kids from permanently harming themselves”. For example, passive safety is like a pool noodle, safe by default even when attempting to wield it as a weapon. Active safety is like a pool drain to prevent small limbs from getting stuck, requiring system-level design (multiple redundant drains) and features (anti-vortex covers, automatic shut-off systems, and safety vacuum release systems) that keep the unit safe. Marc cites previous attempts at microwave-based power transmission that relied on active safety systems as a cautionary tale. The power of microwave systems is “way above the safety threshold” requiring a system that would “keep people out of the beam or shut the beam down if something transits the beam.” To illustrate the challenge of active safety for a microwave, imagine a home microwave with the door off. It would be extremely unpleasant to stick your hand in while warming up a burrito. To be considered safe, the system would need a way to detect if something is in the path of the beam and, if so, disable the beam. That system of detection and disabling is active safety, and it only gets harder to ensure the reliability of an active safety system when the microwave is no longer a box at home but moving in space and pointed at Earth. Some common commercially deployed laser systems do rely on active safety. Many autonomous driving systems rely on LIDAR, which spins a laser. Because the laser is spinning, it would not deliver enough power to hurt a human eye. However, the laser is strong enough that if it stopped spinning with line of sight to someone’s eye, the amount of energy delivered per laser could damage the retina. As a result, Waymo’s control system has an interlock that activates to disable the laser if the spinning stops to prevent injury.

As directed energy weapons — the category of weapons that intentionally destroy or damage with energy in the form of microwaves and lasers—have moved from science fiction to possibility, I ask about the ballpark of the intensity of Overview’s laser to a military-grade one. Marc indulges me with “napkin math”, which is supposed to be rough estimates to check your intuition. “If you want to damage things, melt stuff, blind a sensor, physically damage stuff, the easiest way is to assess the melting temperature of the target material in question. In space, we compare it to how efficiently the material can radiate energy away. 1500 °C will melt steel. That requires radiating 450,000 watts per meter squared of the melting steel, rounding off emissivity. Weapon lasers are in that ballpark. If Overview’s power transmission system is running at peak capacity, it would generate 350 W per meter squared or about 1200 times less than a weapon that can melt steel.”

Instead, Overview Energy aims to be the pool noodle of space lasers by designing their transmission system to be inside the safe exposure for lasers. “You can stand in the beam, stare at it, spend eight hours a day in it for the rest of your life, and it’s fine,” Marc told me. That may establish the beam is safe, but it also must be effective. “Because the beam is relatively low power, you can now make a small aperture with a small wavelength, which is cheap. So you can make lots and lots of them. It’s better to mass produce one design than make one big satellite. With small, attritable systems, it’s much safer, it’s resilient to damage, and you don’t care if you lose one.” To summarize, in order to deliver sufficient power, the system relies on many satellites positioned with line of sight to the target that can beam power simultaneously. Each beam is designed to be safe both individually and collectively.

Understanding why generating solar at night is valuable is straightforward: money. Understanding why lasers are the best tech to transmit power requires some background. Sputnik-1 kicked off the Space Age in October 1957; the first working space laser was built soon after it in May 1960, when it was still capitalized as LASER to represent the acronym “light amplification by stimulated emission of radiation.” Four years later, NASA would use its first laser, GODLAS (standing for Goddard Laser), firing from the ground to check the range of satellites in space. On December 11, 1965, Lt. Frank Borman and Comdr. Jim Lovell (of Apollo 13 fame) demonstrated the first in-space use of a laser to attempt transmitting data, known as the Gemini-7 test, though their experiment was only a partial success. Their task was to essentially hit a receiver, in this case think of it as a telescope but instead of producing an image, it produced a signal that can transmit a wave. Their voices would be embedded in that wave. Using a laser made by the Radio Corporation of America (RCA), once a titan of all kinds of American goods that is now defunct, the astronauts hit their target, but they could not successfully track their target. The contact between the laser and the target was too intermittent to successfully transmit any information. It is hard enough to hit a target from space, and the task is complicated by attenuation through the atmosphere.

In the decades since, control systems have been refined and laser usage has grown exponentially for space-to-space laser data transfer. Starlink, SpaceX’s space-based internet service, passes petabytes per second of data between its constellation of satellites with laser-powered datalinks. In 2025, most space-to-ground datalinks use radio frequency (RF) that are more forgiving with precision and weather but limit how much data can be transmitted. Projects like Laser Communication Relay Demonstration (LCRD) at the Lincoln Laboratory at the Massachusetts Institute of Technology, NASA’s TeraByte Infrafred Delivery (TBIRD), military mesh networks like Space Defense Agency’s Proliferated Warfighter Space Architecture, and products like Cailabs Optical Ground Stations show a future of data transmission via laser to Earth.

While Overview Energy wants to transmit power, not data, the principles of successful lasing are the same: their control system must be able to keep the laser on target to ensure the photons get to the solar panels where they are useful rather than the surrounding dirt where they are not. A laser power transmission system, working successfully, starts with photons ejected by the sun hitting solar panels attached to the satellite in space. The solar panels do exactly what they do on Earth, absorbing the energy from the photons to excite electrons that flow to the laser. The laser converts electrons back to photons, except now that beam can be aimed. In ideal conditions, the beam could be aimed directly at a solar panel on Earth. Lasers do suffer from attenuation through the atmosphere, and significant cloud cover can completely block the beam. But being in GEO means that so much of the Earth is in view that a small change in the satellite’s aim could hit another solar plant where the weather isn’t. The physics are constant, but Overview Energy exists because progress in lasers, space launches, and solar panels have utterly transformed their economics in a way that other wireless transmission technologies have not.

Microwaves are much more forgiving in power transmission than lasers. They do not require precise alignment, and they can penetrate cloud cover. Satellites with solar panels and microwave transmitters could transfer power, but not safely. Every concern from a household microwave applies: it could slowly cook you or start fires if it heats metal beams. The danger grows the tighter the beam is. The Federal Communications Commission enforces maximum permissible exposure (MPE) which limits RF (radio frequency) energy to ~10 watts per meter squared. To transmit safely, the beam would be bigger — much bigger. Transmitting 1 gigawatt passively safe per FCC regulations would require 100 million square meters — or 38 square miles — of continuous ground receivers. And unlike solar panels, which are cheap and plentiful, microwave ground receivers are expensive rectifier antennas that generate no power without an active beam.

Another option is… just deliver more light with giant mirrors in space, specifically in low earth orbits. A startup called Reflect Orbital is attempting exactly this — positioning satellites with large mirrors to bounce sunlight to solar farms after dark. But in Marc’s words, “the issue with mirrors is, the more you look at it, the more problems you find.” For reflected light to be useful to a solar panel, it has to already be night on the ground. As a result, each satellite would have to synchronize with the sun, constantly chasing the terminator — the line between day and night on the Earth. The satellite’s altitude in LEO also poses issues. Mirrors can only add power within a few hours before sunrise or after sunset. Moving into higher orbits allows the mirror to go deeper into the night time, but makes the beam size “gigantic”. Imagine you are floating 500 km over Hawaii, like our Gemini-7 astronauts, but now with a mirror. Hold the mirror parallel to the surface — if it’s not parallel, the beam will form an oval instead of a circle. The tightest beam of light hitting the earth is 5 km or 3 miles across. At higher orbits, the spot size grows to 10-20 km or 6 to 12 miles across, drastically reducing the power density of the beam. The result: you’d need hundreds to thousands of satellites, all crowded into a sun-synchronous orbit. Managing their operations and deorbiting malfunctioning satellites without colliding into others poses a significant engineering and operational challenge. And then there is the light pollution. Stargazing, scientific telescopes, and the creatures that rely on light-based REM cycles (including humans) would suffer with visible light spraying the night sky.

In Overview Energy’s case, the necessary technologies to beam energy onto Earth are all improving dramatically and continuously. Potential customers like hyperscaling AI data centers and even power utilities are desperate for power, especially cheap and clean power. Solar farms are plentiful, so no net new land usage is required. According to Marc, space-based solar power was “kind of a thing that was always the energy source of the future. Nobody could make it work. And if a thing didn’t work, I want to understand why, then see if the situation has changed. If it has, do that thing.”

Lasers improved substantially in the early 2000s, thanks to a DARPA initiative called SHEDS (Super High-Efficiency Diode Sources). Until that research program, laser efficiency ranged from 20-50%. DARPA hoped to cut the energy requirements for high-power lasers, with a “DARPA hard” goal of 85%. Through many optimizations of the design and manufacture of lasers, DARPA brought laser efficiency to 60-70% at room temperature, addressing a longstanding issue to broader laser use. Improving from 70% to 85% required active cooling. In Marc’s words, “cold laser, good laser”. Those efficiency improvements were attainable, but not economically. The power to run the cooling system was more than the power saved running the laser colder, so the industry had to settle for merely triple the laser efficiency as before.

Notably, Overview Energy’s spacecraft design operates in a much colder environment. One side always faces the sun and the other faces the vacuum of space—a heat gradient which allows for ultralightweight radiators and sufficient space to cool the lasers effectively. Marc believes their beam size in GEO will be 2 to 5 km (1.2 to 3.1) miles across and reach MW-class per beam.

The cost for Overview Energy to get a satellite to orbit has fallen significantly, due to the pioneers of the modern space economy, SpaceX. Their reusable rockets continue to lower the cost of launching mass to orbit through manufacturing lots of rockets and launching and relaunching them as often as they can. That virtuous cycle drives down the cost of each launch, savings SpaceX uses to offer the lowest prices to reach orbit. In December 2025, Falcon 9 rideshare missions to LEO cost roughly $6500 per kilogram, an increase from a low of $5000 per kilogram before inflation drove up costs throughout the industry. SpaceX believes their fully reusable Starship and SuperHeavy rocket together could drop launch costs another 10x; Marc believes Overview Energy can be profitable at a cost of $1000 per kilogram to orbit. Only Elon knows how much prices may follow these costs, but for context, proponents of space-based data centers assume costs will reach $350-500 per kilogram in the near future to make their designs competitive with ground-based data centers.

If lowering launch costs for other space companies was not enough, SpaceX has also demonstrated another critical milestone — satellites can be mass manufactured. Before Starlink, satellites were frequently one-off products. Starlink demonstrated that the design requirements for space are within reach of existing, cheap components used in consumer electronics. While the space environment’s temperature swings between -100°C and 260°C, the electronics inside satellites don’t experience those extremes. Interior chambers can be sealed and maintained at consistent temperatures. As Marc explains: “If you compare it to automotive electronics, the temperature range is more manageable. The same pickup truck needs to operate in both the constraints of Saudi Arabia and Alaska. Space can be designed to be one, fairly constant environment that can be designed for. The tradeoff is that, once you put it up, you can’t fix it if anything breaks.” By proving satellites can be commodity products, engineers and investors now have a reference point to design and invest in other kinds of mass-produced spacecraft.

Which brings us back to the Gemini-7 problem—tracking the ground receiver successfully—and explains why the Overview Energy team was cheering in the dark. “Tracking is the hard part,” Marc told me. Overview had to demonstrate they could build an optics and control system that could hit a stationary target from 3 miles above and hold the beam on that target, a lightweight version of the challenge that ultimately flunked the Gemini-7 laser tracking test in 1965. “Integrating complex systems is hard,” Marc explains. “Most power beaming demonstrations to date are fixed point to fixed point, with about 1 part in a 1,000 from a beam reaching a drone a mile away. To hit a receiver from GEO, you need a one kilometer wide receiver [in this case, a solar farm] with a beam within a few hundred meters so it’s not wobbling all over the place.” This demonstration tested their entire control system, including the ground-based trigger. “The receiver has a beacon with upward looking infrared. The satellite uses that beacon to lock onto the location and transmit back down. The airborne test demonstrated that, using the same method, same lasers, and same infrared beacon that we would use in GEO.”

Marc has been training all his life for this challenge. How does a man choose space lasers as a profession? In the 1990s and 2000s, it was not for the money but for the thrills. Both of the traditional sources of funding for space tech were running dry. As the Space Race wound down, NASA’s budgets were under constant pressure; the end of the Cold War ushered in the Peace Dividend, a reduction in military spending. But Marc knew that he wanted to pursue a career at the intersection of aerospace and nuclear engineering. Growing up at the end of the Cold War, Marc was enamored with radical proposals, such as the Strategic Defensive Initiative proposal to position space lasers to intercept intercontinental ballistic missiles. More fundamentally for this aspiring Q, Marc wanted to be at the edge of technology and applied research. The farther out at the edge, the more fun. But technology has many frontiers, so Marc doubled-down on following the action to the highest energy levels: rockets, nuclear reactors, wireless power transmission, and all the components that made those practical. Marc double-majored in aerospace and nuclear engineering at MIT, studying small modular nuclear reactors (SMRs) and the potential for fusion energy. He completed his double-major in just three and half years.

Working at the edge of a field is more like pioneering than a gold rush, and many that wanted to venture into these domains relied on adventurous engineers like Marc to find a path and keep them on it. His professional career included time at the Institute for Defense Analysis working on proposals for space lasers to ward off intercontinental ballistic missiles, consulting for DARPA, developing optics and lasers for the Missile Defense Agency and Raytheon, and consulting for many startups and companies developing SMRs. In 2017, SpaceX’s reusable Falcon 9 rockets dramatically lowered the costs of launching mass to orbit, essentially making it cheaper to reach the frontier and enabling new kinds of businesses. Investors and entrepreneurs would also consult with Marc on their path through this new frontier, and he still serves as a friendly voice to help them on their path. But as he helped others find their path, Marc found one of his own. Based on old ideas about what might be possible with space-based power, Marc thought the time had finally come.

Overview’s next milestone is slated for 2028: beaming from LEO aboard a booked SpaceX flight. This milestone will test the final optics design, validate the design is suitable for space, and demonstrate the tracking control is close to the target needed for optimal economics. “The optics for the LEO satellite are 5-10x bigger than the airborne one, but if the input wiggle is constant and the optics get bigger, the output wiggle is smaller. The same amount of control as we demonstrated in the airborne demo gets a 5-10x improvement with our bigger optics.” Improvements to the control system are all about steadiness. From GEO, Overview will need to demonstrate control within about roughly the angle between your eyes when you’re looking at a coin from a mile away. Now imagine keeping a beam of energy locked on that coin while both you and the coin are moving. The optics alone get them about halfway to that level of control, and other improvements in the LEO design are expected to close the gap.

Marc sees other challenges ahead. The new space industry is hot, but Marc stresses that Overview is here to build cheap, clean power, not hype. They are building for a future of cheap launch costs that SpaceX and others are racing towards, but is not here yet. Few investors can evaluate both space tech and energy, making fundraising complex. While AI is driving a revolution in engineering on the ground (I asked: engineers at Overview Energy use Claude), Marc still sees talent as a constraint: “Can you get the right people at the right time and enough of them that are passionate about what they’re doing? The hard part isn’t making something work at all costs, it’s making it work reliably at cost, on time. We need people that can make space lasers boring.”

Layton served as the chief intelligence officer for Admiral Chester Nimitz, the Commander in Chief of the U.S. Pacific Fleet, and Commander in Chief of the Pacific Ocean Areas’ naval fleet during the Second World War. In the spring of 1942, American Signals Intelligence (intelligence derived from electronic signals and systems used by foreign targets) — SIGINT — intercepted Japanese communications indicating an imminent offensive against a target referred to only as “AF.” The United States had access to the signals, but lacked certainty about their meaning.

That distinction between access and understanding sits at the heart of intelligence work and is often misunderstood outside it. Signals Intelligence is frequently imagined as a technical feat: codebreaking, interception, and collection at scale. In reality, those capabilities are necessary but insufficient. Signals only become intelligence once a human institution decides what they mean and what to do with them. The defining challenge of SIGINT has always been interpretation.

At Midway, American military analysts debated what “AF” referred to. Some believed it was Hawaii. Others suspected a location in the Aleutians. Acting on the wrong interpretation would have exposed the U.S. Navy’s remaining aircraft carriers to catastrophic loss, effectively ceding control of the Pacific to Japan. There was strong consensus in the Navy brass not to challenge the bureaucracy and not to speculate on alternative theories. It is very hard to project doubt against an admiral or general’s expertise, especially as a junior ranking officer.

The resolution came from analytic judgment and trained minds ready to interpret, doubt, and govern the signals — not from additional collection or more sophisticated cryptography. Consider how this played out at Midway. Edwin Layton and his team understood something fundamental: signals do not exist in isolation. They exist in adversarial environments. The Japanese were listening too — to American intelligence. So as a test, the Americans transmitted an unencrypted message stating that Midway Island’s freshwater condenser had broken and the garrison was running dangerously low on water. Shortly afterward, Japanese intelligence traffic reported that “AF is short on fresh water.” The ambiguity collapsed. AF was Midway, and the course of the war was bent by a few military nerds hunched over primitive machines.

That act — testing an interpretation rather than trusting it — allowed the U.S. Navy to concentrate its forces and ambush the Japanese fleet. The Battle of Midway, which happened three months after Layton’s team decoded “AF,” altered the course of the Pacific War. It is often remembered as a triumph of interception, the covert acquisition of signals and electronic emissions, or cryptography, the breaking of coded communications. But it was greater than both of that; it was a triumph of disciplined doubt. The codebreakers proved themselves right in rejecting the common belief that AF referred to Hawaii or the Aleutians.

In the U.S. military and intelligence community, SIGINT is one of several foundational intelligence disciplines, alongside human intelligence (HUMINT), open source intelligence (OSINT), imagery intelligence (IMINT), and a host of others. Each exists to reduce a different kind of uncertainty. SIGINT focuses on communications and electronic emissions: radio transmissions, radar signals, satellite links, digital traffic, and the metadata that accompanies them. The National Security Agency (NSA) is the country’s primary SIGINT organization and practitioners throughout the military and intelligence community mark their work with logos, slogans, and patches denoting the ways in which we “are always listening.” Military organizations prefer skulls and other terror-inducing mottos. But perhaps nothing is more frightening than knowing in an Orwellian sense that everything you are saying can be heard and tracked.

In our hyper-technological age, digital enemy communications are the most critical and valuable type of intelligence, often classified at the Top Secret level. Due to both the scale of data produced and the sheer speed with which we communicate now, the 21st century has thrust SIGINT to the forefront. Our tools and tactics may have evolved over time, but the basic discipline — “trust but verify” — remains the framework that allows us to interpret what is real, what is fake, and, most importantly, how to act on it. Signals have changed dramatically since 1942, as signals data now flows in petabytes rather than Morse code dots, but its disciplined analysis holds its core value.

Institutional Overconfidence

Among the many concerns raised about artificial intelligence — bias, fairness, ethics, robustness — there is one that deserves sharper attention: the risk not simply of hallucination or error, but of institutional overconfidence arising when large systems produce authoritative-sounding outputs without clear human accountability. This is playing out today: the Bank of England and creditors in the UK are using AI to make creditworthiness decisions for consumers, often accepting the algorithm’s output without meaningful review. In fact, I’d argue institutional overconfidence is the most dangerous failure mode of AI. It’s particularly bad when systems that feel authoritative obscuring who, exactly, is responsible for interpreting their outputs.

Midway reveals why this matters. The U.S. Navy could only act decisively because Layton’s team not only intercepted enemy signals, but owned the interpretation, tested hypotheses, and insisted that uncertainty be explicit. In contrast, many modern AI systems produce conclusions with persuasive fluency while hiding uncertainty, sources, and interpretive context.

Recently, in the civilian world, a passenger relied on Air Canada’s website chatbot for information about bereavement fares. The chatbot confidently told him he could apply for a refund after travel — directly contradicting the airline’s actual policy. When he sought the refund, Air Canada refused and even argued that the chatbot was effectively separate from the company. A tribunal rejected that logic and held the airline responsible for what its bot had said. The failure was not that the chatbot was wrong, but that no one could identify who was responsible for its answer. This creates a psychological and institutional trap: users and organizations begin to trust the answer without considering its provenance or limitations. Air Canada is just a civilian airline, but it’s also a typical bureaucracy struggling to understand how to integrate AI into decision-making.

Now consider a military analogue. Russia has been publicly rolling out an AI-enabled command-and-control decision-support system called Svod, explicitly designed to fuse disparate intelligence inputs into a single operational picture for commanders and speed the decision cycle. Russian reporting describes Svod as giving officers real-time access to maps, satellite imagery, weather, and “air and ground situation” data inside a protected information space, with trial “combat operations” reportedly conducted in at least one troop grouping and a plan for broader implementation. Other reporting describes Svod as aggregating inputs from satellites, aerial reconnaissance, intelligence reports, and open sources, then using automated analysis to help commanders choose courses of action under battlefield pressure. But when Svod recommends a faulty course of action, who is responsible? The commander who followed it, or the system designers who may not fully understand Ukrainian deception or spoofing tactics?

This is exactly where institutional overconfidence can emerge: a fused, authoritative-looking “single picture” can feel like consensus even when it’s built from correlated, incomplete, or adversary-shaped signals — and when the system’s uncertainty (or the fragility of its inputs under electronic warfare and deception) isn’t made explicit, responsibility for interpretation quietly shifts from accountable analysts to an algorithmic dashboard.

The risk of institutional overconfidence appeared in intelligence history long before the age of neural networks. There is the famous idea in statistics: finding the “signal through the noise,” but I fear we’re entering a world where everything appears to be a meaningful signal, and we take those signals for action. Intelligence is the process of going that one step further to refine the signal until we have an answer, beyond a reasonable doubt. A functioning apparatus of humans, along with procedures to check their judgment, is essential in building out AI-forward institutions.

When Interpretation Falters: Lessons from SIGINT History

In November 1983, throughout various European countries, NATO conducted Able Archer 83, a routine exercise involving simulated command post activity and nuclear release procedures. To Soviet observers, however, the signals produced by the exercise looked strikingly like preparations for an actual first strike. Intercepted communications and unusual radio traffic — harmless to Western planners — were read through a lens shaped by recent Cold War tensions, like the massive increase in encrypted communications between the US and the UK. (These ciphered comms were actually a discussion about Grenada, but the sudden increase was abnormal for Soviet listeners)

Listening in, the Soviet leadership grew spooked by how realistic the exercise appeared. Coded communications and radio silences — standard military procedures during exercises — were read as genuine operational security measures. (The US military now often announces “exercise, exercise, exercise” before simulations to prevent precisely this kind of confusion.) Soviet institutions, perpetually paranoid and convinced that the threat was real, responded in turn: calling up forces, dispersing nuclear assets, and scrambling fighter jets.

The result was not a clear picture of NATO’s intentions, but intensified fear and misinterpretation. The world nearly stumbled into nuclear war — neither side wanted war, but ambiguous signals were interpreted with excessive confidence and insufficient skepticism. In the absence of mechanisms for adversarial hypothesis testing (asking “could this be an exercise?”), institutional assumptions hardened into existential interpretations. No single Soviet analyst “owned” the judgment that these signals were just an exercise; confidence grew through collective momentum rather than critical evaluation.

A decade earlier, Israeli intelligence faced the mirror image of this problem. While the Soviets saw threats where none existed, Israeli analysts dismissed genuine warnings because they contradicted institutional doctrine.

In October 1973, Israeli intelligence received multiple indicators suggesting that Egypt and Syria might attack. The signals were there — SIGINT, human intelligence, and other data like troop movement, leadership locations, and unique press articles — in retrospect, unambiguous. What went wrong was interpretation. The Israelis believed that Egypt would not act until they had secured air superiority through Soviet aircraft, and this assumption blurred the concrete information intelligence had gathered.

Back in 1972, the Israelis had secretly connected listening devices to Egyptian communications in the Sinai. To avoid detection, the devices remained off unless IDF command gave an order to activate them. During the build-up to the surprise attack in which the Egyptians and Syrians launched a coordinated offensive against Israel, controversial General Eli Zeira, who ran the intelligence directorate, refused to turn on the system. When he eventually allowed a ten-hour collection window, it picked up nothing unusual, and he ordered it switched off. By doing so, he missed the crucial signals broadcast in the hours that followed.

Zeira was following the prevailing assumption that Egypt would only attack once it had Soviet aircraft capable of providing air superiority. This thinking was doctrinal, and the institution failed to challenge it. Though the Israelis had the means to collect further intelligence — and act on it — they squandered both. On Yom Kippur, October 6, 1973, the holiest day in the Jewish year, Egypt and Syria invaded.

These failures underscore a core insight of intelligence practice: data does not interpret itself. Interpretation reflects assumptions, and unchecked assumptions can drown out valid signals.

Signals and AI Today: What Modern Research Shows

The analogies between SIGINT failures and modern AI are not merely rhetorical. Research into large language models (LLMs) reveal concrete manifestations of the same interpretive risks that plagued Cold War intelligence analysts.

One of the most studied failure modes in LLMs is hallucination, where the model produces text that is fluent and plausible but factually incorrect. Hallucination arises not from malice but from the training process itself: models are optimized to produce coherent completions, not to admit uncertainty. As researchers at OpenAI explain, LLMs frequently “guess when uncertain, producing plausible yet incorrect statements instead of admitting uncertainty”— a dynamic rooted in how they are optimized for performance rather than reliability.

Comprehensive surveys of LLM behavior confirm that hallucination is widespread, persistent across model families, and context-dependent. Even state-of-the-art systems can still produce highly confidently wrong outputs when given carefully grounded prompts.

Hallucination is not just a technical nuance. In critical domains like healthcare or law, even rare confident falsehoods can have disproportionate consequences. Research classifying hallucination types in medical and life sciences contexts finds that models produce outputs that are “factually incorrect, irrelevant, or misleading” — all of which undermine trust and efficacy in high-stakes environments. The parallel to SIGINT is clear: like intercepted signals, LLM outputs can be abundant, coherent, and superficially persuasive while masking fundamental uncertainty. In the SIGINT world, interpreters hold security clearances and receive formal training; with AI, these systems are deployed on a civilian populace and legacy institutions that lack any framework for systematic doubt or verification. We have millions of GPUs and the greatest hardware mankind has ever built, yet we don’t have the human software in our own gray matter to match the task.

And the problem is getting worse. As language models become more sophisticated, they become better at producing fluent content that feels authoritative — regardless of whether it’s accurate. Scholars call this AI trust paradox: advancements in fluency make outputs harder to distinguish from truth, even when they are misleading.

This paradox reflects the same dynamic that undid analysts before the Yom Kippur War. Confidence in surface coherence masked deeper uncertainty, leading institutions to treat preliminary assessments as confirmed intelligence. With AI, the trap is structural: every response we get is not an answer but a probability-weighted directional guess at what may or may not be correct.

What distinguishes successful SIGINT practice is not raw data collection, but interpretive discipline. Intelligence analysts are trained to articulate uncertainty explicitly rather than obscure it, test alternative hypotheses rather than converge prematurely on a single interpretation, expect adversarial behavior, and anchor every conclusion to specific human judgment.

Raw data collection matters far less than interpretive discipline. These practices reflect hard lessons from Midway, Able Archer, and Yom Kippur, forged through catastrophic near-misses and actual failures. The contrast with AI is stark: when systems produce outputs that feel right, organizations may stop asking why they feel right, who is responsible for verifying them, and what assumptions underlie them.

Another growing body of AI research focuses on calibration: whether a model’s stated or implied confidence matches its actual likelihood of being correct. Across domains, researchers find that large language models are often miscalibrated — particularly in complex or unfamiliar tasks. Models express high confidence when wrong and uncertainty when right, a pattern that worsens under distribution shift.

From an institutional perspective, this is dangerous. Humans are notoriously deferential to confidence cues. When AI systems produce decisive-sounding answers, organizations treat them as authoritative — even absent any mechanism to verify or challenge them. Over time, responsibility migrates from human judgment to system outputs, creating precisely the kind of diffuse accountability that intelligence organizations worked for decades to avoid.

In SIGINT, analysts are trained to separate confidence from certainty. In AI deployment, that distinction is often erased.

Feedback Loops and the Collapse of Independent Signals

Research on model feedback loops reveal another parallel. As AI-generated content proliferates online, models trained on scraped web data risk learning from their own outputs. This phenomenon, called model collapse, reduces diversity, amplifies errors, and narrows the space of possible interpretations.

Intelligence agencies learned that relying on self-referential signals is a recipe for analytic failure. Independent sources matter not because they add volume, but because they provide friction against shared error. A canonical example is the US intelligence failure surrounding Iraq’s alleged weapons of mass destruction in the early 2000s. Much of the analytic confidence rested on a small number of sources — most notoriously a single human source codenamed Curveball — whose reporting was repeatedly recycled, summarized, and corroborated internally without truly independent verification. AI systems, by contrast, often present internally consistent but externally ungrounded consensus. Multiple “answers” may in fact be variations of the same underlying model judgment. A well-documented example occurred in US federal court in 2023, when lawyers in Mata v. Avianca submitted a legal brief citing multiple prior cases that appeared to corroborate one another. In reality, every cited case had been fabricated by a large language model. The illusion of consensus came from the fact that the model generated several internally consistent but entirely fictional precedents, each reinforcing the same underlying narrative.

Why Technical Fixes Are Not Enough

Much of contemporary AI safety research focuses on improved training, better benchmarks, and mitigation protocols. This work is useful, but it is insufficient on its own to address the problem of interpretation. We need technical and institutional training and controls. Technical enhancements for hallucination detection, uncertainty calibration, and adversarial mitigation can reduce certain errors, but they cannot create interpretive accountability.

Consider the extensive research on hallucination detection and reduction — token-level uncertainty estimation, retrieval-augmented generation, and automated fact-checking frameworks. These improve output quality, but what they cannot do is assign responsibility for interpretation. A more accurate model is just a better signal generator.

In intelligence work, an intercepted signal initiates a structured process of analysis, debate, dissent, revision, and formal assessment. Humans are known to perform better with checklists and clear steps, which is exactly what this process encourages. AI systems collapse this process entirely, presenting outputs directly to users without clear chains of human judgment.

Laplace’s Demon

In the early nineteenth century, the mathematician Pierre-Simon Laplace imagined a hypothetical intelligence — later dubbed Laplace’s demon — that knew the position and momentum of every particle in the universe. Given perfect information and sufficient computational power, this intelligence could predict the future and reconstruct the past with complete accuracy. Nothing would be uncertain; nothing would surprise.

Laplace’s demon is a fantasy, but it remains seductive. In defense and intelligence work, sensor fusion and data integration pursue the same dream: a complete picture of the battlespace or information environment. The difference is that defense institutions have built doctrines around accepting fundamental limits to knowledge. Analysts are trained to operate under uncertainty, to distrust perfect clarity, and to assume that adversaries are manipulating signals.

I fear that the private sector has no such muscle of interpretation. Instead, there is a pervasive assumption that more data and faster models will eventually converge toward omniscience—that the demon is achievable with further scale and capital.

Signals intelligence disproved this long ago. No amount of collection ever eliminated uncertainty. In fact, beyond a certain threshold, additional signals increased confidence faster than they increased understanding. Comprehensiveness was a false idol. This is precisely what is happening with AI today: confidence in model outputs is increasing faster than our interpretive capacity to validate — or invalidate — them.

Emergence, Opacity, and the Limits of Interpretability

Contemporary research on LLMs reveals a troubling paradox: as systems grow more capable, they often become less interpretable in human terms. Studies of emergent behavior show that models develop capabilities — reasoning patterns, internal representations, strategic behaviors — that were neither explicitly programmed nor anticipated by their designers. These behaviors appear abruptly at scale, defying linear expectations.

Interpretability research has made impressive progress in mechanistic interpretability, sparse autoencoders, and feature attribution methods, but even leading researchers acknowledge a fundamental gap. Understanding a model’s internal representations does not translate into understanding why a model produces a particular output in a given context. Insight into the machinery does not guarantee insight into the judgment.

This mirrors a core challenge in SIGINT. Analysts may understand exactly how an intercept was collected, decrypted, and processed, yet still struggle to assess what it means — especially when adversaries deliberately shape signals to mislead. AI systems are not (yet) adversarial in this sense, but they create the same interpretive problem.

Where SIGINT is troubled by deception, AI is troubled by compression. AI systems present outputs as complete narratives. Internal uncertainty, conflicting latent representations, and probabilistic tradeoffs are flattened into fluent text. The result is not mere opacity but a false sense of coherence — outputs that read as definitive answers while concealing ambiguity. This is Laplace’s demon in modern form: the fantasy that enough computation can resolve ambiguity and eliminate the need for interpretive judgment.

Why Laplace’s Demon Still Haunts AI

The enduring appeal of Laplace’s demon is not that it seems achievable, but that it absolves institutions of judgment. If outcomes can be computed, responsibility shifts from decision-makers to systems. Failure becomes a technical malfunction rather than a result of human error.

Signals Intelligence rejected this fantasy out of necessity. Adversaries adapt. Context shifts. Meaning remains contested. No system, however powerful, escapes the need for interpretation. Rather than solving this reality, AI obscures it.

The lesson from intelligence history — and from contemporary AI research — is that interpretation cannot be automated away. Systems can surface signals, identify patterns, and reveal possibilities. They cannot own the consequences of acting on them.

Laplace’s demon promises certainty; SIGINT teaches restraint. The choice facing institutions deploying AI is which tradition they will follow.

From Listening to Hearing

We have built systems that listen everywhere. They monitor vast datasets, generate summaries, predict outcomes, and recommend decisions. The harder challenge is learning how to hear — to interpret with discipline, acknowledge uncertainty, and anchor every judgment in human accountability. The future of AI will be determined not by advances in model performance, but by whether institutions develop the interpretive discipline that intelligence organizations forged through decades of failure and near-catastrophe.

The stakes are existential. The next collision between AI and SIGINT may come when the Chinese PLA operates in the Taiwan theater. Will exercises in the Taiwan Strait be mistaken for all-out war? Will algorithmic pattern-matching mistake mobilization drills for invasion preparations? Will AI-enhanced intelligence systems produce seemingly confident assessments that obscure ambiguous signals? Our immediate challenge is not simply to make AI smarter but to make our institutions wiser — to establish interpretive discipline. Organizations using AI need structured processes borrowed from intelligence tradecraft like alternative analysis and red teaming.

Alternative analysis is a formal intelligence practice designed to prevent premature consensus. Rather than pushing analysts toward a single “best” explanation, it requires the deliberate generation and sustained consideration of multiple competing hypotheses drawn from the same underlying signals. Analysts are expected not only to state what they believe is happening, but to articulate credible alternatives and identify what evidence would support or falsify each one.

This discipline exists because intelligence organizations learned that agreement often emerges before understanding. Time pressure, institutional incentives, and persuasive signals can all drive early closure. Alternative analysis introduces friction into that process. It slows judgment at precisely the moments when confidence feels highest.

Applied to AI-enhanced systems, alternative analysis would mean resisting the impulse to treat model outputs as default interpretations. Instead of asking, “What does the system conclude?” institutions would ask, “What else could explain this pattern — and how would we know if the system is wrong?” Competing interpretations are not a failure mode; they are a safeguard.

Red teaming is the intelligence community’s way of institutionalizing adversarial thinking. Dedicated teams are tasked with challenging prevailing assessments by examining how an intelligent opponent might manipulate signals, exploit analytic blind spots, or deliberately induce misinterpretation. Where alternative analysis focuses on internal disagreement, red teaming assumes the environment itself is hostile.

In SIGINT contexts, red teams ask how signals could be staged, spoofed, or shaped to produce misleading patterns. They explore how an adversary might design activity specifically to trigger analytic models or reinforce existing assumptions. The goal is not prediction, but resilience.

Crucially, red teams are structurally independent. Their value depends on being insulated from institutional pressure and career risk. They are not rewarded for consensus, but for surfacing vulnerabilities early. In AI-enabled organizations, red teaming helps prevent systems from becoming confidently wrong by treating apparent clarity as something to be tested, not trusted.

Only then can we avoid the traps of misplaced certainty and build systems that support judgment rather than override it.

In 2014, Boom Supersonic was founded with the goal of making supersonic travel mainstream. If successful, Boom will achieve something that no American startup has since the Douglas Aircraft Corporation was founded in 1921: enter the commercial airliner market from scratch and challenge an industry dominated by entrenched incumbents. Boom’s efforts have already yielded major regulatory breakthroughs. Since the 1970’s, the FAA has prohibited supersonic flight over land due to sonic booms. In 2025, Boom was instrumental in persuading the White House to sign an executive order directing civil aviation authorities to revisit and modernize these rules.

I sat down with Blake Scholl, the founder and CEO of Boom Supersonic, to discuss the disappearance of supersonic passenger flight, the consolidation of the aerospace industry, and his attempt to reintroduce founder-led innovation into commercial aerospace. Through Boom Supersonic and its proposed Overture airliner, Scholl is betting that faster transoceanic and overland travel can become commercially viable at scale for the first time in history. What follows is a transcript of our conversation.

CB: What did it take to build the US aerospace industry into what it is, with the primes, with the state of our technology, with the considerable power our aerospace industry allows us to project?

BS: I think this is a story of the role of entrepreneurship in innovation — what happens when it dies and what happens when it comes back. If you think about the first 50 years of flight, from the Wright Brothers through to the 1960s, this was entrepreneurially led innovation. The Wright Brothers were bicycle entrepreneurs who wanted to try their hand at making something new and important. And there was a golden age. Company creation started in 1903 with the Wright Brothers’ first flight, then went to 1921 when Douglas Aircraft was founded. By the way, it was another 95 years before the next commercial aircraft company was founded. And that was Boom. We had a half century of these founder-led companies doing incredible things. The last founder retired in the 1960s — and again, I’m talking commercial aviation here. If you look at private jets, it went on a little longer. But if you look at every major acceleration in aircraft, it came from a founder-led company. De Havilland actually introduced the first jetliner in the late 1950s while de Havilland himself was still running his company.

One trend was the retirement of the founders, which happened in the ‘60s. And another trend was this pivot in what was driving the vectors of innovation. I think we had entrepreneurially led — which really means customer- and use-case-led — innovation. What entrepreneurs do is figure out what people need, if only they knew how to invent it themselves, and build it, whether for a commercial customer or a defense customer. And so we went from this tiny little biplane to the most incredible intercontinental jetliners. And I think we peaked in some ways with the SR-71 Blackbird, also in the 1960s. By the end of the ‘60s, though, we moved from an entrepreneur-led, use-case-led model of innovation to a glory-and-national-prestige model of innovation.

I have a pretty controversial opinion here. In 1969 we landed on the moon, and also in 1969 we flew Concorde for the first time. And I think both of those were massive mistakes, and we would have had far more progress if we hadn’t made them. What happened is instead of having use-case, capitalistically driven innovation, we started doing things because they looked good on the world stage. We literally landed a man on the moon. Miraculous technical accomplishment, don’t get me wrong. Super inspiring. But we did it to beat the Russians there, and then we didn’t know what to do next. And it’s certainly not sustainable to spend 4% of the federal budget continuing to put more flags on the moon. We can’t afford it. We built an entire technology behind that that is entirely unaffordable and unsustainable. I don’t think you could have found someone in 1969 who was so pessimistic that — if you asked the most pessimistic person in 1969, “Tell me about air travel and space travel in 2026” — you’d find a single one who said, “You can’t go to the moon, you can’t fly supersonic.” It would have seemed absurd.

CB: It seems like it’s pretty much identical today, except maybe the food’s a bit worse, in terms of commercial air travel from the late ‘60s.

BS: I mean, we have made progress in optimizing the machine. Safety has gotten better, efficiency has gotten better, and affordability has gotten better. And those are real gains, but the basic capability hasn’t gotten any better.

Boeing’s latest airliner, the 787, is a highly optimized version of their first jetliner, the 707. It’s far more optimized, but it’s the same machine, the same concept, the same speed. And there’s been no radical rethink. It’s just optimization.

CB: I think one of the lessons of the Soviet Union was that centralization is kind of a bad thing. It’s not good for innovation, right? And it seems like since the Cold War ended, we’ve gone in that direction.

BS: Fundamentally, why does communism not work? It’s a lack of freedom, which means you turn off people’s brains, right? If you’re not free to do, you’re not free to think. And I think the human mind is the source of all progress. Then you’ve got, on top of it, central planning, which means that if you do have any brains involved, it’s a small number that are isolated from the real impact, and whose motivations are disconnected from the real impact.

But then what did we do in the ‘60s? We were competing with the Russians. And how did we compete with them? We started all these centrally planned aerospace projects. Apollo was a communist-style project. And I know this is sacrilege to say, but it’s really true. This was a government-spec project where the goal was to plant an American flag. It was not capitalistic in any sense of the term.

And Concorde, which was a joint venture between the French and British governments, was the exact same thing. It was like, “Hey, let’s show that Western technology beats Soviet technology.” The space race was bipolar. It was the Soviets versus America. The supersonic race was actually tripolar at first. There were the Soviets, the Europeans, and the Americans. And we did the same thing with supersonic in America that we did with space. There was an FAA-spec, taxpayer-funded supersonic transport. Concorde was Mach 2, twice the speed of sound, 100 seats. The American competitor was supposed to be Mach 3, 300 seats. And by the way, they had no idea how to make the affordability work such that 300 people could afford to fly on it. It made no economic sense. Concorde made no economic sense.

At the time Concorde was built, you’d find roughly a couple dozen first-class seats on an airliner. And then Concorde proposes to have 100 seats with fares a multiple of first class. It makes no sense. You just can’t find that many passengers. This is Econ 101. The more expensive the thing is, the fewer people will buy it. The more expensive the fare, the smaller the airplane needs to be. Again, no capitalist would make this mistake.

CB: What went wrong with the Soviet supersonic airliner?

BS: People called it Concordski. It was basically the same airplane concept, roughly the same speed, roughly the same size, maybe not as well built. It definitely didn’t make any sense economically. In the Soviet Union you couldn’t find 100 people willing to pay four times first class to go Mach 2. And then the airplane didn’t work super well either. There was a famous crash at an air show, and so it didn’t even last as long as Concorde did.

CB: It’s an interesting paradox. You can have a fantastic, even groundbreaking technology that’s just a terrible business, and you can also have a great business that’s built on some kind of primitive or very simple, non-innovative technology, right?

BS: I think Marc Andreessen coined, or at least popularized, the term “product-market fit.” And the most important thing is: What are you making? Who is it for? Why is it any good? And these things have to mesh. I think entrepreneurs basically exist for the purpose of creating new things that have product-market fit. But when the taxpayers write a check, there’s no such thing as product-market fit.

The important thing is: Is it valuable to an audience that matters? That’s what private innovation does. Because if you don’t make something that’s valuable to an audience that matters, you don’t get paid. You go away, and the capital allocators don’t give you any more capital.

This is the free-market system. On the other hand, there is no feedback loop like that with the government. Glory projects get rewarded for glory. And you’re absolutely right: those guys can do technologically impressive things. Concorde was technologically super impressive. Apollo was super impressive. And some things were probably invented faster than they otherwise would have been. But if you pull the camera back and ask, on net, was it good? I think the answer is, sadly but obviously, no. Apollo did not lead to more space exploration. It led to less. Concorde did not lead to more supersonic travel. It led to less.

CB: What does it take for a country to have a real aerospace industry?

BS: Aerospace is really hard, so you’ve got to have a robust talent pipeline. And that’s a flywheel that is hard to get spinning. Advanced jet engine technology is one of the few things that China has not been able to copy from the West yet. And we can make turbine blades in the US and Europe that radically outclass anything made in China. That is a statement you cannot make about very many kinds of products. That knowledge flywheel — which is really about what’s in the minds of the engineers and how they teach the next generation — is super, super important.

CB: What were the primary constraints in founding Boom technologically, from a business perspective, and from a regulatory oversight perspective?

BS: Here’s the shocking thing: there were no technological constraints. When I started working on Boom about 12 years ago, one of the most important early discoveries was that we already had the technology required to build an economically viable supersonic transport. There was a book that NASA had published called Commercial Supersonic: The Road Ahead. And it was kind of a systematic, technology-by-technology breakdown: where’s the state of the art, and where does it need to be?

They were looking at two different product ideas. One was a supersonic private jet for the ultra-wealthy, and the other was a 300-seat supersonic jumbo airliner. And they said, well, you can’t do the private jet because private jets fly mostly over land. And if you can’t fly supersonic over land because you haven’t solved sonic boom technically or regulatorily, that’s stuck. There’s no market. Decades of R&D would be required in order to solve sonic boom and make a supersonic private jet viable.

And then you can’t do the 300-seat supersonic jumbo because nobody knows how to make that efficient enough that the airfares would be affordable to enough people to get 300 at a time on board the airplane. So their conclusion was that decades of R&D were required.

I remember reading that and thinking, are those really the only two ideas? And it seemed fairly obvious that today roughly 50% of international air travel dollars are spent in business class, and that’s where about 80% of the operating profit comes from for international airlines. And these are people who are paying top dollar, sometimes even $20,000 for a round-trip ticket with a flat bed, because the flight is long and they want to sleep through it. And so the airlines charge a lot for flying beds.

By the way, if your flight is faster, you don’t need the bed. And you can improve passenger density. So that was the founding idea of Boom: let’s not assume any new technology. Let’s not assume any new regulations. Let’s use existing technology, and let’s build an airplane that is commercially viable at fares similar to what people already pay in subsonic business class.
And it turns out you don’t need to invent anything fundamentally to make that happen. All the technology is there, and it had been there for about 10 years before the company was founded. Theoretically, Boom could have been founded 10 years earlier.

CB: That kind of confirms the great man theory of history, right? Sometimes it just takes a single actor with some agency to make something significant happen.

BS: I think it’s true generally. And I think in Silicon Valley people tend to tell you that if your idea is any good, there are already several teams working on it, and that if nobody else is working on it, there’s probably something wrong with the idea. The “why now?” question is another version of the same thing. Why now? Why is this only just now possible? The assumption is that things get done at the earliest moment they’re possible. And I think that’s just not true.

CB: I think that assumes a lot of efficiency that probably doesn’t exist.

BS: It doesn’t exist. And the fact that people hold this theory actually reduces efficiency. It effectively creates a bystander effect.

CB: How does the commercial air travel business model work, and what are the primary inefficiencies in the current model?

BS: Commercial air travel is really a bundled three-class product. If you look at the anatomy of a Boeing or Airbus long-haul airplane, you’ve got business class, you’ve got some premium, and you’ve got some coach. And the whole thing works because it’s a bundle of three classes. People say you make money up front and recover costs in the back.

Boeing and Airbus are both in the business of selling these three-class airplanes. And today, we don’t have the technology yet — though I intend to create it at Boom — but we can’t do supersonic at economy fares. We can do it at business-class fares. Basically, we’re cutting off the business-class cabin, pulling it out of this big airplane, and building an airplane that’s all business.

It’s an interesting case of disruptive innovation, because if Boeing did this, they would blow up the economics of their wide-body products. They’d have to cannibalize their system product line. And ordinarily companies might say it’s better for me to cannibalize myself than to let somebody else cannibalize me. But in practice, only a few companies are willing to eat themselves for the future. Steve Jobs, for example, was willing to destroy the iPod business to create the iPhone business.

CB: I understand you have manufacturing facilities in Colorado and North Carolina?

BS: Yeah, we’re live in Colorado now, and we’re standing up North Carolina.

CB: What kind of machine tools do you need to build an aircraft factory? Where do you even get machine tools from these days?

BS: There’s an enormous variety of different production processes to build the different kinds of parts that go into airplanes. If you focus on the engine, which is primarily metallic, we’re talking about castings, investment castings, forgings, all of which basically get ground or machined down to net shape.

The equipment you need for that comes from all over the world, and often not the US. Japan and Europe are, in many cases, leaders in machine tools. We have a number of Makino and Mazak CNC machines and mills that are designed in Japan, sometimes built-made in Japan, sometimes built-made in the US.

CB: What were the lessons you learned in building XB-1, the supersonic demonstrator that you tested last year? What surprised you once you actually had the data from the test flight and from seeing it become a real, tangible aircraft?

BS: The overall plan was: step one, demonstrate that we have the technology for a commercial supersonic airplane by going and building and flying the first ever independently developed supersonic jet made out of airliner technology. We did that with XB-1. Mission accomplished.

Step two — and this was a surprise along the way — was to solve sonic boom and repeal the ban on supersonic flight over land in the US. That was a consequence of the XB-1 program. That’s done.

Step three was also a surprise. Take our engine technology that we’d originally planned for our airliner and deploy it first on the ground for power generation. So we are delivering our supersonic engine technology first to AI hyperscalers that are using it to power AI data centers. We’ve got over a billion dollars in backlog for our power turbine product.

The next step is taking all the money and the knowledge from that product and using it to finish development of the airliner. And then we all get to fly supersonic. Ultimately, that becomes a whole family of products, from large airliners to small private jets. And I think supersonic replaces subsonic, ultimately, for every passenger on every route.

What was surprising was that we solved sonic boom. We founded Boom with the idea of just focusing on international routes first, putting the booms over water. You don’t have to solve it over land. It turned out that it was much easier than we thought. We were able to solve it really in software. We demoed that, and then we found ourselves — with the combination of the demo and a kind of deregulation-oriented political environment — taking only 115 days from the demo to the regulatory change, which has got to be some kind of speed record in and of itself.

CB: How soon will Overture, your flagship supersonic airliner, be a reality?

BS: On the order of 5 years, maybe a little bit less. We’re doing this as fast as we possibly can. But it’s one of the most complex, safety-critical machines ever made, and we’re doing it as a brand-new company.

CB: The world has changed a lot every time travel times have halved. It used to take two months to cross the Atlantic, then a week, and so on. What do you think the implications will be when travel times are halved by your product?

BS: The really cool thing is that we’re not smart enough to predict or even imagine that. The best clues are if you look back and ask, what were the second-order consequences of the jet age? Because this has happened before. We’ve doubled the speed of air transit before.

What happened? It turns out it opened up a bunch of destinations that otherwise wouldn’t have been destinations. It made Hawaii a destination for the first time. It made it possible for families to vacation in Europe from America. There was a whole bunch of cultural cross-pollination that otherwise wouldn’t have been possible.

Companies like Nike got their start. What’s the connection between Nike and jets? I stumbled across this while reading Phil Knight’s memoir. In the late 1960s, after business school, he fell in love with Japanese-style running shoes. And Nike got its start importing Japanese running shoes to America. There was no such thing as a chance trip to Japan after business school without jet travel.

There was no Major League Baseball west of the Mississippi before jets. For 100 years, from the founding of Major League Baseball, it was this cluster of teams in the Northeast and it never changed. Then suddenly it changed overnight, and we got teams coast to coast. If you enjoy any kind of major league sports west of the Mississippi, it’s because we’ve got jets that can move teams around.

So you don’t have to get on an airplane to benefit from a faster airplane. There was a story in The Wall Street Journal a few weeks ago about how the NFL is closely tracking supersonic because they’re asking whether this makes it viable to put a team in Europe.

CB: How difficult is it to scale speed once you’ve broken the sound barrier? Is it significantly different to build a Mach 1.2 commercial aircraft versus one at Mach 2 or Mach 2.5?

BS: There are step changes in difficulty along the way. What we’ve tried to do is be on the favorable side of a significant speed increase, and initially on the favorable side of step changes in technical difficulty. But it will absolutely ratchet up over time.

Overture One we’re designing for Mach 1.7, because that’s roughly a doubling in speed versus today’s airliners. And yet it’s within a turbofan architecture that can still fly efficiently at that speed and meet takeoff noise compliance, which turns out to be a big design driver.

But we have a technical roadmap to push to far higher speeds than that. And I think ultimately the limit is actually human comfort.

CB: What do you think are the prize routes that are going to be fundamentally transformed by this?

BS: Our modeling says there are going to be at least 1,000 that are significant. Everyone remembers New York to London from the Concorde days, because that was really the one route that almost worked as it was. It was rock stars and royalty who were willing to pay top dollar. And you could leave New York in the morning and make a dinner meeting in London, and if you wanted, come back that same day.

You can generalize that to the eastern US traveling to Western Europe. Effectively, you’re taking what would be a red-eye flight and turning it into a daytime flight. And then coming back to the US, you actually land before you take off in local time. With maybe a five-hour time difference and a three-and-a-half-hour flight, you land 90 minutes before you took off. It’s going to be really awesome.

Some of the really exciting routes are the ones that never worked with Concorde, like the US to Australia. That’s such a terribly long flight today — 16 hours. Even if you manage to sleep on the airplane, when you wake up there’s still a whole other day of misery on the airplane ahead of you. That turns into about an eight-hour flight with Overture, say LA to Sydney. That’s actually a reasonable-length red-eye. Sleep on the airplane, wake up, and you’re there.

The time zones also work out beautifully in some cases. You can go to sleep in Sydney and wake up in LA eight hours later.

There are routes which don’t get much traffic today but suddenly start to sing with supersonic. What about vacationing in Tahiti? What about the Maldives? These are places where the flight time is a real barrier. When you cut flight times in half, all of a sudden it makes more sense.

And then with boomless cruise, there’s also domestic travel. We can shave about 90 minutes off a coast-to-coast flight. That means you could leave New York at 9 a.m. and land in San Francisco at 9:30 a.m. local time. I think there will be scenarios where commuting becomes viable in ways it isn’t today.

Multiple projects were proposed to mark this new era — modular stations, joint missions. But costs mounted, and the Soviet Union collapsed. Eventually, Russia, which inherited much of the USSR’s spacefaring infrastructure, and America combined their efforts. The result was the International Space Station (ISS) in September 1993, the most expensive structure ever built at over $150 billion. It was assembled piece by piece in orbit over more than a decade, beginning in 1998 and reaching major completion in 2011. It was a triumph of post-Cold War globalization: America and Russia provided the bulk of the components; European, Canadian and Japanese modules gradually joined to form a single structure. In total, it took roughly thirteen years of continuous assembly and international coordination for the ISS to become what it is today: a functioning and continuous human outpost in LEO. The only other operational space station in orbit is China’s Tiangong, launched in 2021.

The ISS is now twenty-five years old, suffers from material fatigue and costs taxpayers roughly $3 billion annually. NASA plans on decommissioning it by 2030. The United States has not built another space station since.

✺

At a 90,000-square-foot facility in Long Beach, California, Vast is vying to become the ISS successor with its proposed space station, Haven-2.

The guest lobby at Vast feels like the inside of a very luxurious space shuttle — rounded, tall, grey, futuristic, with a central tree growing up into the sky. Vast repurposed former distribution warehouses for aerospace use, with three buildings spanning manufacturing, testing, integration, and astronaut training, Standard operating procedures are strictly followed on the factory floors. Earplugs and safety goggles are neatly arranged in cabinets, with clean room suits on racks available at multiple checkpoints. There are pillars here and there holding protective headgear. Each building has its own communal cafeteria, and lunch is being served during our visit. People are very excited about the burrito station.

Vast was founded by entrepreneur Jed McCaleb in 2021. When I asked Vast CEO Max Haot about the company’s name he explained: “Jed named it Vast because space is vast.”

Haot was born in Belgium. He taught himself to code as a child and moved to the UK after graduating high school. In London, while working at the sports management company IMG, Haot pursued his interests in live television and the internet by developing websites, scoring and real-time video platforms for organizations like Manchester United and Wimbledon, learning English along the way. He moved to the United States after Verizon acquired a content-management system he developed. “I wanted to follow the American dream,” he told me. Haot landed in New York City in 2005, founded the livestreaming platform Livestream in 2007 and sold it to Vimeo 10 years later — his second liquidity event.

Following the acquisition, Haot finally had the resources to pursue big-picture ideas for humanity. “We are expansionists — we need a frontier. In 10,000 years, if we are still around, we’ll be multiplanetary. And if that’s true, then the most important event ever would have been Sputnik, and then the moon landing. The first time we left. And so I’m born at minute zero.” Haot founded Launcher, a rocket and satellite launch company, in 2017, the same year as Livestream’s sale. He then scaled it from a small New York team to a major operation in Hawthorne, California. After meeting McCaleb, he sold to Vast in 2023 and joined as President, becoming CEO a few months later.

Currently, the Vast facility is buzzing: the team is putting together Haven-1, Vast’s flagship space station. The structure is in its integration phase, following successful pressure and load testing in 2025. Vast will have invested about $1 billion, a combination of capital provided by founder McCaleb, and revenue from customers, by the time the first crew to Haven-1 is launched. “That’s actually not a lot of capital compared to what’s been done on space stations so far,” Haot explains, referring to the cost of the publicly funded ISS. Vast operates on what Haot calls a “leapfrog strategy” — step-by-step development funded privately. “Build an actual space station, and launch it as quickly as we can, and become the company building the world’s first commercial space station.”

Haven-1 is Vast’s proof-of-concept to win NASA’s CLD Phase 2 funding. (Awards are expected to be announced mid-2026). CLD — Commercial Low Earth Orbit Destinations — is NASA’s program to partner with private companies to build and operate the next generation of commercial space stations, intended to replace the ISS after its planned retirement in 2030. Phase 1, which has been ongoing since 2021, provided smaller funded or unfunded agreements for early design and technology development — Axiom and a partnership between Blue Origin and Sierra Space received funded awards in the hundreds of millions. Phase 2 will award $1 to $1.5 billion to at least two winners. Unlike earlier fixed-price contracts, the program now uses Space Act Agreements, with NASA providing direct funding for hardware demos, crewed tests, and prep work through 2031. We caught Vast in the run-up.

Haven-1, designed with guidance from former Apple Industrial designer and Vast advisor Peter Russell-Clarke, is a compact cylindrical module about the size of a subway car, with “Earth tone” interiors, a domed observation window, personal crew quarters, and always-on Wi-Fi via Starlink. The station is able to support up to four crew members for missions of 10–30 days (up to 40 days maximum). It also includes the Haven-1 Lab: 10 standardized payload slots (each up to 30 kilograms and 100 W) for microgravity experiments. Food scientist Zach Rosenthal is leading a team to develop food systems to meet the nutrient demands for long term missions. The station is planned to launch uncrewed on a SpaceX Falcon 9 rocket from Cape Canaveral, Florida in Q1 2027.

“We went from 120 people when I joined” — 40 at Vast, 80 at Launcher — “to a thousand people. We went from empty buildings, to fully built facilities, where we believe we have everything we need to build America’s next space station,” said Haot of Vast’s rapid growth. I ask if McCaleb’s personal investment has given Vast a competitive advantage, allowing them to take risks unavailable to VC-funded competitors like Axiom Space. “The payback for human space flight is not instant, and the amount of technology you have to build, and the amount you have to invest, you just can’t bootstrap it. The barrier to entry is so high, the moat is so high, that it needs this kind of private investor willing to place such a large bet.”

Foreign governments have already signed on: Colombia, Uzbekistan, and the European Space Agency, among others, have signed memorandums of understanding with this private American space company to explore human spaceflight opportunities and research on Haven-1 and future stations. The change, from trusting the US government to trusting a private operator, is significant. “We’re lucky that there’s an amazing precedent,” Haot reassures me. NASA’s move from ownership to a more supervisory role, via its “cost-plus” programs, in which contractors are reimbursed for expenses plus a guaranteed profit margin — removing incentives to control costs — to fixed-price commercial contracts was validated by SpaceX, whose reusable Crew Dragon spacecraft now flies astronauts with an “impeccable safety record” under close NASA oversight, a model that is now being extended to private space stations.

In order to validate key systems that will power Haven-1, Vast launched Haven Demo into orbit in November 2025, ridesharing on SpaceX’s Bandwagon-4. Haven Demo is boxy, roughly 3 × 4 × 3 ft. It successfully demonstrated core systems like propulsion, navigation, and control, directly informing the final design of Haven-1. It performed well even during the northern lights phenomenon in December 2025, which produced intense auroras that can disrupt spacecraft transmissions. Demo currently transmits onboard footage in 4K, viewable from the mission control center room at Vast HQ. Its success marked Vast as the only private company to operate hardware in space created specifically for commercial space station development.

By sending up low-cost pathfinders equipped with telemetry systems, Vast avoids the high-stakes first-flight failures common in new space hardware, while gathering empirical data in actual orbit. This mirrors the SpaceX-style progression COO Kris Young emphasizes — “uncrewed engineering tests before crewed demos” — and mirrors Vast’s broader philosophy of iterative development. Young joined Vast in September 2025 and is an aerospace engineer with roughly 18 years in spaceflight, including roles at Northrop Grumman and more than 14 years at SpaceX focused on human spaceflight. At SpaceX, he led engineering for Crew Dragon and later oversaw flight and launch operations, supporting around 19 human spaceflight missions and 150 rocket launches in a single year, which was then “a record for the company.”

Young frames Vast’s North Star: “humans are endeavored to be among the stars,” and thus Vast’s goal is to make space “safe and efficient for humans,” from LEO to the Moon and Mars. He emphasizes that this begins with foundational demonstrations like Haven Demo, advancing through Haven-1 as a learning platform for human spaceflight, and culminating in Haven-2 as a successor to the ISS, enabling “continuous human presence” and a 21st-century environment for living, working, and conducting research in space.

To this end, Vast deliberately explores long-horizon ideas — like artificial gravity stations and fully closed-loop life-support systems — because microgravity profoundly affects the human body: “you actually end up losing about half your blood volume,” Young tells me, along with muscle, bone density, and balance. This philosophy guides their Environmental Control and Life Support Systems (ECLSS), which recycle air, water, and other resources to reduce resupply needs. Haven-1 starts with a simple, open-loop ECLSS that prioritizes reliability, venting waste and relying on Dragon to bring crew aboard. In parallel, Vast flies closed-loop experiments on every mission to gather real data and iterate without risking crew safety. The plan is to mature this system by Haven-2 so that these life support systems have no dependence on Earth.

Haven-2 will be fully equipped for long-duration, continuous human occupation with up to 12 crew members and approximately 500 cubic meters of habitable volume. (For comparison, the ISS has a known maximum capacity of 13 people, with seven being the standard crew count). Vast’s longer-term roadmap envisions building artificial gravity stations generated by rotation: the station will spin at roughly 3.5 revolutions per minute to produce Earth-like gravity along its outer rim, so that astronauts can live and work under conditions similar to those on Earth. Vast has had an unfunded Space Act Agreement with NASA since June 2023 for collaboration on microgravity and artificial gravity technologies, giving Vast access to NASA expertise, facilities (e.g., testing at NASA’s Neil Armstrong Flight Research Center), and validation, helping de-risk their experimental concepts without cash funding.

The Haven-2 station will be assembled in LEO through sequential launches beginning with the first core module targeted for 2028 — contingent on winning NASA’s CLD contract. Vast aims for Haven-2 to provide seamless continuity for NASA-sponsored science and international collaboration in a post-ISS era.

I ask Young what prompted his decision to join Vast. “I spent a long time working on crew transportation and just getting humans out of Earth’s atmosphere into orbit. And, whilst it’s never easy, it’s working. We have a safe and reliable transportation system.” He alludes to how SpaceX has made transportation in and out of Earth’s orbit almost commonplace. “Right now, the number one limiter for flying [to space] more often is not having a destination. What inspired me to join Vast is that they are gonna be the first destination.”

“I have a lot of pride in just being able to build capability for the country. I’m really proud of some of the technology we’ve developed,” Young tells me. The one story he wants to share with me is about the team developing in-house control moment gyroscopes (CMGs), devices that allow a spacecraft to rotate and orient itself without using fuel. For long-duration human presence in orbit, the vehicle must maintain attitude — attitude meaning its rotational orientation in space — without using propellant. CMGs solve this by using electricity to generate torque, providing fuel-free attitude control that’s reliable over years. Vast discovered the sole US supplier for CMGs was slow and expensive, Young explains: “We’d love to buy some. Well, first it took them a while to actually answer the phone… then it was like, okay, pay us 50 million bucks and we’ll see you in four years… What? Are you kidding me? We’re gonna fly Haven-1 before that’s even done.” So Vast designed, built, and tested the first set of CMGs themselves in roughly six months, and they’ve been spinning continuously for two years with additional sets built since then.

Speaking on how Vast will be different from the ISS, Haot explains, “we need to be very respectful of what’s been achieved in 25 years, the capability, the safety. No one wants a LEO settlement to just disappear. Our biggest contribution I hope will be that we make it sustainable, profitable, save taxpayer money. We want to achieve more science.”

It feels very special, yet equally humbling, to stand next to, and see, Haven-1 being assembled. The structure demystifies space technology for me: I see pieces of metal being put together by human hands, similar to mine. The pieces that have come together to complete the station and its components — mainly aluminum and some steel — have been ordinary throughout my whole life as well. What feels most extraordinary is the timescale. Nearly 300 people from 26 nations have visited the ISS; many of the people who will one day live inside Haven-1 and Haven-2 haven’t even been born yet. I wonder when our ambitions and dreams will outgrow Haven-2, or even LEO itself. When I ask Kris Young if the sheer scale of what they’re attempting is daunting, he calmly replies, “No. It’s exciting.”

Alright, I’m kidding. For as long as technology has been changing work, education has been important to help individuals and societies adapt. It’s still true. Learning about code and computers is very worthwhile, and it’s tremendous fun. AI and coding may present a unique conundrum because in the history of education, the subject you’re learning has not been able to teach itself. “Chemistry” can’t teach chemistry. But “AI” can most certainly teach AI. AI can teach you to code, because it knows how to code, which is begging the question... why bother at all?

It’s a question that few people have thought more about than Lee Robinson — “leerob” for short on GitHub and X. Robinson grew to be a well-known figure among developers as one of the main public faces of Vercel, the company that develops the open-source Next.js framework for web applications; today, he’s VP of Developer Education at Cursor, the $30-billion company whose eponymous coding tool has enjoyed one of the most impressive takeoffs of any AI product since the launch of ChatGPT.

I met Robinson in San Francisco on one of his regular trips from Iowa, where he lives (and where I grew up, too). He flies in every few weeks to spend time at Cursor’s offices, of which there are now several in San Francisco’s North Beach neighborhood. He spends a few days, then flies home. It’s a rhythm he settled into during his five years working at Vercel, and one that tells you a lot about how he thinks. “When I’m here in San Francisco, there’s this amazing buzz in the office and energy with things happening. I love it for a week, and then by the end of the week, I feel like I can get work done better when I’m back in my cave in Iowa.”

“I used to have an office upstairs,” he told me as we walked around North Beach. “Then we had a daughter and that was not gonna fly. So I moved my office into the basement. It has a closed door, and I invested probably more than any reasonable person would in having an immaculate camera setup. I plug in a couple things, turn on a couple lights, and I’ve got studio quality video in my home.”

✺

Robinson grew up in small-town Iowa, went to Iowa State University for computer engineering, and stayed for his first jobs as a software engineer—including at Workiva, a publicly traded software company based in Ames (where Iowa State University is), and Hy-Vee, an Iowa-based grocery that is the dominant grocer in Iowa and the leading brand in Nebraska, South Dakota, and Minnesota, too. One of his last projects as an engineer at Hy-Vee was standing up a grocery delivery platform in 2020.

Robinson had nearly left Iowa in 2018. He had a job offer to be a product manager at a startup in San Francisco. In the end he didn’t take it. At the same time, he began to appreciate teaching people about code in his free time — helping people “to understand why we’re doing all this stuff in the first place. It’s not just about writing a bunch of funny characters into a computer, but it’s about building something great.”

Yet he noticed a particular content gap for the type of web development work he was doing with React (a Javascript framework maintained by Meta) and Next.js.

So, he made the content himself, first short video tutorials and blog posts, and eventually paid courses. Because Next.js was still young and almost nobody else was producing material about it, he became a de facto ambassador. “It wasn’t really intentional,” he told me, “but I kind of became the sole result when you Googled Next.js. Basically the only results back in the day were my videos.” The first course that got traction had a hundred paying customers, and the feeling stuck with him. “At the time, that was insane to me.”

The team at Vercel — just a few dozen people at that time — noticed and asked if he’d ever considered doing it full-time. “I mean, actually, no, I haven’t ever thought about doing it full time,” he recalled thinking. “I’ve just been kind of doing this on the side. What does that even look like?” The answer, it turned out, was a lot of different things. And over five years, he helped define a role that is now in high-demand at the most important technology companies engaged in a brutal contest for the love and attention of developers and users. And he managed to do it while staying put in Des Moines.

Vercel is a fairly unique model in tech: a venture-backed company built atop an open-source framework that anyone can use for free, including some of the largest businesses in the world. The tension is obvious, because the business can only thrive when people pay for what are essentially optional services (Vercel has hundreds of millions in revenue from its hosting services).

“It’s funny when you watch people try to copy Vercel’s business model,” Robinson told me. “It’s kind of an insane thing to do because it’s incredibly hard to do well, to have both a successful open source framework and community and then build a successful venture scale business on top of it—and make them not conflict with each other.”

One gets the sense that maybe what other other businesses lacked but which Vercel had was… someone like Lee Robinson, a figure whose personality and presentation exude trustworthiness. “I was not doing it just for the love of the game. I wanted the company to succeed,” he told me. And the success of Vercel, to Robinson, was proof enough that you could succeed on both dimensions.

✺

The release of Anthropic’s Claude Sonnet 4 model in May 2025 represented a turning point in Robinson’s thinking about what DevRel was going to mean, and what role he was going to play. “Now is the time to change,” he recalled thinking. “Otherwise, I don’t know what developer relations or my job will look like in a couple of years. I can keep talking to them about Next.js and React. I’ve been doing that forever. But the AI models are getting very good at that. So what do they actually need help with?” He answered his own question: “They’re going to need to figure out how to navigate this strange, weird AI journey in a way that’s not too hypey but not pessimistic either.”

He joined Cursor in July 2025, when the company had around 80 employees. Its workforce has since grown to several hundred.

“There’s a lot of marketing around AI that developers are allergic to,” Robinson said. “They see the press releases and the launch videos and the product demos, and they just don’t trust it.” He said he wanted to do things a little differently.

“My goal in joining Cursor was to try to give people a pragmatic, cautiously optimistic view of this new world and to bring along these millions of developers who hear these news headlines. Developers are worried about their jobs; they’re worried about the future of this industry; I want to show them there is a path to prosperity where we will be able to do incredible things.”

In the same way that “Uber for X” once described an entire generation of startups — not always positively — “Cursor for X” has become a stand-in for the type of AI tools that makes a complex domain feel suddenly accessible. Cursor for law. Cursor for accounting. Cursor for architecture. Cursor for marketing. The name has become synonymous with the idea that a professional tool can be powerful and approachable at the same time. For someone whose career for the last few years has been about making complicated technology feel accessible, Cursor is a great perch for Robinson, and it keeps him connected to both millions of developers and some of AI’s top luminaries.

“I’ve had CEOs reach out who haven’t been coding for ten years who say, ’thank you for explaining this in a way that was accessible and digestible for me.’ And that’s my job: to cut through the noise and try to explain things in a way that a beginner would understand but that an expert would also find helpful.”

Robinson pointed out to me that in 2025, the top model on SWE-Bench — a benchmarking test that grades AI systems on their ability to solve real software engineering problems — changed more than a dozen times. “I think it’s great,” Robinson said, “because with so much competition, the best product truly wins.” And building the best products for coding is Cursor’s focus (the company develops its own language models, but all major models are available in Cursor’s products).

Robinson traces three major shifts in AI-assisted coding. The first was autocomplete: AI finishes your line of code, not unlike lane assist in a car. GitHub Copilot pioneered this, and Cursor followed with its own model, Cursor Tab. Developers were still writing code mostly by hand, but the AI could complete a line. The second shift was putting a chatbot inside the editor, replacing the need to search Google or Stack Overflow. The third shift was autonomous coding agents. “When you are working with an agent that is autonomously taking actions for you, you’re trying something that’s closer to self-driving, where the car is mostly driving itself. You’re still giving feedback and you’re nudging the wheel, but primarily you’re charting the course for where the car goes.”

Early agents could modify code inside a single file for tightly defined tasks like renaming a function, or moving a block of code around. As the models improved, they could work across multiple files, run terminal commands, and handle the kind of arcane operations that developers used to have to memorize. “Running shell commands, using a terminal — these are things that developers had to do by hand and remember all of these magic combinations of characters,” Robinson said. “But now, AI models can remember all of that stuff for you.”

Continuing the metaphor a bit, Robinson compares the new generation of coding agents to a fleet of cars unleashed on an entire city and allowed to drive freely: “We’re starting to reach the point of full self-driving for agents, where they can run for hours or weeks across entire code bases with millions of lines of code.”

When I asked whether looking at code would remain the primary way software engineers build software five years from now, Robinson didn’t hesitate. “Probably not. You would probably be looking at something that is much more like English. There will still be code. You will still have to think about the code. You might have to debug the code and dig in. But the primary interface will look more like English if the current trajectory continues to play out.”

And the ways in which people can interact with code using Cursor are growing in number, too. “If you want to make changes to your website from your phone, in English, Cursor has an app for that. You can tag Cursor in Slack and instruct agents to go fix bugs that users notice.”

Robinson himself welcomes this new paradigm, even though he isn’t exactly sure where it’s headed. “As an engineer, I loved code because of what it helped me build. I was never in love with looking at the code. But there are people who love coding for the artisanal, handcrafted code, and I think they’re having a bit more of an existential crisis right now.”

✺

Talking to Robinson, one notices that like most great teachers, he has a habit of explaining his opinions by recounting the questions he asks himself. He can marshal history and metaphors in his explanations. And it’s very clear that part of why he enjoys teaching so much is that he enjoys learning, including — or especially — when it’s a slog. “AI models have this funny way of tricking you into thinking that you learned something when you didn’t really actually learn something,” he said. “It’s like this short term gratification of seeing that it works, without the long struggle in the middle to make it work.”

Robinson doesn’t think there’s a shortcut around that struggle, and he doesn’t think there should be. “Learning anything difficult is a struggle. Like, it’s gonna suck. But part of accomplishing and learning really hard things is the grind not understanding, then trying and figuring it out. If you rely on AI too much and you outsource the thinking, you don’t really have a great foundational knowledge for how it works.”

When he said that, it reminded me of one of the first things he’d said as we began our walk from Cursor’s original office down toward the San Francisco waterfront — that he fell in love with teaching code in part to remind people that engineering isn’t an end itself, but a tool to build “something great.” And that, in a single idea, is the school of Lee. So, build to your heart’s content. Make Cursor for Architecture, or Cursor for Boats. Airbnb for Dogs, or Uber for Cats are options — the age of AI has made it exceptionally easy to build these new things — but the ease of it all presents some really hard choices. Life wants you to figure out what you actually care about, what you think “great” means. So, the one thing you might want to hold off on is “Cursor for Thinking,” because that’s worth doing the hard way.

At the very end of our walk, Robinson gave me this thought: “I always thought that at some point I’d have to move here to San Francisco, that it’d be the only way I could achieve the success that I wanted. And surprisingly, that hasn’t been the case.”

When Robinson started working at Vercel in 2020, San Francisco was in something of a nadir. Companies were openly contemplating leaving. A number of prominent individuals did. Staying in Iowa at that moment was easy to explain — why would you move toward the chaos? But in 2026 as a major figure at one of AI’s top companies, Robinson’s basement cave in Iowa is a far more significant choice. San Francisco is resurgent, the most important nexus of talent and capital in the history of the technology industry.

But for him, it is a considered choice, and I can say with confidence he didn’t outsource his thinking to anybody else. “I’ve spent more time than I’d like to admit in the Denver airport on layovers,” he lamented to me in a way that I understand very well as a fellow Iowan, having taken direct flights to my hometown only a handful of times ever (and certainly not from San Francisco).

“Iowa is not a travel destination, it’s not a luxury resort. But it’s a great place for me to raise my family, to be close to my parents and to my wife’s parents. It’s a slower pace of life and I do appreciate that a lot. But I also love San Francisco. I love coming here and being involved in the thick of building the future and the incredible talent density. I feel very fortunate that I can do both.”

China’s commitment to this ambitious project reflects a strategic vulnerability. The country imports more than $300 billion annually in semiconductors and semiconductor manufacturing equipment. This figure exceeds its spending on imported oil and gas. An American export ban on foreign-made microchips would cripple China’s industrial base. And if China can’t develop its own EUV machines, it will lack the advanced chips needed for AI model development if the U.S. decides to impose strong export controls.

Many analysts believed it would take far longer for China to assemble even a prototype. The task requires recreating an extraordinarily complex supply chain. ASML draws on more than 5,000 specialized suppliers to build close to 100,000 components for a single machine, which is roughly the size of a school bus. Many suppliers occupy narrow niches with no real substitutes.

EUV lithography depends on a light source created by firing a high-power carbon-dioxide laser built by TRUMPF at microscopic droplets of molten tin inside a plasma generator developed by Cymer, an American firm acquired by ASML. Each droplet vaporizes into plasma heated to temperatures hotter than the surface of the sun, which emits light at a wavelength of 13.5 nanometers. That light must then travel through an ultra-high-vacuum system supplied by Pfeiffer Vacuum. Even trace amounts of air would absorb the beam entirely. The light then reflects off a handful of mirrors polished and multilayer-coated to sub-atomic precision by Carl Zeiss SMT.

At the tolerances EUV lithography requires, the machine proves so sensitive that minute variations in gravity across its length, effects traceable to the curvature of the Earth itself, introduce measurable error. Every step in the process introduces noise and randomness that ASML’s control systems must model these variations in advance, correct them in real time, and mathematically cancel them before they corrupt the pattern. The machines are a statistical miracle.

Industrial espionage has been central to China’s EUV project. U.S. courts found that former ASML engineers at startup XTAL stole more than 2 million lines of the control software required to work these statistical miracles before its leadership fled to China. According to a 2023 report by the Center for Strategic and International Studies, Chinese intelligence services employ up to 100,000 open-source intelligence analysts who harvest publicly available information about technology of interest to China. These analysts map EUV supply chains and help identify recruitable technical talent. China lures these specialists to China as part of recruitment programs with promises of signing bonuses of up to 5 million yuan (roughly $700,000) and subsidies to purchase homes. Once in China, they work in secret facilities under aliases to conceal their involvement in the EUV initiative.

But recruited talent isn’t just copying. Lin Nan, a leading optical scientist, left ASML’s light-source technology group for the Shanghai Institute of Optics and abandoned the company’s approach entirely. His team developed solid-state lasers achieving 3.42% conversion efficiency, meaning that more than three percent of the input laser energy is converted into usable extreme-ultraviolet light. That performance reaches half of ASML’s commercial standard but exceeds what major Western labs have achieved. In other experiments, his team reportedly hit 50% efficiency. Solid-state systems are smaller, cheaper, and on a trajectory to surpass the CO2 lasers ASML relies on.

Industrial espionage has always accompanied shifts in technological power. The dominant power guards its critical technology. The challenger steals it, then tries to build something better. But as China has discovered with its EUV project, what matters most is not access to stolen designs or even intact subsystems. What matters is whether substitute or copied subsystems can be made to function together reliably and at scale. Western firms spent decades building that capacity, beginning with the assembly of the first EUV machine at Sandia National Laboratories in Livermore, California in 2001.

The Chinese EUV prototype, even if successful, addresses only one bottleneck in the supply chain. China would still face challenges in photoresist chemicals, where Japanese firms dominate. Ultra-pure fluoropolymers come from German and American suppliers. Dozens of other specialized inputs remain out of reach, with Zeiss optical systems being the most difficult to replicate. Building a prototype differs categorically from manufacturing at scale with acceptable yield rates. ASML took fifteen years to move from its first EUV prototype to reliable production. China’s timeline depends not just on technical capability but on whether it can compress this scaling process.

China’s EUV effort is not historically unique. What it exposes is a recurring feature of technological competition between great powers where industrial espionage helps accelerate development, but is seldom decisive. A survey of the history of industrial espionage is instructive as to why that is the case.

The Historical Pattern of Technology Transfer and Power

The world’s first great industrial power began with a great theft. Spain controlled the Atlantic in the 16th century through navigation secrets. Spanish pilots combined magnetic compasses, astrolabes, and celestial observations to cross the open ocean. The Spanish crown guarded this knowledge as a state secret while English mariners lacked this expertise and stuck to coastal waters.

That changed when an English spy obtained Martín Cortés de Albacar’s Art of Navigation. The book appeared in London in 1561 and England immediately invested in applying the stolen knowledge. John Dee taught mathematical navigation to sea captains. The English crown funded exploration voyages such as Frobisher’s Northwest Passage attempts in the 1570s and Drake’s circumnavigation starting in 1577. Thames shipwrights spent two decades developing “race-built” galleons with superior maneuverability.

In 1588, an outnumbered English fleet faced a 130-ship Spanish Armada escorting an invasion force intended to overthrow the English regime. English captains used superior maneuverability to hold windward positions through the Channel battle, firing from ranges beyond which Spanish guns couldn’t respond. About half the Armada never returned to Spain while England lost almost no ships. Stealing Cortés’s manual took one spy. Building a navy that could beat Spain took 30 years.

After England’s civil war ended in 1651, the country lagged its Dutch rivals in shipbuilding and finance, German states in metallurgy, and French competitors in luxury manufacturing. Over the next 150 years, Britain smuggled in machines and experts with the process knowledge needed to help them catch up, such as French silk weavers and German metallurgists, despite export bans imposed by rival nations.

But stolen knowledge alone didn’t create British dominance. Industrial growth fed the Royal Navy, which became the world’s largest industrial consumer. This created a cycle. Naval demand drove industrial capacity while stolen techniques became British innovations. Growing production led to ever greater state and maritime power, which enabled Britain to expand outward and control 25% of the world’s landmass at the peak of its empire.

It was the arrival of the United States on the world stage, with its heretofore unimaginable advantages, that would break the British model. The newly independent United States ignored Britain’s export bands and patent laws entirely. One of America’s first great industrialists, Samuel Slater, memorized designs of British spinning frames, emigrated illegally, and rebuilt them from memory. American agents recruited British workers, reverse-engineered machinery, and systematically acquired foreign technology.

Carroll Quigley argued in The Evolution of Civilizations that nations rise when their institutions expand productive capacity. He defined this as the ability to turn labor, resources, and knowledge into usable goods and services. Britain developed powerful instruments of expansion, but the United States inherited that instrument without Europe’s constraints. Vast natural resources, navigable rivers, a large mobile low-cost labor force, and a continent-sized internal market removed material bottlenecks.

American manufacturers pioneered the “American System of Manufacture,” which relied on standardization, interchangeable parts, and precision machining to compensate for a lack of craft expertise. By 1916, the U.S. had 30-50% higher output per worker than Western Europe and its economy exceeded the entire British Empire in size. By the 1930s, U.S. labor productivity was 50-90% higher than Western Europe.

The WWII Acceleration

Its higher productivity rates enabled the U.S. to become the largest airplane manufacturer in the world during World War II, producing more planes in 1944 than Japan did during the entire war. But it still lagged behind German and British aerospace technology until the war enabled the Americans to access their secrets.

Through the Tizard Mission in 1940, Britain transferred its most advanced technology to the U.S. because it lacked the industrial strength to develop this technology at scale and it wanted to entangle U.S. industry in its war effort. This technology included the first practical realization of jet propulsion, the Whittle W.2B engine. General Electric (GE) copied this engine design, enabling Lockheed to build America’s first jet fighter plane in only 180 days in the closing days of World War II.

As the war reached its endgame, the Americans and British would join together in their pursuit of Nazi “wonder weapons,” such as the V-2 missile and ME-262 jet fighter. Over 3,000 British and American technical experts accompanied Allied forces in their push towards Berlin in 1944. On the British side, Ian Fleming, who would go on to author the James Bond novels, led top-secret missions as commander of the 30 Assault Unit. The U.S. Air Technical Intelligence unit combed through rocket factories carved into mountains and secret Luftwaffe wind tunnels where experimental plane designs had broken the sound barrier.

After Berlin fell in spring of 1945, the Air Technical Intelligence unit sent recruited elite Luftwaffe aerospace engineers and test pilots to a luxury German mountain spa where they stayed until they could join Operation Lusty (Luftwaffe Secret Technology) at an Army Air Force base in Ohio. There, the Germans helped accelerate U.S. aerospace R&D by four or more years through wind tunnel design, jet engine reverse-engineering, and swept-wing development.

GE and Pratt & Whitney engineers synthesized British centrifugal jet engines with German axial-flow compressor research, creating jet designs that went from 2,000 pounds of thrust with tens-of-hours lifespans during the war to over 10,000 pounds of thrust with 1,000+ hour service lives, enabling a generation of supersonic fighters and long-range bombers. Other GE engineers were in the deserts of west Texas as part of Operation Fireball, where they made major improvements to the V-2 missile’s guidance and control systems, subsystem performance, and system integration. Engineers from Rocketdyne improved the V-2 engine design, ultimately building the F-1 engine which remains the most powerful single-chamber liquid rocket engine ever flown, reliable at 1.5 million lbf (pound-force) while the German engine could only generate 25 tons of thrust and routinely destroyed itself via pressure oscillation. The V-2 rocket failed more than 30% of the time it was launched during the war; American engineers brought the frequency down to under 5%.

By the mid-1960s, approximately 5% of the U.S. GDP funded new weapon systems and NASA’s Apollo Program, transforming U.S. industrial capacity. The Atlas Intercontinental Ballistic Missile was deployed in five years and required simultaneous breakthroughs in rocket engines, lightweight structures, inertial guidance, and reentry physics. The Apollo moon landing came eight years after zero human spaceflight capability. The Archangel spy plane program went from concept to the first flight of a plane that could fly 85,000 feet in the air at speeds exceeding Mach 3 in four years. Boeing took the 747, the largest and most advanced commercial aircraft ever built, from initial concept in 1965 to planes rolling off production lines 28 months later in a project that required building a factory that remains the largest building by volume in the world today.

The English novelist J.G. Ballard would later compare the 747 to the Parthenon, stating that both gave physical shape to “mathematics, aesthetics and an entire geopolitical worldview.” The geopolitical dominance the U.S. gained through aerospace and the semiconductor and technology industries that emerged from space and defense spending proved extremely difficult for rival nations to duplicate.

In an attempt to catch up, the KGB’s Directorate T coordinated deep-penetration espionage through Line X into Western R&D programs, scientific institutions, and high-tech industries. The scale was substantial and often effective. Soviet intelligence obtained detailed designs, manufacturing processes, and software for advanced computing, aerospace, and industrial control systems.

The difficulty faced by the Soviets lay not in collection but in application. Soviet industry lacked production flexibility, precision tooling, and institutional incentives to improve foreign designs. For example, despite extensive intelligence on Western semiconductor design, Soviet chip fabrication suffered low yields, high defect rates, and an inability to scale down transistor size.

Directorate T suffered a decisive disruption in 1981, when one of its senior officers, Vladimir Vetrov, provided French intelligence with thousands of classified KGB documents. The material, later known as the Farewell Dossier, revealed the extent of Soviet industrial espionage and enabled Western intelligence services to dismantle networks and exploit Soviet dependence by allowing compromised designs to be acquired. In the early 1980s, sabotaged industrial control software used by the Soviets for a Siberian gas pipeline resulted in the largest non-nuclear explosion ever observed from space.

Britain and the U.S. built industrial power on stolen knowledge. But their ability to incorporate this knowledge into powerful instruments of expansion enabled superpower status. Soviet inability to do the same led to technological stagnation. China’s rise matters because it appears to have solved the institutional problem that defeated the Soviet Union.

Why China Is Not the Soviet Union

According to Dan Wang, author of Breakneck (2025), the Chinese Communist Party has closely studied the history of two countries as it planned China’s rise: Japan and the Soviet Union. Much of its industrial policy playbook, including strategic state investment, industrial espionage, and forced IP transfers, is a supercharged version of the playbook used by Japan in the post-war era. From the Soviet Union, China learned how not to run an authoritarian state and where market forces could produce better outcomes than central planning achieved.

China’s hybrid system combines state direction with market discipline. The country provides a firehose of cheap capital to companies in strategic industries and removes regulatory barriers, but it forces them to engage in a brutal survival-of-the-fittest competition. There are 137 electric vehicle companies in China, but only a few of them are currently profitable.

Huawei illustrates the Chinese developmental model. Over three decades, China provided Huawei with an estimated $75 billion in state support. This support took several forms. State-owned banks extended preferential loans to Huawei at below-market interest rates, while government procurement contracts for China’s $1 trillion national telecom infrastructure buildout — which began a few years after Huawei’s 1987 founding — favored domestic firms. This support allowed Huawei to dominate its home market by replicating stolen Western IP. Domestic profits then funded R&D that produced incremental improvements to this technology. Within just over a decade of its founding, these refinements enabled Huawei to offer a full range of network infrastructure products that matched or exceeded Western alternatives in capability while undercutting them by around 40% on price.

Huawei distributes its $22 billion annual R&D budget across more than 15 research centers globally with nearly 100,000 people working in R&D roles, which has enabled it to hold over 110,000 patents and create arguably the best 5G technology in the world. Its progression from equipment reseller in the early 1990s to the largest telecom equipment provider globally by 2012 demonstrates its institutional capacity.

Chinese Premier Xi Jinping wants to duplicate Huawei’s success in every major industrial sector where the West, Japan, South Korea, and Taiwan currently lead. He identified using scientific and technical innovation to create “new productive forces” as the central development goal of his administration in 2023. This meant transforming traditional industries through automation while nurturing ten strategic sectors ranging from electric vehicles and robotics to aerospace and medical devices. China identified these sectors in 2015 as part of its “Made in China 2025” plan, which was an attempt to formalize a state-directed strategy of technological catch-up, substitution, and eventual dominance in high-value manufacturing.

The only realistic way to collapse R&D timelines that quickly is through industrial espionage. In the two years before announcing Made in China 2025, China’s primary intelligence service, the Ministry of State Security, doubled in size in preparation. In July 2020 testimony before the Senate Judiciary Committee, FBI Director Christopher Wray stated the Bureau was opening a new China-related counterintelligence case approximately every ten hours. Almost all of them involved economic espionage or technology theft.

The American Institutional Challenge

Preventing Chinese industrial espionage and denying access is important, but rebuilding the institutions that once allowed rapid industrial scaling is more important.

Boeing’s aircraft development timeline illustrates the transformation of American industrial capacity. The 747 program was delivered in 28 months. The company’s next clean-sheet commercial design won’t arrive until the middle of the next decade, roughly 30 years after its last clean-sheet design, the 787 Dreamliner, entered production. The problem isn’t lost technical knowledge. American engineers still understand how to design aircraft. The problem is institutions that no longer function as they once did.

In stark contrast, the Chinese semiconductor equipment firm SiCarrier formed in 2021. Four years later, it unveiled over 30 products named after Chinese mountains that covered most wafer fabrication processes except lithography, each representing decades of accumulated Western and Japanese process knowledge that China collapsed into a single developmental push. The disparity in commitment shows in the numbers: per the IMF, China invests 4.0% of GDP on industrial policy while America invests 0.39%. Compounded over decades, that gap built manufacturing ecosystems that cannot be replicated on demand.

It should be acknowledged that China’s developmental model contains flaws. China’s province-level industrial policy has generated what economists call “involution,” or destructive competition that produces massive overcapacity. Political economist Victor Shih estimates China’s debt-to-GDP ratio at 200%, while the Financial Times found 12% of Chinese firms are “zombies” unable to service debt. China now produces 14 million more cars than its market absorbs annually and manufactures double the solar panels global demand requires, which has resulted in its solar manufacturers suffering $60 billion in losses. There are unsustainable features of both Chinese and American developmental models and it’s unclear whose dysfunctions will be harder to fix and whose will end up being fatal.

Many current U.S. policies reflect assumptions formed during a period of uncontested technological dominance by the United States that followed World War II. The United States built a coordinated system, including research universities, national laboratories, DARPA, and a defense industrial base supported by almost unlimited spending. That system degraded after the Cold War ended in 1991, when the United States no longer faced a peer rival and it allowed its defense industrial base to contract while other manufacturing moved offshore.

The U.S. still has time to recover many of the capabilities it lost, but its assumption of technological advantage needs to be abandoned now that it faces a Chinese rival that has organized its state systems around systematic technological catch-up. Institutional reform matters most, rebuilding or developing new institutions, companies, and industrial policies that create new instruments of expansion. These instruments could include greatly increased spending on industrial policy, requiring IP transfers as part of joint venture agreements for firms wanting to build factories in the U.S., applying AI to industrial problems, developing entirely new R&D systems, or even engaging in industrial espionage for commercial advantage using the same talent recruitment, cyber espionage, and human intelligence methods China does.

China hasn’t just stolen technology; it’s rebuilt the American system of the 1960s while America dismantled its own. The Chinese EUV “Manhattan Project” reveals a competitor treating technological supremacy as existential while America treats it as one priority among many.

The U.S. built its first ICBM in five years because doing so was viewed as a matter of national survival. If the U.S. wants to remain a superpower that controls its own destiny, it must recover that intensity.

European powers continued to intrude, most dramatically France, which imposed a short-lived puppet monarchy in Mexico during the American Civil War. Meanwhile, Britain retained naval predominance in the Caribbean and Spain held Cuba and Puerto Rico. Only with the rise of American sea power in the late nineteenth century did doctrine meet capability. The defeat of Spain in 1898 announced American hegemony, and after European powers blockaded and bombarded Venezuela over delayed debt repayments, the Roosevelt Corollary of 1904 formalized Washington’s claim to police the hemisphere, legitimizing repeated American interventions in Caribbean and Central American countries to secure political order, credit, and trade.

This order more or less persisted through World War II when hemispheric priorities shifted to area denial against Axis forces and securing the maritime approaches to North America. The 1947 Rio Treaty was conceived as a codification of this logic, a hemispheric counterpart to NATO’s principle of collective defense. Its purpose was to keep the Americans in, non-hemispheric powers out, and Latin America stable, preventing its long tradition of revolutions, coups, and debt crises from inviting outside intervention.

In 1959, the very scenario that the treaty was meant to prevent came to pass when Fidel Castro toppled the Cuban strongman Fulgencio Batista. Cuba has sought to undermine American primacy in the Western Hemisphere for over six decades since. It used its real estate to host the Soviet Union’s nuclear-armed IRBMs, along with naval and air assets. Its SIGINT bases at Lourdes hosted the Soviet (later Russian) intelligence services from 1962 to 2002, while a similar facility at Bejucal is rumored to operate with the help of China’s PLA. Cuba’s formidable intelligence apparatus has dealt numerous blows to American security by recruiting well placed spies, like Ambassador Manuel Rocha and the Defense Intelligence Agency’s Ana Montes, and has trained and supplied revolutionary leftist groups across Latin America, from Nicaragua to Bolivia and Colombia.

After a long period of relative inactivity, the conflict between the United States and Cuba reached a crescendo on January 3, 2026, in Caracas, Venezuela, where dozens of troops representing Cuba’s armed forces and Ministry of the Interior were killed by Delta Force in the raid that captured Nicolás Maduro. The Cubans were providing personal security to the Venezuelan president, whose oil revenues and exports were a crucial element of Cuba’s hemispheric strategy. Now, with its key oil supplier gone and its principal line of regional influence severed, Cuba’s government faces a more uncertain future than at any time in the last half century.

For nearly three decades, Félix Rodriguez served on the front lines of the covert war waged between the United States and Cuba. Born in Havana in 1941, Rodriguez went into exile after Castro’s revolution and later took up arms against the communists. Throughout the 1960s he served as a CIA paramilitary in Latin America. In 1967, Rodriguez took part in the capture of the Argentine revolutionary Che Guevara, later serving in Vietnam and volunteering with the Salvadoran Air Force during that country’s civil war. In light of the escalating tensions between the United States and Cuba, I decided to reach out to Félix, now 84 years old, to discuss his life.

What follows is a transcript of our conversations, which took place on January 9 and 29, 2026. The transcript has been edited for length and clarity.

✺

Carson Becker: Could you tell me a bit about your family history in Cuba? I understand you’re related to Alejandro Rodríguez Velasco, the first elected mayor of Havana.

Félix Rodriguez: That’s correct. Alejandro fought with Máximo Gómez. He was a two-star general in the Cuban War of Independence. And then he became the first mayor elected in the city of Havana.

CB: What are your most striking memories of Cuba before the revolution?

FR: I enjoyed my youth in Sancti Spíritus, my hometown. We used to have a recreational farm for the whole family that I used to visit. I enjoyed having breakfast at four o’clock in the morning, bringing my coffee and getting the milk directly from the cow and adding some sugar and everything, you know, it was happy. I used to ride horses and do hunting and fishing on the farm.

CB: Are there any noticeable differences between Espirituanos and Habaneros? Do they have a different kind of culture in those two cities?

FR: Santi Spíritus was more traditional. It was founded even before Havana, so it was a very traditional city in the middle of the island. I really, really enjoy being considered an Espirituano. I was born in Havana because there were problems with my birth, so they had to get the best doctor for my mother, and that was the one in Havana, so I was born there. But the following day, I went back to Santi Spíritus, so I consider myself lucky to be an Espirituano. And I really enjoyed living there. Every family knew each other. It was a fantastic feeling.

CB: I believe your uncle was a minister for Batista. You wrote in your memoir you actually met Batista as a child.

FR: My uncle was a senator in Cuba. He helped Batista assume his senatorship in our province, even though Batista was not from my province. So they became very close friends, and when he went and took over Cuba, he made my uncle the Secretary of Public Works. And then later they made my uncle the president of a government institution which owned a lot of the equipment to build roads. It was a combination of the people and the government in building roads in Cuba. The government would provide the equipment, and the people would supply the labor. They did a fantastic job all over the island, and it was very effective. It opened the roads to trade for the farmers, so they could take their goods to market.

CB: What was Batista like when you met him?

FR: When I was a little kid, I think four or five years old, he came to our farm with my uncle, near Santi Spíritus. That was when I first met him.

CB: How was Cuba doing in the 1950s? There were clearly some social problems, but it also seemed like it was on track to become a much more prosperous country thanks to tourism and development.

FR: Oh yes, communism destroyed the whole thing. But in 1958 we were probably one of the most developed countries in the hemisphere. We had sugar, which dominated the world market, and we had tobacco. We were way ahead of many Latin American countries, including Argentina. Everything communism touches, it destroys. And that’s exactly what it did. Look at Venezuela, for example, formerly the richest country on the continent. Now, fortunately, they got Maduro out, but Venezuela was destroyed when communism took over the country.

CB: Why do you think the revolution had so much support from working-class people? What do you think was the key to Castro’s success?

FR: Being up in the mountains, all of that created a real charisma. And he was a charismatic guy. We cannot deny that. A lot of people thought that they could do better and supported him. And then everything that he promised, he didn’t go through with it — he actually destroyed. And the first thing was democracy. He disappeared people in Cuba. He became a dictator. That was sad, and that explains the downfall of the economy of Cuba; all of that was due to his regime.

CB: What inspired you to potentially risk your life by joining the Anti-Communist League of Trujillo? You were quite young at the time, and yet you chose to do that.

FR: Well, I was in boarding school at the time, in Pennsylvania, and the first thing that impacted me was the firing squads. Cuba didn’t even have the death penalty. One guy was executed during World War II for being a spy for Germany. That was the only death by firing squad. And then Castro comes in and kills thousands of Cubans who opposed him. That would inspire anybody to try to save their homeland, the place where I was born. So that’s why I dedicated my life to it.

CB: The Cuban diaspora in South Florida at that time was full of people spying on behalf of Castro. Why so many infiltrators?

FR: Well, they didn’t know who he really was. He promoted one thing and then did something else when he came to power. He gave people hope that things were going to get better. He promised a free and democratic election within a year. People had faith in him at the beginning, and thought he was going to be the savior. It turned out to be the other way around.

CB: What was it like for you in the Dominican Republic, where you had a lot of anti-communist fighters from all over the world?

FR: I was in Mexico at the time because my parents had moved in 1958 from Cuba to Mexico. That was the homeland of my maternal grandmother. I used to visit them from school back and forth. And one time there was this captain who was recruiting people for the Dominican Republic, where they had the Anti-Communist Legion of the Caribbean. I thought that was what I should do. I had already been accepted to the University of Miami for engineering. Instead of doing that that year, Captain Cortez recruited me, and then I went to the Dominican Republic. We were training and all of that.

CB: What kind of people were signing up for this adventure?

FR: We had like 150 Spaniards, over 100 Cubans. There were about 50 Yugoslavians, and then 50 from different countries all over the world that Trujillo got into the country. It was interesting. I was only about 15, 16 years old when I went there.

CB: When did you first make contact with the CIA?

FR: I had been accepted to the University of Miami for engineering. When I came to Miami, that’s when I learned they were training people in Guatemala to liberate Cuba, and it turned out to be the operation that later on became the Bay of Pigs. That was started under the Eisenhower administration, after he saw how much influence the communists had in Cuba. We had no idea it was the CIA. We were told they were rich people from Cuba financing the operation.

CB: What year was this?

FR: Late 1960.

CB: How did you get to Guatemala from Miami?

FR: They sent us to Opa Locka. And from there they sent a plane to pick us up. They flew us in a C-54 plane from there to Guatemala. Before we arrived in Guatemala, because our runway didn’t have any lights, we had to stop in an area called San José. It was a Guatemalan paratrooper base. We had to wait there until daylight, and then from there we flew to our destination in the C-54.

CB: How were you received upon arrival?

FR: When we arrived, they had already completed the base. The group that arrived before us had to participate in building the base, making the barracks, and all of those things.

CB: What went wrong with the Bay of Pigs?

FR: The Eisenhower administration started this operation with the idea of guerrilla warfare in the Escambray Mountains, and having enough people in arms to go and announce a provisional government on the radio station, and then be recognized by the OAS [Organization of American States] and the United States. And that would be the end of Castro. That was the original concept. Eisenhower had a great idea to take over the city of Trinidad. The city was very much anti-Castro; it was right next to the Escambray Mountains.

When Kennedy came to power, he decided differently. And when they ran the Bay of Pigs, the only way that we could really be successful was if we controlled the air. They started under those assumptions. There was an initial air attack that destroyed 90% of Castro’s air force. But then the second one was stopped because of Adlai Stevenson. He wasn’t briefed properly.

CB: During the period leading up to the Bay of Pigs, you were inserted into Cuba. Could you tell me the full scope of this mission?

FR: Well, first of all, we trained in Guatemala, along with everybody else. They took a selected group called the Grey Teams, infiltration teams, and took us to Panama for additional training. Then we went inside Cuba to resupply the resistance, and then support the invasion.

CB: How close did you come to being captured on this mission?

FR: I was lucky. There was only one time we were going in a car, and it was stopped by soldiers. They were doing a search, and we were all there. Three of us had participated in blowing up some oil trucks. But they stopped us, and I put my head out and said “pasar con compañeros” [passing with friends] in a very strong way. I guess they thought we were one of them, so they let us go. That was a close call. This was at one o’ clock in the morning, coming back from trying to exfiltrate from Cuba, but nobody came to pick us up.

CB: How did exfiltration work?

FR: We had a radio operator inside Cuba, and they would tell us time and date and location to be able to exfiltrate.

CB: You were picked up by boat?

FR: Yes, by boat.

CB: You made a couple of different trips to Cuba to try to assassinate Fidel Castro. What was the closest you ever got?

FR: Not even close. He had a very secure perimeter around him. I understand from one of his bodyguards that I talked after the fact that he had a security group numbering up to 10,000 people. For example, if he visited a square in some town, there would be two of their security people in every single window facing that area. So there was no possibility of anybody shooting him from anywhere. He had some of the best security, even better than the President of the United States. It was almost impossible to get him.

CB: What kind of weapons did you bring with you to Cuba?

FR: We would bring machine guns, hand grenades, and other rifles.

CB: Why did the United States adopt the policy of assassinating Fidel Castro, and what role did you play in that?

FR: It was never meant to be. They talked about it, but they never approved it.

CB: The United States never approved his assassination?

FR: Right.

CB: But Cuban exiles organized independent efforts?

FR: There were, but they were not professional.

CB: What was the last attempt on Castro from the exile community that you’re aware of?

FR: One time they claimed they were going to go to an island nearby; they were going to have a boat with a .50 caliber machine gun to shoot him, but he never came through.

CB: In 1967 you got the call for the Che Guevara mission.

FR: We got a call from an Agency guy by the name of Larry Sternfield. He met with several of us, like 16 of us, and he selected two of us. The reason was they knew Che Guevara was in Bolivia, and there was a prohibition from the US ambassador, Henderson, that no US citizen could participate in combat there, because there were so many bodies coming back from Vietnam. They didn’t want to start having that in Latin America. Since we were not even residents, we were selected for that operation, because we didn’t violate that restriction from the ambassador. They selected us to participate directly with the Bolivian troops to track down Che Guevara.

CB: How did you find Che’s location?

FR: At that time, a lot of people believed, including in the Agency, that he had been killed in Africa, when he was in the Congo. When they were able to verify that he was in Latin America, they immediately prepared a training team from Panama to train the Second Ranger Battalion of the Bolivian Army, because they had no expertise. The Bolivian Army had no idea how to combat the guerrillas. And Poppy Shelton was the major from the US Special Forces in Panama who started training the battalion. At the same time, they sent two of us to help with intelligence.

We were lucky then because we had already captured one guy, who told us how Che was moving around. He told us, for example, that when Che moved from point A to point B, he would send a vanguard of ten men about one kilometer ahead of him. He would go in the middle with the majority of the troops, and then one kilometer behind would be the rear guard. In case there was an ambush from any place, he would be protected in the middle. One time, there was an encounter between the regular Bolivian Army and Che’s group, and they killed three members of the vanguard, and when they checked out the names, they knew that Che was in the area. That’s when we were able to convince Colonel Zenteno, the commander of the division, to go ahead and release the special second Ranger Battalion to the operation.

There was a captain of one company who sent people, always in civilian clothes, to talk to the farmers, because sometimes farmers are afraid of the uniform. They found out where Che was hiding, and that’s when they surrounded that area with one company, and that’s where the fight took place, and they immediately captured him.

CB: What was Che like when you met him?

FR: I had this image that was created about him. Then, when I saw him the way he was, sometimes I wasn’t even paying attention to what he was talking about. I just remember a picture of him when he visited the Soviet Union, where he visited Mao in China. And then to see him the way he was, he looked like a beggar. His clothes were torn. He didn’t even have a pair of boots. He wore some kind of sandal. He was in deplorable physical condition. I really felt sorry for him as a human being.

CB: Tell me about your conversation with Che.

FR: At the beginning, Che said nobody would interrogate him. Seeing that attitude, I told him that I didn’t come to interrogate, I just wanted to talk to him, that I admired him. He was a leader of a state, and yet he’s here fighting because he believed in his ideals, even though I know they are mistaken. We started talking about different subjects. He really didn’t want to talk about Cuba or anything like that. He said that Bolivia was an excellent place for a revolution because you have boundaries with five countries. If you take Bolivia, it would be easy to export the revolution across the borders. They had a very poor army, and he thought the United States was not going to be too concerned, because Bolivia was a very poor country. If it were a rich country like Venezuela, the United States would be much more involved than in Bolivia. They always have that fascination about: if it’s a rich country, the v will defend it; if it’s a poor country they won’t.

CB: Were you able to get any meaningful intelligence out of him?

FR: He was very depressed because he was captured.

CB: Was he afraid?

FR: I guess at that point in time, people are, but he conducted himself well. I came and told him he was going to be executed. He said, “It’s better that way. I should have never been captured.” And then he gave me a message for his widow, to tell her to remarry and try to be happy.

CB: He was wearing a watch, right? He was wearing a Rolex GMT at the time.

FR: Every single Cuban had a Rolex. When he was captured, Che had his Rolex and another Rolex of a Cuban who had died in combat, and he was saving it to give it to the widow in Cuba whenever he returned. But that was taken by the captain of the Bolivian Army who captured him. He kept Che’s Rolex and the other guy’s Rolex. There were all kinds of Rolexes there, because every single Cuban that was there had a Rolex. I had my own Rolex, and I exchanged it with one of the soldiers who told me he had Che’s, but it turns out it wasn’t Che’s at all.

CB: What happened to his body after he was executed?

FR: Well, after the execution, there was a meeting, and they decided, first of all, they were going to cut off his head to prove that he was dead. And I told the general, you cannot do that. I said I would cut one finger, and we had a fingerprint from the Argentine Federal Police, so it could be checked. He ordered both hands to be cut. At about three o’clock in the morning, when there was no press, they went with a doctor, and they cut off both hands, and then they took his body and two others. He was buried at the end of the Valle Grande runway, where they had a bulldozer that was expanding the runway. They dug a big, big hole in there, and they dropped the bodies of Che and two of his soldiers. He stayed there for a long, long time.

CB: What kind of world was it that you think Che wanted?

FR: He wanted to be his own man. In Cuba he was always going to be number two, Fidel number one. Che wanted to have his own revolution, but it was a total failure.

CB: Why do you say that?

FR: Because he was the worst guerrilla you could find. He could not even recruit a single guy the whole time he was in the field.

CB: Was he relying mainly on Cubans who were accompanying him?

FR: Some Cubans accompanied him, but he was the main individual in the whole thing. But he didn’t have the support of Cuba, because Castro was told by the Soviet Union — they didn’t want him to succeed — because Che was pro-Chinese and Cuba depended solely on the Soviet Union.

CB: Across all of your missions, from Cuba to Latin America to Vietnam, what would you say was the most dangerous?

FR: Well, in Vietnam I was shot down five times in helicopters, and that’s why my back got all screwed up, and then finally they evacuated me for medical disability.

CB: How long was it before you were able to fly again?

FR: I flew with the Salvadoran Air Force from 1985 to 1988.

CB: In El Salvador, were you there on your own business, or on behalf of the US?

FR: I went as a volunteer. Nobody was paying me.

CB: Not even the Salvadorans?

FR: The only thing that I got was a place to stay and food to eat.

CB: When you were in Nicaragua and El Salvador in the 1980s, what kind of Cuban involvement did you encounter?

FR: Cubans were involved, and they were supporting them with weapons. All the Salvadorans had M16s from Vietnam, and Cuba was the one who channeled those rifles to them.

CB: Were they active on the ground?

FR: Yes, they sent some mortar experts.

CB: How are Cuban operatives influencing things in Latin America? Is it all directed by Havana, or is Havana sometimes taking orders from Moscow and Beijing?

FR: Havana runs itself, even though they may have requests from Russia and China. But Cuba controls the whole thing, and they have a good intelligence service. We cannot deny that.

CB: Even recently, a former US Ambassador was arrested for spying for Cuba.

FR: Yeah, I know. He was a friend of mine.

CB: Interesting.

FR: I think he was working with them for a long time. I didn’t realize that because we had him down as a strong anti-communist. Manuel Rocha. And I met him because at one time he was in charge of the US Interests Section in Havana at the Swiss Embassy. He was supposed to be very, very much anti-communist. He was probably recruited in Chile a long time ago. He surprised the hell out of me when I found out because he’s a very, very intelligent man. I can’t understand how he could cooperate with those people. Knowing what they have done to the Cuban people, they destroyed our economy and everything, it really surprised the hell out of me when I learned that Rocha was doing that.

CB: Are there any portions of your career that you are unable to ever disclose?

FR: Yes.

CB: If Cuba’s regime were to change, what would be the first thing you would want to go back and do?

FR: I’d like to go back and see my hometown before I die.

I spent a long career at the CIA but I rarely write about it. The myth is too pervasive and exciting, so that anything I said on the subject would sound like mere prose. In Latin America – during my time at least — not a sparrow fell, but the CIA was responsible for its cruel untimely death. Here in the United States, many people think of the Agency as a shadowy prime mover, hiding behind the façade of our democratic government and “terminating with extreme prejudice” anyone who gets in the way. No doubt the grandees on the seventh floor at Langley wield the image to their advantage, but these fantasies are entirely self-induced. There’s something about turning secrets into a multi-billion-dollar business that makes people crazy.

In everyday life, keeping secrets is associated with a guilty conscience. Something bad has been done that must be concealed. People who cheat on their spouses go to great lengths to conceal their actions, for example. By extension, a secret organization like the CIA takes on an almost orgiastic aspect, staffed by tens of thousands of spouse-cheaters enjoying a lavish lifestyle while literally getting away with murder. At that point, a predictable moral inversion occurs. Cheating and lying under an official seal of approval becomes, to the public mind, irresistibly alluring. One may condemn what seems like the institutional immorality of the CIA, but one can’t help but look. One can’t help but dream, either. It’s espionage porn.

Enter James Bond. And Jason Bourne. And George Smiley. I should say, enter Hollywood, which long ago discovered that the myth of the spy, with all its moral unsavoriness, could be commodified for big bucks. According to the AI platform Claude, Hollywood has produced approximately 1,500 movies about spies and their dark secrets, including 27 from the inexhaustible Bond franchise (which is about 20 too many), 5 centered on Bourne and his kind, 9 directed by Alfred Hitchcock, and 10 — some of the best – based on the novels of John Le Carré. Only the Western and the detective story genres surpass the spy flick in sheer volume, yet the spy film, let me suggest, stands apart, because it tramples on the sacred formulas of American cinema. From Stagecoach (1939) all the way to Unforgiven (1992), Westerns delivered heroic cowboys and outlaws who stood and fought for their honor. Movie detectives like Sam Spade in The Maltese Falcon (1941) and Jake Gittes in Chinatown (1974) skirted the law and ticked off the cops, in a quest to satisfy a personal moral code. 

But Bond was just a cad. He slept around, drank too much, and told little jokes when he killed an opponent. He didn’t appear to care about the British Empire, winning the Cold War, or anything much. That was what we loved about the guy. He moved through a secret dimension in which the moral polarities had flipped, and his immense cool left him immune to commitment.

The public has received a profound and detailed knowledge of the spy business from Hollywood. It’s useless to argue. Every denial just proves the point. Working for the CIA, everyone knows, must entail an amazing amount of superficial razzle-dazzle: driving an Aston Martin with rotator machine guns in the headlights, making love to beautiful women in exotic locales, fighting off evil masterminds on top of Mt. Rushmore or the Statue of Liberty, having your body chemically altered so you can become…well, Matt Damon. And if that job description sounds blatantly insane, the tendency is to believe that some downscaled version of it is real — maybe it’s a RAV4 instead of an Aston Martin, and those machine guns don’t really rotate.

The film-makers’ true obsession, however, is with the inner rot of the spying life — the constant betrayals, the cheating and lying, the death of trust. Probably, it reminds them of working in Hollywood, only with a noble motive, patriotism, and an actual license to kill. A spy is trained to betray, but what’s the stopping-point? In The Good Shepherd (2006), the protagonist betrays his mentor not once but twice, supposedly for the greater good: the latter “knew too much.” But what does that mean — and who decides? 

A person who lies to family and friends no less than to hostile operatives, who often conducts business under an assumed name and fake identity, must suffer the dueling temptations either to climb out of the lies for a gulp of fresh air or to drown in an ocean of self-deception. All this is a golden gift to the scriptwriter, who can attain what in the film world passes for moral complexity without having to create believable characters.

✺

The alert reader will have figured out by now that I’m not a fan of the spy genre, but there are exceptions. A personal favorite is Three Days of the Condor (1975), in part because it has a true hero, played by Robert Redford, who survives by his wits rather than his lethal toys, but also because Redford is shown working in a fictionalized version of my old CIA unit. He isn’t a spy, any more than I was. He reads books, looking for unusual patterns. We monitored the world’s news media, looking for interesting developments. 

Three Days of the Condor delivers the full package — mind-boggling conspiracies, endless betrayals, heaps of dead bodies, and a 24-hour love affair. None of those things happened to me. I just drove to work in the morning and drove back to my family in the evening; if there was a crisis, I worked overtime. The distance between Redford, with his chiseled features and magnificent hair, and the rest of us schmucks who worked in my outfit, can stand as the ultimate symbol of the difference between Hollywood’s red-carpet ideal of espionage and the suburbanites who in fact toil at the Agency.

Condor begins with a massacre and ends with an assassination. In between, lots of people die, casually and wantonly. That never happened to me either. I never killed anyone and here I am, mostly alive in my decrepitude. And think about it: even the mafia doesn’t kill unless it has to. It’s bad for business. If your profession is to lead a secret life, leaving a trail of corpses is a sure way to get pushed out of the closet. The carnage, however, is cinematically necessary to display the moral stain that sticks to working for the CIA. 

The most sympathetic character in Condor, other than Redford’s, is the assassin, Joubert, played with sinister aplomb by Max von Sydow. Joubert is a dispassionate professional, willing to kill for whatever side pays best, amused by the observation that, in this game of lies, the players must lie first of all to themselves and believe in a noble cause. “There is no cause,” he tells Redford. That’s the pervasive mood of the movie. There is no cause. The enemy is us. Inscrutable kingpins manipulate the public like puppets. With notable exceptions like Argo (2012), that, too, has been the theme pushed by spy films in recent decades. Plots demonstrate that idealism is childish if not dangerous. Serving Uncle Sam is dirty business.

Was I lying to myself when I worked at the Agency? By definition, I can’t tell. But I was naïve enough to believe in the cause of freedom. And when I think of an institution that sells itself for money, promotes delusion and pretense, and is wholly lacking in principles and morals, the CIA isn’t the first name that comes to mind. The glamorous but amoral movie spy never existed in history. The whole genre is a mirror to its makers — it’s Hollywood, projecting.

The Reality

I am often asked how the CIA managed to “recruit” me. The question evokes whispers in foggy street-corners with faceless strangers wearing trench coats. In reality, I answered an ad in the newspaper. I took a few tests, endured a long and rigorous vetting process, including the damn polygraph, then I was in.

So I come to you from the inside bearing startling news: the CIA isn’t really about spying. It isn’t about covert action or terminating anybody. The mission is to get the best information in front of the right official — usually the president — in time to make the smartest decisions on behalf of the United States. 

Now, I say “information” but insiders prefer “intelligence.” For example, the National Security Act of 1947, the Big Bang to a large chunk of our government, assigns to the CIA the duty “to correlate and evaluate intelligence relating to national security.” Yet the distinction is meaningless. It sounds like a joke, and maybe it is, but nobody inside the Intelligence Community has ever come up with a satisfactory definition of “intelligence.” Ask an intelligence collector or an intelligence analyst what intelligence means, and you’ll get a shrug and some version of “Well, you know, it’s our stuff.”

When I signed on, the Agency was set up in three directorates: for operations, technology, and analysis. To confuse the enemy, the names have since bounced around, but the business areas never changed.

Operations housed the spies. A spy is a collector of a special kind of information — secrets. But it should be obvious that it’s the content of the information that matters; secrecy just means it’s harder to get. Unfortunately, this truism was often lost on CIA management. One of my directors, George Tenet, once told the troops, “Our business is stealing secrets.” Besides being bizarrely misguided — the Agency’s business is to understand the world — the statement told us Tenet had watched too many Bond flicks. (He also said that finding Iraq’s weapons of mass destruction would be a “slam dunk.”) If secrecy is the standard of value, then all self-respecting intelligence officers will engage in an arms race of classification, until even the janitor will label the broom closet Top Secret. Since operations is the gravitational center of the CIA, and since spies deal in secrets, that is precisely what happened.

In my limited experience, most Ops people were courageous and highly skilled. A few were arrogant and ignorant of foreign cultures. All served a valuable purpose. Nations keep secrets other nations desperately wish to learn: the Chinese plan for the invasion of Taiwan, for example. In the struggle for survival, virtually all nations run spies to try to get at that. The derring-do of the movies can happen, but rarely. Some environments are tough to crack, and an occasional Top Secret Ops cable will curiously recall what one just read in the New York Times. Otherwise, keepers of secrets in foreign places can be bought with money or women. The United States always enjoyed an unfair advantage over other nations, because many foreigners with access to secrets considered us a bastion of freedom and were willing to risk, and sometimes give, their lives to help our cause.

✺

Technology, Marshall McLuhan wrote, is an extension of the human sensory apparatus. Spy technology enables the collector to see, hear, remember, conceal, and reveal beyond what’s humanly possible. The CIA’s relationship with technology has followed an interesting trajectory. In my early years, it was cutting edge, having engineered marvels of overhead photography and miniaturization. Then, led by Silicon Valley, the private sector left the Agency in the dust. There was no way for the government to compete. It was reduced, for example, to mandating the maximum resolution allowed to corporate satellite photography — but that could be enforced only on American companies. If you were Japanese, you could pick out a grain of sand on the beach. The attempt to suppress competitive technology collapsed from sheer inadequacy — and thus Google Earth was born.

The internet, too, posed a fundamental challenge. The cult of secrecy presumed a tiny amount of extremely valuable information. The internet flooded the system with near-infinite volumes of variably useful stuff. If practically every intelligence question had an answer online, who needed secrets? The CIA was wholly unprepared for, and culturally traumatized by, the internet. The infrastructure, built for secure communications, got in the way. For some time, I had to do my searches at home. That improved later, but there was always the unspoken dread that we had stepped on black ice and were about to fall on our hind parts. The spirit was willing. We were urged by our managers to embrace the web, but we always ended up embracing our preconceptions. Attempts to build an “intelligence search engine” were baffled by the basic problem of not knowing what intelligence was.

Yet I sense that spy tech is on the upswing. In Palmer Luckey, founder of Anduril, the intelligence and defense communities have acquired their own slice of entrepreneurial genius. Luckey’s obsession is autonomous drones — and the flock of drones that descended on Nicolás Maduro’s defenders in Venezuela was positively cinematic. The “sonic weapon” deployed in that confrontation, too, seemed straight out of science fiction. It remains to be seen whether Luckey or anyone else can get the culture past the digital age and into the wild frontier of AI — but it appears, at least, that the CIA is approaching its 80th birthday having rediscovered the cool gadgets of its youth.

✺

Sherman Kent, who in the 1950s gave form and substance to the analysis wing of the CIA, was a disciple of Walter Lippmann’s. Like Lippmann, Kent believed in Platonic truth. Every intelligence question was like a jigsaw puzzle, with one and only one right answer. Kent also agreed with Lippmann’s insistence that “the power of the expert depends on separating himself from those who make the decisions.” In the world according to Kent, the analyst arrived at truth by rising above passion and policy. To his political masters, he delivered an “estimate” of reality and issued “warnings” of the future.

I want to separate the people who perform analysis at the CIA from the method they are stuck with — the burden of Kent’s legacy. Almost without exception, the Agency analysts I encountered were smart, knowledgeable, and articulate. They did amazing work under stressful conditions, some of which saved lives and promoted our country’s interest. Could they predict the future? Of course not — nobody can. Could they cobble a single interpretation out of branching and disputed facts? Only through a deeply bureaucratic process not unlike what Communists went through when they arrived at the “party line.” Could they tell the president, who cared only about how to implement successful policies, that their analytic remit was limited to the pursuit of pure truth? There began a trail of tears that typically concluded with the president feeling betrayed by the Agency and the Agency whining about “politicization.”

The CIA was shocked to the core by the fall of the Soviet Union. I was there. Our biggest strategic antagonist for 45 years seized up and died, and we had no idea it was happening. The CIA missed the initial test of the Soviet atom bomb — and India’s bomb, and Pakistan’s as well. 9/11, the sort of disaster the Agency was erected to prevent, came as a complete surprise. In hindsight these episodes appear inevitable and thus predictable, but in fact most historic discontinuities are extreme low-probability events. Place the filter of Platonic truth over them, and they disappear from sight. It would have been career suicide for an analyst to brief, “Mr. President, we estimate there’s a 1 in 1,200 chance that terrorists will crash airliners into the World Trade Center and the Pentagon.” Money and glory attend to the immediate and obvious.

Trapped in an impossible situation, analysts developed survival mechanisms. For one, they wrote too much. A blizzard of classified material blew out not just from the CIA but the entire Intelligence Community. Nobody read the stuff, but if something unforeseen occurred, we could be sure that at least one document had mentioned the possibility. The analytic style was also a hedge against failure. Robert Gates, then head of the directorate of analysis and later director of CIA and secretary of defense, commissioned a logician to scrutinize the language of the President’s Daily Brief, or PDB. The logician discovered a large measure of unclarity in the PDB. The same words had different meanings across time. It was frequently hard to tell whether a prediction was being made or not. The analysts who did the writing tended to be brilliant wielders of the English language. It was the blessed Sherman Kent who contorted their work.

✺

I want to register one final observation about CIA personnel. When they go out into the world, they wear a bull’s-eye on their backs. I don’t know if the Agency hires real-life versions of the assassin Joubert to kill its enemies, but I do know its officers get killed in the line of duty. Many died during my tenure. The stars covering the wall in the lobby of Headquarters Building, each symbolic of a nameless death, more than doubled in that time. A special quality is needed to know the risk, as everyone does on the inside, and still get on that airplane to Abidjan. The CIA, without question, is a Washington bureaucracy — but it’s not the Department of Transportation.

The Potential

The superhero edition of the CIA would provide the exact information needed for the president to triumph in the world, everywhere and all the time. But that’s a Hollywood fantasy. The real world is full of ugly choices, and failure is built into the human condition. The goal is to maximize advantage and preclude catastrophe, and for that, a firm grasp of our government’s strategic priorities — the “requirements process” in intelligence jargon — is a precondition. Since the end of the Cold War, the Agency hasn’t prioritized particularly well — but let’s pretend otherwise. Let’s imagine an organization willing to roll with the 21st century.

What would change?

The most meaningful improvement at scale for the Agency would be to kill the cult of secrecy and redirect resources towards the dominant information structures of our century: the web and AI. The government will never lose its appetite for secrets. While technology can satisfy much of this hunger, we aren’t about to pension off our spies. It’s a question of perspective. “Stealing secrets” is expensive and carries great human and diplomatic risk. Covert sources can play us false — that’s what happened in Iraq. The culture must be liberated from an addictive dependence on classification; Top Secret should never correlate to great authority. This is particularly true in the age of sexting and performative elites, when enough secrets get spilled online to make a grown spy cry.

As Robert Redford’s character in Three Days of the Condor would remind us, even before the internet the immense majority of intelligence material was collected from “open sources” — news media, books, government and corporate publications, etc. After the arrival of the web, the disproportion ballooned exponentially. Open information is faster, nimbler, cheaper, and much less dangerous to obtain. The Agency knows this, and occasionally will acknowledge it with a wave of the hand. But it has never acted on it, never put its money there. Although criticism after 9/11 and Iraq forced the establishment of an Open Source Center — my home turf — the unit was ridiculously underfunded and subservient to operations. 

If the CIA’s business is to understand the world, then a major part of that mission should be to understand the web at great depth. For every digital utterance, the analyst must be able to penetrate beyond author and site to provider, location, funding, ideology, past history, connection to similar posts elsewhere, affiliation with state and non-state actors. Analyzing video should have primacy over text — this is alien to government thinking but it’s the way of the web. The digital universe is a huge and shifting target. Powerful AI applications will keep track of billions of moving parts, constructing a dynamic map of digital space in the manner of the 16th century explorers, placing the warning when appropriate, “Here be monsters.” 

Skeptics will argue that all online material is horribly tainted — that the internet is the mother of lies. That would be accurate and all to the good. To the propaganda analyst, disinformation is a moveable feast. Among many benefits, it can provide an answer to the most difficult intelligence question to ascertain: intent. The point, after all, isn’t to strive after Platonic truth but to extract knowledge about how the world works.

✺

In another systemic change, the Agency will tear down the Berlin Wall between intelligence and policy. This can’t be accomplished universally — one can’t ask the crusty analyst who’s been counting beans for 30 years to brief about how best to overthrow the government of Cuba. A new class of hybrid specialists must be brought into being whose careers will be threaded through both intelligence and executive assignments. Organizationally, they will be housed in a single CIA unit — call it the Policy Intelligence Team, or PIT for short. Physically, they will reside in the White House, where they can be consulted by the president at will. The latter will indeed ask, “So how do I overthrow that wretched regime in Cuba?” The answer will be, “Here are the possibilities and our assessment of their merits.” 

PIT members will be free to brief on high-impact but low-probability events like 9/11, so that the president can take whatever preventive measures he sees fit. In the White House, they will be looked on as helpful advisers rather than sporadic visitors who mainly talk mush. Members of the team will rotate in and out according to the president’s interests. When needed, subject area experts — the bean counters — will be brought in to brief the president on an ad hoc basis.

It will be noticed that PIT will make the traditional early-morning PDB briefing redundant. That’s okay. Let the president sleep late. The PDB long ago ceased to be a means of information exchange and became, on both sides, a status symbol. It’s been called “the president’s newspaper” — and the daily newspaper is going extinct.

✺

One last suggestion about “finished” intelligence: there should be a lot less of it. Mass production, self-evidently, is a dumb model for our nation’s secrets, since the more there is, the more that will be buried, leaked, or lost. Analysts shouldn’t be rewarded for how much they publish but for how useful they have shown their knowledge to be: in the Intelligence Community as in academia, publish or perish has resulted in vast masses of published trivia. (There should probably be fewer analysts too, but I’m not going to say that out of loyalty to the tribe.) This applies with particular force to Top Secret material. The label should be applied sparingly, to protect sensitive content and not to promote its importance.

On every topic of interest, at every level of classification, items will be linked and synthesized by AI. Rather than a scattershot of data and analysis, the Agency will have a running story on, say, Russian war-making capability or Iran’s efforts to resurrect the bomb. A constant criticism of the CIA is failure to “connect the dots.” The accusation is vague and probably unfair, but it should never again be heard in the era of the thinking machine.

This past holiday season, conspiracy theorists around America rejoiced when what may be the greatest conspiracy theory of our lifetimes had, at long last, been made hyper-legible: the saga of Jeffrey Epstein. Epstein, a financier of murky origins who cultivated relationships with the rich and powerful, was arrested and indicted on federal sex trafficking charges in July 2019. But he died by (apparent) suicide in jail before his trial on August 10 of that year, denying the public answers to questions that theoretically would have been answered in a court of law in a public trial — like how Epstein made his money, and the nature of the involvement of his network of princes and presidents. In the absence of a trial, much of what the public knows comes from the “Epstein Files,” a collection of documents released through lawsuits, court orders, and investigative reporting. These files are the documentary residue of his life (emails, flight logs, photos) that were kept under lock and key by lawyers and government bureaucrats.

The Epstein Files Transparency Act, signed into law by President Trump on November 19, required the Attorney General “to release all documents and records in possession of the Department of Justice relating to Jeffrey Epstein” by December 19, 2025. The Department of Justice did not meet this deadline, though the first wave of these Epstein Files hit the internet on November 12, when the House Oversight Committee released over 20,000 pages of files from the Epstein estate. But although the full scope of the files remains unavailable to the public, a large fraction of the Epstein Files has been released, including a new drop of about 3.5 million pages of material on January 30, 2026.

Technically, all of the information in the first tranche of files was “public” following the November release. But these 20,000 pages — dumped into Google Drive nesting dolls of folders, .txt and zip files — were extremely difficult to navigate. (My Google Drive froze when I tried to scroll through subfolder 001 in the TEXT folder just to gauge how many documents it contained.) Legibility was not the first priority when these documents were uploaded; the sheer amount of files, combined with the government’s lack of technical finesse, made them nearly impossible to parse.

But these files were once hyper-legible: the ordinary emails, receipts, and other digital detritus of a man who, despite the extreme nature of his “work,” used the same software to run his life as everyone else. So, what better way to make Epstein’s emails legible than to log in to his inbox as Epstein himself? A group of San Francisco twenty-somethings, led by 26-year-old Luke Igel and 24-year-old Riley Walz, are now allowing users to do just that.

Luke and Riley call their project “Jmail”; swapping Google’s ‘G’ for Epstein’s ‘J.’ Luke, Jmail’s ringleader and the CEO of Kino AI, told me over a 6pm iced coffee that he and Riley made Jmail overnight. “We just stayed up on Wednesday night [November 19], and Riley and I made it in Cursor. We ended up being completely correct that this is what the world wanted to see.”

Back in the mid-2000s, part of Gmail’s original novelty was making email searchable, which was impossible in the early days of email because there was no central system that stored and indexed all emails. Today, Jmail’s novelty is making Epstein’s emails — and now, his documents, photos, and flight logs, to name a few of the features of the extended Jmail “suite” — searchable and intuitive to view. The government’s declassification of the Epstein Files did the bare minimum of what was required to make these documents legally “public.” Jmail, by contrast, is actually designed to make these documents legible, and understandable, to a general audience.

Luke had been following the Epstein case since August 2019, after his freshman year at MIT, when he heard the news that Epstein died. “I was at the Formula Student driverless competition in Hockenheim, Germany, living in a military tent for a week.” Immediately after the news broke, Luke’s friend burst into their tent while he and his teammates were eating canned soup, and asked “‘Did you guys hear that that coward Epstein just killed himself? It's like he just got away with it!’ And then the mystique just grew and grew and grew.”

Ever since Epstein died in jail, many people have wanted authoritative proof that Epstein was killed by a shadowy network of powerbrokers to protect their own reputations, believing that the circumstances of his “suicide” are just too perfect to believe otherwise. And it seems obscene that, if Epstein was killed, those who orchestrated his death should be able to get away with such an obvious crime.

For those who believe Epstein’s death was not a suicide, the files may offer a set of clues to solve the mystery of whether some of the elites in Epstein’s network wanted him dead, and succeeded in killing him. And for everyone else, these files are a record of how the ultra-elite communicate when they don’t think anyone is reading: struggling to get a reservation at Gary Danko, recommending private chefs, or arranging meetings with Vladimir Putin. And now, the details of what is likely the greatest crime of the 21st century are available for anyone to read, in bright, Googley technicolor.

When I open Jmail, I get stressed out because I have over seven thousand unread emails. Of course, these are not my own emails, but Jeffrey Epstein’s; still, the prospect of reading them all is overwhelming, a task I have not yet attempted. The logo for Jmail is the classic Gmail ‘M’ envelope, except with a beachy hat on top — colorful and irreverent. What makes Jmail so captivating is that it breaks down the mystique of the Epstein case with its friendliness. The question “what if your Gmail was on vacation?” was the “creative inspiration” for the logo, Luke told me. He used Nano Banana Pro, Google Gemini’s AI image generator, to create the logo. And so Jmail was born.

Jmail.world went live on November 21 with an X post from Riley Walz: “We cloned Gmail, except you're logged in as Epstein and can see his emails.” The pickup was immediate; Riley’s original post has 134,000 likes and 22 million views (and counting). The same day, the San Francisco Standard described the website as “the easiest way to read all the Jeffrey Epstein emails.”

For the Jmail team, getting access to the most recent files in the first place is an uphill battle. “We were able to really quickly trudge through tens of thousands of distinct documents that we had from the White House’s December release. But this one is over a million, and it is in the hundreds of gigabytes, whereas the one in December was less than ten gigabytes. But everyone, all journalists, everyone who is into data hoarding, everyone on Reddit is pointing out that those downloads just get cancelled,” Luke told me. “Riley joked about this last time: it feels like you're on Ticketmaster, because there's a queue. If there's too many people visiting justice.gov, there's like a queue that says your turn is ready.” When I talked to Luke about the January 30 release shortly after it went live, he told me that the Jmail team had spent dozens of hours chasing access to the new files, "sourcing really sketchy download links” and “cross-examining everything.”

Together, Luke and Riley have a long track record of making inscrutable information legible. In San Francisco, Riley is already notorious for his Find My Parking Cops app, which tracked San Francisco’s parking cops using (then) publicly available data from recent tickets, allowing users to evade them, since locations of newly issued tickets inadvertently revealed where the parking cops were. San Francisco’s Municipal Transportation Agency soon turned off the tap for the real-time data feed, which stopped the app in its tracks. But, Riley — who also leaked JD Vance’s Spotify and opened a fake steakhouse in New York for one night — remains legendary in San Francisco for his shenanigans. Luke often spoke about the “Riley Walz touch” in his stunts: “the whole point of the great Riley stunts is that they're polished.” The duo behind Jmail were obsessed with the elegance of their product; if their interface were clumsy or coarse, it wouldn’t have caught on. “We're getting 10,000 people using the Jsuite every single day, which translates to quite a few million users per month. And we peaked at well over a million people visiting in a single day” — that was December 23, soon after the Jsuite expanded following the statutory deadline on the files’ release.

Thanks to Luke and Riley’s work, many “Epstein heads,” as Luke calls these somewhat-redeemed conspiracy theorists, who spent years insisting that there was more to the Epstein story than the official narrative allowed, share information largely via Jmail and other Jsuite products.

As more and more information became available following the December 19 statutory release deadline for the Epstein Files, Luke and Riley kept hard at work. So came the “Jmail suite,” or what Luke refers to as “Jmail 2.” He explained: “Jmail 2 is what happened after Will Depue” — an engineer at OpenAI with a popular presence on X — “was like, ‘is anyone gonna do a build?’” after the Epstein Files Transparency Act’s first tranche was released. “And I was kind of offended by it. And I was like, ‘Fuck you. We already did a build.’” Right as the startup offices of San Francisco were emptying out before Christmas, Luke found himself with a group of hackers who wanted to do something with the newly released files. “And then a bunch of people invited themselves to my house. This happened the day of December 19.”

Aside from the December 19 batch of files, the success of Jmail 1.0 led to access to even more documents: the login to a previously unknown Epstein email, jeeproject@yahoo.com, was added to the Jmail interface after “some guy figured out Epstein’s password — the word ‘Ghislaine’ in all lowercase.”

With some time, Luke’s expanded team managed to display the litany of documents in a full suite of Epstein-themed tools, each mimicking a familiar Google product. “Jdrive” organizes court files and photos by their origin (folders are titled “doj-disclosures,” “house_oversight,” and “estate-production”) and allows users to search by person mentioned in each document. Jdrive also links to Kino, Luke’s AI video editing startup, where government footage of Epstein’s cell is available for viewing. “Jflights” visually tracks Epstein’s private jet and uses manifest records to list who was on each flight. (For obvious reasons, you can’t book flights on Jflights like you can on Google Flights). “Jemini” is an AI chatbot that answers questions about the Epstein Files. (When I asked “How many flights did I take with Ghislaine Maxwell?” Jemini responded: “There were 13 flights that Ghislaine Maxwell was a passenger on. Would you like to see them displayed?”) “Jamazon,” though technically not under the Google umbrella, tracks Epstein’s Amazon orders via his receipts: some of his most recent purchases in the months before his death included The Annotated Lolita, delivered on May 24, 2019, and “Kidoozie Musical Stack & Learn Rainbow - Stacking Activity Toy for Infants and Toddlers 6-24M,” delivered on June 2, 2019.

Of course, there is something uncomfortable about logging into any of these Epstein-flavored apps, but this discomfort is downstream of the evil of Epstein’s crimes — evil that is now legible to anyone with a browser.

When I spoke with Luke about building the Jsuite, it was evident that the project wouldn't have been possible in such a short time a year ago (or even months ago). Much of the Jsuite was “vibe coded,” meaning AI was heavily used to program the different applications. Luke also used Reducto — his MIT fraternity brothers’ startup — to convert “ugly, handwritten PDFs” to parsable files ready for data analysis. “This is technology that has only been around for a very short amount of time, as in, completely open-ended PDF extraction being used to populate an app that you just made with thousands of actual data points in whatever format you want.” Additionally, Igel’s co-conspirator Diego Rodriguez, the CEO of an AI imaging startup, made use of Apple’s new AI model that “came out the day that the Epstein Files came out.” “Diego was like, I love Apple. I have an overpowered Mac Studio. I’m going to use the new Apple algorithm.” And so he used the model to create EpsteinVR, “A full 3D model of Jeffrey's Upper East Side mansion” which was generated using Apple's new SHARP Gaussian Splat model. “Every single startup that could have been involved, we tried to include them and give them a shout out” on the Jmail website, Luke said. “The less vibe coded stuff,” which included manual redacting, “was the stuff we were able to release in the days and weeks afterwards.”

Luke used his own proprietary software from Kino to render the images in Jphotos to “render really large amounts of footage really quickly. It was really nontrivial to be able to render this many photos.” The fast pace of development — often overnight — allowed the Jmail team to quickly react to and publicize new documents and news events, keeping them on the bleeding edge of the news cycle. What once required a team of engineers and weeks, if not months, of work can be built by a hacker or two in hours.

In building Jmail, Luke took pride in “being punkish in a way that is very technically elegant and technically elaborate.” And by punk, Luke doesn’t mean wearing plaid or blasting the Dead Kennedys. Being “punk” means embodying a spirit of anti-institutional irreverence: sticking it to “the man” (whatever or whoever “the man” may be at your particular place in history) without letting your sense of humor be crushed in the process. When I asked Luke whether he got any flak from Google for associating their product with the largest scandal of the 21st century, he laughed. “I respect the hell out of Google. You can really mess with them.”

Conspiracy theorists may be decidedly anti-institutional, but by and large, they don’t have a sense of humor about themselves or the objects of their conspiracies — perhaps because the standard topics that make up conspiracy theories are so horrifying (child sex trafficking rings, manmade pandemics, mass censorship, secret networks of anti-democratic bureaucrats). But while the Epstein case may be a true deep-internet conspiracy theory come to life, there is also something deeply satisfying, and fitting, about making the details of the Epstein case so easily accessible to the masses. When I asked Luke if he had made any tradeoffs when designing a product that could go viral (accessible to “the median voter,” as Luke put it), while also trying to make meaningful “punk” art in software, he told me he was happy with Jmail’s reception. “Punkishness that goes viral has always, always been very close to my heart.” This case study of elite excess and inscrutability can now be viewed by anyone, in primary-colored, user-friendly graphic design.

One of my favorite films about the escapes of elites after hours is the legendary director Stanley Kubrick’s final film, Eyes Wide Shut (1999), in which a doctor in his thirties (played by Tom Cruise) finds himself at a masked psycho-sexual ritual at a mansion outside of New York City. The gathering is populated by the city’s ultra-elite, who engage in secretive, anonymous sexual rites. (It has long been theorized that Kubrick’s death soon before Eyes Wide Shut’s release was the work of Hollywood executives who did not want the contents of their own masked rituals to be revealed.) Part of the film’s appeal is its voyeurism: Cruise’s character, while successful by any normal standards, is squarely upper-middle-class, and only hears of the gathering through an ultra-wealthy patient, after which he succeeds at sneaking in. Through his eyes, the audience is able to gaze at a private world otherwise sealed off. Jmail may be Eyes Wide Shut for the digital age: a brief glimpse into the private activities of the ultra-elite, shown through the eyes of an unlikely intermediary. But the reality of elite conspiracy is far less glamorous than Hollywood fantasy. Scrolling through the Jsuite, one can’t help but be a bit disappointed by the banality of Epstein's day-to-day life: Amazon receipts for rugs and headphones, daily Quora digests, and, like any modern executive, a mountain of unread emails.

He found lawsuits claiming Care.com wasn’t conducting the background checks it advertised. “I thought, there’s one way to test this for myself: try to sign up as Harvey Weinstein,” he told me. It was October, and the avalanche of stories about Weinstein was in full swing. He used Weinstein’s photo, and the email ‘harveythebabysitter@gmail.com’.

The platform asked for his consent for a background check. “I say yes, and I’m documenting this because I wanted to be able to prove that I did this.”

To his surprise, “Harvey the Babysitter” was approved on the site. He made several other gag profiles, and wrote a report about it. The stock of Care.com fell modestly; in what may have been a coincidence, a board member left the next day. For Dorsey, it was the beginning of a major saga. It began when Stanford’s dean of students asked to meet with him.

“When I met with them, they initially said they wanted to meet with me about Wi-Fi policy violations,” he said. “I kind of assumed it was Care.com related, but wasn’t 100% sure.” The meeting revealed that a Care.com co-founder had contacted Stanford. Stanford told Dorsey that he had violated school Internet policy by pretending to be someone else, in this case… Harvey Weinstein!

Stanford warned that further complaints could lead to an ethics investigation, and Dorsey felt it was obvious they were asking him to take down the article. “As a young person, it was kind of terrifying because I had not gotten into any trouble at that point,” he said. He’d go on to publish a second article in June 2018 recounting both the Weinstein gag and a litany of other problems. By this time, he’d taken a position against the stock. He included in the report over 1,000 pages of consumer complaints relating to fake cheque scams, unauthorized billing, safety concerns, etc.

• Care.com approved babysitters are linked to the deaths of at least five children (two of these babysitters had prior criminal histories, another was operating an unlicensed daycare)

• Care.com babysitters have repeatedly abused children

• Care.com’s caregiver screening appears to be nonexistent. For example, I was able to sign up and apply to babysitting jobs as Harvey Weinstein

• Care.com’s background checks have approved convicted felons, prostitutes, and people on probation

• Scamming is common on Care.com and some babysitters have received death threats

• Care.com bills users for unused services and repeatedly bills customers after they cancel their accounts

Dorsey, for his part, was totally vindicated. A year and a half later, a headline in the Wall Street Journal read: ”Care.com Puts Onus on Families to Check Caregivers’ Backgrounds—With Sometimes Tragic Outcomes.” The stock cratered and the company was taken private less than a year later.

While Stanford never officially punished Dorsey, he encountered problems as a rising senior that, to this day, he thinks were not mere coincidences. In Stanford’s notorious housing draw, which no longer exists but for years determined housing for sophomores, juniors, and seniors, he received the worst-possible number out of all male seniors: 999 (1 is the best). So, while the vast majority of seniors had a draw number low enough to get a room to themselves, Dorsey ended up in a converted computer room not by himself but with two others. At one point that year, he lived in a motel for several weeks. “It was the lowest point mentally of my life,” he said. “I kind of just went insane from lack of sleep. I totally became the worst version of myself.”

Building The Bear Cave

Out of this turmoil came The Bear Cave, a newsletter Dorsey started in February 2020 to publish both roundups from the world of short research, and his own detailed reports. “I started the newsletter mainly as a way to get hired because I knew from my past experience that writing stuff online will get you attention,” he said. The Bear Cave started to grow, and eventually he decided he’d make it his job, rather than try to use it to get a job.

The Care.com episode was the first and last time that Dorsey actually took a position against a company on which he was publishing research. “I didn’t love the feeling of having a position against a company financially while also criticizing them,” he explained. So, he stopped, and monetizes his research by charging for access. In other words, he’s an activist but not a short-seller. “It’s a completely unique model, but I think it does come with higher integrity and a different set of incentives.”

At $640 per year with thousands of paid subscribers on Substack (and almost a hundred thousand total subscribers) I don’t need to tell you that The Bear Cave is doing well. At the same time, there’s no doubt that he’s leaving money on the table given how many companies featured in his reports have gone to zero.

The seemingly neverending bull market since 2008 has not been kind to anyone who is broadly bearish. American public equities are probably the greatest assets of all time. But what’s true in the macro picture is not always true in the micro picture, and it’s as good a time as ever to be looking for problems. Short-sellers on Wall Street have attracted an immense amount of attention, not all of it positive.

Michael Burry, the reclusive doctor-turned-investor was immortalized in The Big Short for having bet against the US housing market on the eve of the Great Recession — making billions for his firm and investors. In what may be the most well-known segment of business television in the 21st century, Bill Ackman and Carl Icahn clashed on CNBC over Herbalife and Ackman’s $1 billion short against it. Icahn called Ackman a “crybaby in the schoolyard” and pledged to create “the mother of all short squeezes.” Icahn won, and Ackman would exit the position five years later with a loss of about $1 billion, about as much as Icahn’s gain.

In 2021, GameStop became the epicenter of a massive, coordinated short squeeze by retail investors on platforms like Reddit’s WallStreetBets. Hedge funds shorting GameStop, of which there were many, were wrecked as shares soared from ~$20 in early January to a peak of $483 by January 28. Those retail investors were delighted to see the short-money collapse, not because of some earnest optimism about GameStop’s future cashflows, but to see those they perceived as elite market manipulators get a taste of street justice.

I bring up all of this chaos as a way to explain why Dorsey is quite happy to publish his detailed research and leave it there, rather than deal with the conundrum of short-selling in public.

“In a market built on froth and exuberance and everybody making money, shorts are some of the few people really incentivized to root out the bad conduct,” he said. “I think of the 22 companies I wrote about in 2021, all but one ended down on the year. There was so much low-hanging fruit.”

One such business that went was a drone company, AgEagle, whose stock soared in 2021, valuing it around a billion dollars. Dorsey decided to take a look. “There’s 10 employees, they only spent $40,000 on R&D. I looked at the product and it was literally a remote control plane and a GoPro. It was so clear it wasn’t a real company. Those types of companies only get pumped up when there’s a lot of froth and excitement and there’s making money and retail’s piling in.” When Dorsey published his report on February 18, 2021, shares traded above $10,000 each; today the stock sits at about $1.

To surface issues, Dorsey combs through public records, company filings, and consumer complaints acquired via FOIA requests. “The one thing I’m uniquely good at is going to state AGs and getting consumer complaints people file against businesses,” he says.

Other theses — the ones for which Dorsey has become more well known — aren’t so much about misconduct as they are about long-term trends. Like Hershey, of which Dorsey wrote in 2023: “Investors view Hershey as America’s dominant chocolate company, a safe bet with over a century of consistent growth, and a richly valued royalty on chocolatey cravings. Those days are over. Today, Hershey faces rapidly growing competition from one of the world’s youngest, most talented, and most influential entrepreneurs: 25-year-old YouTube star Jimmy Donaldson, AKA MrBeast. His new chocolate brand, Feastables, has waged an all-out war against Hershey in retail and on social media — and is winning. The Bear Cave believes Feastables rapid growth will soon take a major bite out of Hershey’s profits.”

MrBeast himself responded to the report, saying, “I wouldn’t recommend shorting a company, seems lame. But I will say these next few years between Feastables and Hershey’s will be interesting once I actually ramp up.” And ramp up he did: in 2024, MrBeast’s Feastables did over $250 million in sales, with profits of about $20 million (1% of Hershey’s profit in the same year).

“It was one of the most mocked reports I ever did,” Dorsey said. “You can just see so many people calling me an idiot.” He stands by the thesis. “My writing about how Mr. Beast's Feastable brand is going to take share from Hershey's over time is not going to impact Hershey's stock from day to day. It's not going to send the stock down 10%. So, that wouldn't work in the activist short model. It does work in the newsletter model, because the thesis is about long-term disruption.”

In the last three years, Hershey has significantly underperformed the S&P 500; its market capitalization today is about $35 billion, versus the $50 billion when Dorsey published his thesis.

A similar Dorsey thesis was Draftkings, which Dorsey thinks stands to be seriously hurt by prediction markets. “The prediction market model is a lot less profitable than the traditional sports book model for the company owner. So I think DraftKings is hugely in trouble.”

One large company where Dorsey wishes he’d held his powder was Palantir. “The worst report I ever did was critical of Palantir. I would say for me that was one where I kind of relied more on what friends in the tech industry were telling me than really understanding it myself.”

“I described it as a black box,” he told me. “but in this case there was a lot of value in the black box. I focused a lot on its deals with SPAC companies that increased its revenues. It was all factually accurate but it kind of focused on an issue that didn't matter at the expense of one that did.”

“Bigger companies can take a lot more time to get right,” Dorsey said. “There are large companies that I've followed for many years now, and I still don't think the timing is right, but one day there will be an article.” On the other hand: “Smaller companies tend to be much shadier and have much more egregious issues than big companies. So I can write a Bear Cave on a small-cap with egregious issues pretty quickly.”

As it happens, Dorsey’s biggest actual stock holding is in a casual pizza restaurant chain. I remembered Dorsey telling me this a few years ago, and thought it was an amusing fact. So, I asked him why: “It's very rare for a $50 million company to have a clean balance sheet, profitable operations, and a very talented CEO. So that's something that just excites me."

The Prediction Market Frontier

A recent interest for Dorsey is prediction markets, where he sees a new frontier for truth. “The product of prediction markets is information,” he says. Unlike media outlets with competing incentives, prediction markets reward accuracy. “There is one and only one incentive, which is being right about the information or the question at hand.” He’s an active participant, and thinks his income from prediction markets may soon exceed income from The Bear Cave, which is itself safely in the seven-figure range. “There’s a huge societal benefit to having good quantitative data on these important questions,” he said, “like whether some legislation is going to be passed. That’s hugely useful to executives trying to plan.”

And soon, he says, “the world expert in certain things isn’t going to be some guy on CNBC. It’s going to be the guy who’s in the middle of nowhere but has predicted the CPI prints perfectly for the last three years. Prediction markets are going to find those people.”

When we flew home from El Salvador towards Houston last summer, our first trip to the country, we could see the mega prison. From the plane window, the Centro de Confinamiento del Terrorismo (CECOT) was unmissable, really—stark, fluorescent, conspicuous. The prison was self-consciously terrifying; it caught me off guard, like you weren't supposed to be able to see a place like that. 

Flying to San Salvador in March from Miami, the crowd was different than last year’s: a lot of spring breakers. The landscape is inky dark. I can see stars from the plane. I can see the occasional town. More strings of light as we approach San Salvador. 

The El Salvador International Airport is forty-five years old. Sparkling and glittering and new when conceived in 1980 mid-civil war between the U.S.-backed, military-dominated government of El Salvador and multiple left-wing guerrilla groups, its opening preempting the promise of a surf town that never came to be. El Salvador remained fraught with turmoil and gang violence even after the 1992 Chapultepec Peace Accords that ended the war. Welcome To Surf City: El Salvador, the sign before customs says in a faded blue and white font. 

Those familiar with El Salvador used to say not to arrive at night here. Even two years ago, they told my boyfriend David not to arrive at night, but we were reassured that now, it’s safe. "It's visibly better every time we come here," David’s colleagues told me at the Palestra Society Conference last summer. The new president is doing a good job. Infrastructure. Safety. We arrive in pitch black. Customs goes quickly. We pass photographed portraits of President Nayib Bukele and the First Lady hung above a chair resembling a throne, where tourists can sit and pose for a photo opportunity assembled for a photo opportunity, and then an outdoor food court reminiscent of a midwestern mall: Papa John’s, Pizza Hut, some teenagers selling soda. We Americans are right at home.

Nayib Bukele was elected president of El Salvador in 2019, running under his own new party, Nuevas Ideas, championing a populist investment in El Salvador’s economic and cultural development. Born to a wealthy Salvadoran family of Palestinian ancestry, he quickly gained popularity and notoriety for his iron-fist policies, seeking safety, law, order in a nation that had long been riddled first by war, and then by gang violence. He was elected again in 2024 with a whopping 84.7% of the vote after he autrocouped the constitutional judges in judiciary court and replaced them entirely with pro-Bukele judges—then revised the constitution to allow for consecutive terms. Tough on crime, and the self proclaimed “ coolest dictator in the world” and “Philosopher-King” in his oft-changing X-bio, Bukele has transformed El Salvador from the Murder Capital of the World to the Safest Country In The Western Hemisphere, bringing homicide rates from over 100 per 100,000 in 2015 to around 1.9 per 100,000 in 2024. Now 1.6% of El Salvador’s population of 6.4 million is currently in prison. And in keeping with the young president’s verve, prioritizing innovation, social media, and cultural capital, El Salvador became the first country in the world to adopt Bitcoin as legal tender in September 2021. 

A rapidly transforming pro-autocracy pro-Bitcoin nation had begun to pique the interest of some foreigners abroad. In certain spheres of cryptocurrency, tech, and esoteric right-leaning internet communities, El Salvador had begun to be imagined as a new frontier. 

The goals of Palestra Society were still somewhat obfuscated to me, but seemed to tow the lines between imperialism, philanthropy, and an almost spiritual faith in Bitcoin. The Palestra Conference we’d arrived for seemed to represent a meeting of a certain type of mind. On Palestra Society’s X page, AI-generated images of Greco-Roman architecture, stone sculpture, flowing water, ancient weapons, and sculpted bodies crafted an aesthetic vision that towed the line between futurism and tradition. “‘Uncounted ruins, cloaked by dense growth, await excavation.’ Yucatán, 1969,” one post read. The conference attendees seemed to view the new El Salvador as both impressive and malleable. Disappointed with how modernity had panned out by-in-large, many of them had sought refuge online. Now, they were seeking this, but perhaps something more; influence, escape, autonomy, here.

A tall man carrying an iPhone-sized vape picks us up at the airport. I know Sitrym from New York: he is one of David’s colleagues. A somewhat nomadic biohacker loosely based in Austin, Sitrym would serve as our guide and host for the trip this time.. "Sitrym assures me that it's a good thing we're running late," David tells me, hoping to stop for a beer of his own. “Sitrym has been at the food court, drinking." 

***

The Palestra Society had first been introduced to me at a party in New York in August 2024 as a “neoclassical gym” in El Salvador. “Do you want to go?” David had asked me. “Go where?” I had asked. “El Salvador,” David had responded, 

David, a computer programmer working in cryptocurrency, had been advised to attend the upcoming Palestra Society Conference by an acquaintance at the New York party. We had received other strange invitations like this before. The year prior, we had been invited to a charter city called Prospera on the island of Roatán, Honduras that operated with economic, legal and regulatory autonomy as a Zone for Employment and Economic Development. (Recently, Prospera has run up against significant legal tension with the government of Honduras). In New York, there were often Praxis parties thrown in downtown bars or lofts; celebrating a still hypothetical and ephemeral city-state with plans for construction in the Mediterranean as the world's first “Internet-Native-Nation.” Many of my boyfriend's colleagues were nomadic, the target audience for this sort of initiative, shuffling from Patagonia to Austin to New York to Costa Rica to here, locating friends and community in social groups largely formed online, and seeking to exit from modern frameworks of society they felt were folding in or failing around them. Though my position in this world had some characteristics of “insider”—girlfriend, writer, plus one, this did not often translate to clarity on the nature of these trips when I attended. Boys club, work trip, cult. I would arrive knowing almost no one. However, there would be a free place to stay. 

My research in August had revealed only what Palestra Society has to say about itself: which was to say, not much. “Palestra Society is a private research organization centered on the principles of natural law and sovereignty,” I wrote on my laptop, in Manhattan. “Membership online advertises subcategories of Community and Content for $499 a year,” I note. “Palestra Society is a futurist movement built on ancient principles.” A 2024 video produced by Palestra from a previous Palestra Society conference titled Age of Light: Dr. Jack Kruse depicted a group of men in white linen, standing mostly barefoot on a foggy hillside in front of a black screen and a glowing circular image, which looked something like a cross between a planet, a wave, an alien, and an eye. Dr Jack Kruse was introducing a draft of a new medical “law” for El Salvador, which would include legal protection against mandatory medical interventions, state regulation of “electromagnetic field exposure,” and the establishment of “light spectrum standards for health.”  Jack Kruse, a neuroscientist and now-influencer had gained popularity and notoriety online as either genius or snake-oil-salesman. He had made his exodus from Florida to El Salvador in 2022, seeking freedom from medical research restrictions, and healing from the sun.  

In practice, the premise of Palestra seemed to be a kind of new frontier society for foreign castaways (mostly male; mostly working remotely in tech) in an up-and-coming nation where the dollar went far, and  Bitcoin was king. In this idealized Palestra, the Bukele regime would both welcome the arrival of wealthy tech-forward foreigners, and share in their visions for an autocratic red-tape-free future. But in the interim, “Palestra” seemed to hover somewhere between a  retreat, think tank, and aesthetic vision of the future. It was unclear to me which of these visions was more real.

When I first met David, I had noticed that many of his colleagues hoped to live forever. “Will you indoctrinate me into a cult?” I asked him, soon after we met. “I’m a programmer,” David told me, our unspoken understanding being that the culty sides of these industries involved ideas and topics he chose to largely ignore. 

Closing my computer after my initial scouting of Palestra,  we flew to El Salvador the following morning for our first trip to the utopia-in-progress.

***

The first time I visited Palestra for a week was in August 2024. The trip was centered around Palestra’s annual conference. We had convened at the airport with a group of men in linen shorts, and then made our way to the sprawling Palestra mansion, an art deco style compound built into the hills of Escalón, one of San Salvador’s wealthiest neighborhoods. At the mansion, we were greeted by a college-aged young man who introduced himself as Palestra founder Michael McClusky’s assistant. The terrace had been blanketed with Red LED Light for the conference. Dr. Jack Kruse, who Palestra had promoted, had been a guest of honor. His bio displayed on the Palestra website read Exiled Neurosurgeon: Health is about Light NOT Food. The crowd was eclectic, if culturally predictable: we met the founder of a new magazine (that was mostly about Tennessee) and a sculptor who told us how he had been cancelled for making art about “force and form.” However, the only evidence of cancellation I was able to find online were the sculptor’s own comments lamenting the whole sordid affair. Better to be cancelled, I guess, than unknown. 

I had left Palestra in August still unsure about the premise of the place. Health center or cultural exodus. City or Airbnb. So when David was invited back in March 2025, to spend the week coding with some colleagues, I decided to accompany him again. 

Arriving again in March 2025, most of the guests from the annual conference have gone home or moved onto other temporary living situations. However, the red light remained, and became a permanent fixture of the house, even as the mansion itself has drifted into a ghost town of sorts, sparsely populated by a few unrelated characters who vaguely aligned themselves with the Palestra “vibe” — drawn to the sun, vaguely right-wing online sentiments, utopic aesthetics of water and light and and ancient architecture. The emptiness of the place now dispels any illusion of Palestra as a growing town in its own right at least in its current state.

While Palestra seemed desolate, things in El Salvador proper were different. Sitrym, driving us from the airport at 80 miles per hour, explained how “the traffic has gone from bad to worse with more cars on the road,” Though, there are not too many cars as far as I can see—we’re out late and, heading into the city, we’re driving against traffic, “Why are there more cars on the road?” I ask Sitrym. Sitrym shrugs. “More prosperity,” he says. People feel safer. 

As we drive up towards San Benito, where wealth rises visibly with altitude, Sitrym is talking about the highway project, where Bukele is trying to expand the number of lanes, bury phone lines, and boost capacity for business and tourism traffic along the coastline. He's talking about the passport initiatives for foreign artists and philosophers, which offers Salvadoran citizenship to foreign investors “highly skilled individuals” in exchange for financial contribution.  Sitrym is talking about the changes being made and the things that still must be done to make El Salvador a better country. “You want to go upstream and fix a lot of the things that are causing air and water pollution,” he says. “Infrastructure is fixable, but it requires a wholesale change in the relationship between state and society, and state and the land that they run." 

"I imagine they just want to import a new class of rich people from elsewhere," event manager Matt is saying. 

"Eh, not really, " Sitrym says. “They're trying to bring in people who can solve big problems." 

There is a sense of synchronicity between Sitrym’s stated goals for Palestra and the interests of El Salvador that surprises me, in contrast to the isolationist spirit of the charter city. He would like to improve the nation’s education, stop deforestation, establish permaculture—to optimize the country rather than to wrest away control of it. The interests seem parallelly libertarian and autocratic—the myth of a place where one can do whatever one wants, under the reign of an enlightened leader. Here, the Bitcoin pioneers found themselves already aligned—here, they’d be working with what they viewed as a budding utopia, rather than confront the bureaucratic red tape and subsequent challenges of building an imagined paradise of one's own within a sovereign nation. 

The conversation turns to the topic of “un-schooling”—a model of education derived from John Holt’s notion of harnessing children’s natural curiosity for learning, rooted in play and practical work. In El Salvador, grade school is only in session for a half-day. Some affiliates of Palestra have been working with a town north of Surf City to launch an Un-School campus to teach the kids drone technology, history, ecology. 

I lay my head on the luggage in the backseat, and I ask Sitrym if I should talk to a SeaSteading expert for my story. The movement, which envisions online communities of like-minded people expanding into autonomous physical territories, has been a hot topic in tech circles these days. The creation and habitation of autonomous territories abroad were particularly appealing for those who viewed Bitcoin as salvation and the regulations that typically accompany citizenship to a sovereign nation as a roadblock to their research. Sitrym takes a sharp turn on the road, and addresses my question about the Seasteader of interest. “You’re talking about Patri Friedman,” he says.

“Patri Friedman,” I note. Milton Friedman’s grandson. Moving away from seasteading in favor of strong, capitalist, sovereign leaders, like in El Salvador. 

“I went to his post-eclipse party and everyone but me and one guy did ‘circling’,” Sitrym is saying. “It’s hypnosis, basically.”

***

We take a few sharp turns into the hills of the wealthy Escalon neighborhood and the foliage is more muted in the darkness but the headlights still illuminate the road’s steep bends, mansion walls traced with compound-like gates, mostly-abandoned guard outposts, palm groves, a roundabout. This mansion, the Palestra House, would look like a compound, if it weren't all so open air. 

I drag my bag inside. I notice tall orange stucco walls and a garage door that could seal the whole place off, but the house residents aren’t closing it these days, preferring to maintain a spirit of openness. I walk through interior thick wooden doors and then see high ceilings, eight bedrooms, a stone wall fountain, and an expanse of terrace built into the hillside. The infinity pool seems to stretch over all of San Salvador, the population density mapped out by pockets of speckled light. Darker in the distance. All hazy over the volcanoes. 

The last time I was here, someone had taken the necessary steps to mark Palestra House on Google Maps, but this landmark status disappeared by the time I returned. So too, has the Palestra Banner in the garage, most of the Lindy Juice (Palestra branded fresh-squeezed orange juice) in the fridge, the packs of Zyn and the Mastic Gum of Gods (tough chewing gum designed to chisel one’s jawline) and the groups of young men on the terrace hitting Geek Bars in bermuda shorts, linen, sunglasses. 

I decamp to the room where we’ll be staying, which is described to me as “the gym.” Eating t cashews while the guys cook steak outside under dim red light, I take stock of  bed, a dresser, a single cinder block weight machine, and one elliptical, and put to rest the theory of Palestra as an ancient revival style fitness community. 

Sitting on the floor, I transcribe the recording of a call with Palestra’s co-founder, Michael McClusky, from a few weeks back. Clusk, as he liked to be called, had left his job as a Stripe programmer f in 2022, moving to El Salvador, and founding Palestra with another man who remained anonymous and who is no longer involved. The project had been part fresh squeezed orange juice company, part social club, and part effort to gain influence and assimilation within the Salvadoran government. Recently, Clusk had been working with the El Salvador Ministry of Culture on archeological initiatives, using laser scanning technology to map archaeological sites and cultural heritage previously hidden beneath dense vegetation. 

“Palestra is a place that is connected to nature, and that is also a model for the type of place we want to see in the world,” Clusk is saying in my headphones. “Palestra is a research center for most of the year, and a place to meet now and then.” He continues: “Palestra is mostly an art project for now.”

***

Time for dinner. Steak, obviously. David is manning the grill, and the fire is dangerously flaring. A skinny long haired programmer is complaining that the cleaning ladies keep throwing out his cup of cigarette butts. Sitrym is telling me about the Randonautica app—you use your mind to set an intention, and the app then gives you coordinates for where to go. The conversation is confusing. “They are looking for a platonic city that does not exist,” I note.

Hot humid air. It’s too hot for steak. A small cluster of programmers and Bitcoiners are seated around the table. “Has there been any heat lightning?” I ask one of the men. He looks at me blankly. “Heat lightning isn't real.” 

David pours me a glass of rum, and then one more. I am eyeing the swelling fire on the grill suspiciously and he is telling me to be less neurotic. David pours me one more glass of rum. “Do you know what we call this?” Sitrym asks me. “What?” I responded. “Unschooling!” 

Last August’s conference had been framed as an investigation into “The State As a Work Of Art” — a reference to the work of Jacob Burckhardt, Renaissance historian and intellectual forefather of Nietzsche. 

“Good morning, San Salvador,” Curtis Yarvin had opened his lecture last summer. “I am so happy to be here in the only monarchy in the Western Hemisphere.” 

The premise of our trip to El Salvador being cryptocurrency, Yarvin’s attendance — monarchist philosopher, father of neoreaction, father of Urbit — had initially surprised me, but the tone of the lectures had quickly revealed themselves as dealing more in culture, idealism, thought experiments, and even zealotry, than in anything suggesting real pragmatism. 

The conference had been held at Teatro Nacional last summer, a stately historical building constructed in 1911 in the city center. The Palestra House — technically the home of a wealthy Salvadoran family leased on an incremental basis —  had been the social hub. I had spent my evenings on the terrace, half eavesdropping. I overheard young men talking about the automobile, the impact of speed, “seeking speed, road technology, and red light therapy. “I just don’t see the average man as something to hold up, spiritually,” the cancelled sculptor I’d spoken with earlier had said, confidently. Last summer, they had been talking about a plan to stimulate the arts in El Salvador by providing studio space and economic incentives for artists who had left the country during El Salvador’s most violent days and then returned. There was a real philanthropic tone to much of the conversation, somewhat in contrast to most of the attendees' dual libertarian instincts and fondness for autocracy. 

“Is Palestra going to be a charter city?” I interjected into one conversation last summer. A college-aged boy in a baseball cap wrinkled his nose. “The problem with a government granting autonomy immune to regime change is that when the regime changes, the new regime just overthrows the regime change law,” he said. I nodded, hesitant to ask about long-term considerations if the current El Salvador regime were to change. Sensing my confusion,  the young man clarified: “the Salvadoran government is pro-business, so the upside of an autonomous zone is low.”

As a plus one, I was able to drift around the house mostly uninterrupted. One morning, while I sipped branded Lindy Juice at dinner and drank beer on a terrace above a bar, sweltering heat, a tall guy in a baseball hat introduced himself to David. “Are you based?” he asked. David hesitated. Before he could respond, the man turned to me. “Do you know what that means?” he asked. “Do you know what that means?” 

***

In my next visit to Palestra  in August, I spent my mornings in traffic, in taxis, headed towards downtown San Salvador. “Downtown was hot gang territory right up until the state of exception,” my ride-share companions told me. “There are visible infrastructure transformations every few months,” according to a Bitcoiner standing on the balcony of the Teatro at the conference. There were No Smoking signs on the first morning, but they were gone by the next. Cigarettes were constant here. The biohacking impulse was one towards true immortality, but these health-obsessed chain smokers were seeking loopholes. 

“People ask, ‘why don’t you just make a more based version of postmodernism?’,” the cancelled sculptor said, during his presentation. “Why would you stand in quicksand? We need to exit.” 

At the conference lunch break, I grabbed a Lindy Juice and stepped outside to walk the city center with a group of other stragglers: Matt from New York and Curtis Yarvin’s young assistant. The city center had been hot and bustling. We’d walked past the National Library of El Salvador—modern, glass, built with aid from China as a “monument of friendship” between the two countries recently erected in November 2023 . We’d circled the National Palace. Stumbled upon advertisements featuring women breastfeeding. “Look at these pro-natalist posters,” Matt beamed. Back at the Teatro, Yarvin’s assistant pointed me towards a barefoot and shirtless middle-aged man whom I recognized as the ‘Bitcoin Doctor’ I’d been told would be attending. “That’s Jack Kruse,” he said. “He has like seven malpractice lawsuits against him in the state of Louisiana. That’s what you could talk about if you wanted to write a hit piece.” I responded: “I don’t want to write a hit piece.”

Kruse himself was emerging as the hot topic of the conference. The utopian slant of the whole thing extended beyond, evidently, ideas of political, social, and cultural alignment and was becoming, instead, an opportunity for Salvation in a more biblical sense. In the Sun and the Heat and the Safest Country In The Western Hemisphere, you could leave behind bureaucracy, taxes, and social control to pursue research, freedom, and utopic visions uninhibitedA later panelist displayed photographs of an old wooden door knob alongside a modern metal creation. Dismissing the metal iteration, he told us that we could leave behind anything ugly, here, too. Bathed in red light and veins pumping with methylene blue, you could leave behind illness. The focus of Palestra was not explicitly on biohacking to the degree of a place like Prospera, an alternative crypto city-state, but I got the sense that many of the attendees were also hoping to cheat death itself. In contrast to the disembodied or ephemeral nature I had previously understood many notions of techno-futurism to take on, the ethos of Palestra Society was strikingly grounded in body and place. Artificial intelligence might threaten to make the human form obsolete, but the bio-hacking-forward-Bitcoiners wanted to immortalize it, make it stronger, and keep it safe forever. In contrast to most new city states, with their relentless emphasis on “building,” the Palestra apostles seemed to view El Salvador as a pre-existing Eden of a kind. There existed a perfect world, right here, heaven on earth, and they had just been smart enough to find it and move. Escape to Utopia, rather than create it. 

On stage, later, Kruse, still barefoot, his lecture not yet begun, was talking to an audience member about what would become of the children of the Salvadoran gang members who had all been locked up. 

“Make El Salvador Great Again,” Kruse ruminated, once his talk began. “We are the misfits. El Salvador is a nation of misfits.” 

“The past and the future meet in the memory of the dead. The sweetest and brightest link in the chain that stretches over the past, binds us to the dead.” – from the diary of George B. Taylor, Virginia, 1863.

The road from Northern Virginia to Gettysburg in the springtime is a parade of white dogwoods, purple serviceberries, and endless miles of grass waving in the wind. Brick homes with proud classical columns and big red barns dot the landscape. The orchards are in full blossom, and it’s hard to imagine that this bucolic, bountiful quarter was once the crucible of the worst violence our nation has ever seen. I was traveling to Gettysburg to tour the battlefield with a group of Navy SEALs, and try to learn from the Civil War’s pivotal battle, the bloodiest three days of our nation’s worst conflict. Our spring visit was 160 years, almost to the day, since the end of that war.

I run a venture capital fund out of Silicon Valley. I live in the young West, far from the old battlefields of the East. I almost never visit cemeteries. My trade is really about the future. But I wanted to visit Gettysburg because of that link in the chain that George Taylor described, and that in learning deeply about a critical moment of the past, I’d have some new knowledge about how we might get past our present disunion.

America is in a discordant moment. Our politics are deeply polarized, even deranged at times, with hatred of the opposing party intensifying all the time. Political violence has been sadly normalized. Talk of anything from impeachment and prosecution to outright war is not uncommon as a response to the crisis of the week, whatever that happens to be. So often it feels like we’re walking a tightrope over a precipice. And at the same time, we try constantly to avoid reckoning with serious crises, like our ballooning national debt. Our cities are not a point of pride for our civilization, but are often embarrassing messes.

In Gettysburg’s quiet fields, the unseriousness of our day stings badly. If there is one seminal lesson from the past, it is that we have an important job to do. All we possess—our freedom, our wealth, our long lives, our ease—was passed on to us, not earned. We have to do justice to these gifts. It is that “great task remaining before us” that Lincoln spoke of when he visited Gettysburg in November 1863.

Now as then, the place of the dead can be made a school for the living.

***

Jocko Willink served in the SEALs for twenty years. His unit, Task Unit Bruiser, is the Iraq War’s most highly decorated Special Operations unit. Together with the “Ready First” Brigade of the U.S. Army’s First Armored Division, Willink commanded operations against insurgents and al-Qaeda in Iraq that helped win the Battle of Ramadi in 2006. Willink and his colleague Leif Babin brought their battle-tested lessons from Ramadi back to Coronado, California — the West Coast SEAL headquarters — where they created a leadership training program for SEALs who would soon be deployed.

Since hanging up their Navy uniforms, Willink and Babin have taken those hard-won lessons to the wider world. They’ve penned a #1 New York Times bestseller, launched a popular podcast, and started a consultancy called Echelon Front. Willink’s most watched clip on YouTube – with 14 million views – is a monologue called “Good” in which he expresses his resolute optimism by saying “good” like a manic mantra in response to any setback. “Didn’t get promoted? Good. More time to get better.”  Some can’t help but chuckle at the masochistic fervor, but for most listeners, it has become a defiant staredown with fate, a way to turn loss into learning. “Didn’t get the job you wanted? Got injured? Sprained your ankle? Got tapped out? Good. Got beat? Good. You learned.”

The first night of our meeting, we gathered in a private dining hall at the 1863 Inn of Gettysburg, located in the heart of the town’s historic district. There were about 35 of us, and our seats were assigned by group. Most of the tables were filled by whole teams from firms who had come to learn how to become better leaders. They tended to be mostly male and from industries like construction and even milk production. My table was the island of the miscellaneous.

Willink has a crew cut and broad shoulders; he is just shy of six feet tall, with hands that could dislocate your shoulders. He is a black belt in Jiu Jitsu, a booming voice, and a soldier’s eyes etched with lines. Willink’s colleague Babin writes in one of their books that “Even among SEALs, Jocko was a big, mean-looking, intimidating guy.” He is clearly not your average leadership guru, having really been to the edge of violence and endured.

He opened the night by talking about the challenges of leadership and the labyrinth of the human heart. “Welcome to Gettysburg, everybody. Y'all traveled a long way.” He scanned the room, taking everyone in.

“I was flying out here, and I was sitting on the plane,” he began. “I have a really chaotic brain sometimes, and I'm thinking…what is leadership? You can look it up in the dictionary, and you’ll see a bunch of very academic and cerebral definitions of what leadership is. I had all these very verbose definitions floating around in my head. But I don't really like verbose definitions very much. And so I broke it down as simply as I possibly could. It's getting people to do things. That's all it is. It's just getting people to do things. So it shouldn’t be that big of a deal, right?”

The room chuckled, half nodding in recognition, half at the glaring obviousness of it.

“I'm glad some of you chuckled at that. Because getting people to do things is so complicated. I think it might be the most complicated thing that there is. And the reason it's so complicated is because human beings are so complicated. Leadership is the ultimate challenge.”

Willink is at his best, his most humorous, and most insightful, I find, when he riffs his way into a monologue. In this case, it was about how getting your children to clean their room or getting a Navy SEAL platoon to secure a hill are all really the same problem. The plan of action is clear, but getting people to act on it is hard.

“You have to do what I tell you to do because I outrank you” doesn’t work in the military, and it doesn’t work with your kids. It doesn’t even work with yourself! We fail to lead ourselves to do the right thing all the time. If all it took were reading the right business book or the right self-help manual, the world would be full of billionaires, sages, and happy families. But it’s not. Because even if the principles of leadership are written down, even if the best piece of advice is a billboard-sized slogan right in your face at the moment you need it, it is how we act our principles in the moment that really matters. Not to mention that we’re lazy, weak willed, and our egos get in the way.

Adding other people to the mix complicates it even more. The same error patterns within teams occur again and again. Principles, rules of thumb, and standards that are supposed to reduce these errors are often simple but not always easy to follow. This became a recurring motif for the next two days. “Simple, but not easy.” The other main theme was ownership (hence the name of Willink’s book, Extreme Ownership). Willink told us the best-performing SEAL units had leaders who accepted responsibility for everything: every mistake, every failure, every shortfall. The worst performers blamed others.

It was a lot to digest, let alone commit to memory, but we would walk the battlefield tomorrow to see where those principles won the day, or where ignoring them left bodies in the earth.

***

In the Spring of 1863, the Confederacy needed a big victory in the East to balance its impending loss in besieged Vicksburg, the last major Confederate stronghold on the Mississippi River. That loss was a big threat. (When Ulysses S. Grant’s siege at Vicksburg was complete on the morning of July 4, it cut the Confederacy in two, and gave the Union control of the Mississippi River, a strategic and logistical boon). But if Robert E. Lee could win a decisive battle in the East, it wouldn’t matter what happened in the West. Lee had decided it was time, in own words, to “assume the aggressive.”

The war had been dragging on for two years. For Lincoln, the disastrous battles of the last year made him seem a failure as commander-in-chief. The Confederate Army under Lee was on a roll. The Union had lost at Fredericksburg, Virginia in December. And in May, it lost again at nearby Chancellorsville. The Union was on its third chief general in less than two years. As a result, the coalition supporting Lincoln was fraying. Republicans who lost their seats in Congress blamed Lincoln. Moderates in border states like Maryland and Kentucky bitterly regretted the Emancipation Proclamation while abolitionists in Massachusetts cursed Lincoln’s hesitation to end slavery outright. In Maryland, Lincoln censored the press, suppressed freedom of speech, and put political dissidents in jail. They are controversial moves to this day that had little to show for the sacrifice.

The Election of 1864 was approaching fast. Support for the war in the border states had plummeted, and the peace platform of the Democratic Party was gaining ground. General Lee thought a Confederate victory in the heart of the Union — Pennsylvania — would convince voters that the war was going on too long, at too high a price. Lincoln would have to come to the negotiating table.

Then there was a simple brute fact. Armies at that time were like locusts, eating anything and everything in their environs. Lee’s army had been stationary for so long that all the farms and orchards in Virginia were stripped bare. His men needed food and there was plenty of it in the North. So on June 3, 1863, Lee broke the calm of the spring and invaded the North.

The three days of July 1, 2, and 3 in 1863 at Gettysburg marked a turning point in the Civil War, the beginning of the end of the Confederacy, and with it, slavery in the United States. The two armies had 57,000 casualties total, with 9,600 dead and the rest wounded or missing. To this day it remains the bloodiest battle in the history of the United States.

***

What can we learn from the past?

Willink and his team were telling us that, at a minimum, history can illuminate the pitfalls, by revealing the common mistakes we repeat and what we could have done differently. But so many academic historians, those guardians of serious opinion, recoil at the thought of “What if?” questions, the idea that things might have been different, if only significant actors had made different, better decisions. E.H. Carr, in What is History?, a classic book from 1961 that has sold hundreds of thousands of copies, captures the disdain of professional historians when he dismisses attempts to ponder “What if?” questions as a mere “parlour game” and a “red herring.”  His colleague E.P. Thompson goes further, calling the whole enterprise “Geschichtswissenschlopff, unhistorical shit.”

History, Carr concludes, “is a record of what people did, not what they failed to do.”

That spirit continues today, slightly altered, in the work of historians like Jill Lepore, a professor at Harvard and a writer for The New Yorker whose book, These Truths: A History of the United States, ascended the bestseller lists amid critical genuflections. There, she exalts the anonymous swell of social movements over the isolated gestures of individuals. In a 2019 Rolling Stone interview to promote the book, she derided vulgar histories fixated on battles or presidential action, stating “Most popular history is either military or presidential and has little sense of the incredible force and political power of social movements and protest movements…”  As if individuals could ever bend the arc of events. (Incidentally, Harvard claims its mission is to educate “citizen-leaders for our society” but doesn’t offer a single undergraduate class in leadership. Maybe Lepore convinced them it didn’t matter.)

And not only academics, but even Tolstoy, in his great novel, War and Peace, was possessed by the idea, quite monomaniacally towards the end of his sprawling masterpiece, that examining decisions and action is a futile pastime. For Tolstoy, the act of decision-making on a battlefield was delusional. Actually, in fact, it was worse than that. He saw a tragic irony in action: the more you strove, the worse you did. So the fatalistic Russian concludes, “Those, however, who tried to understand the general course of things and wanted to take part in it with self-sacrifice and heroism, were the most useless members of society…”

It’s fair to say that we live in a post-heroic age, heirs to the disenchantment of Carr, Lepore, Tolstoy, and other futilitarians and determinists. Our contemporary philosophy of history centers on the idea that history is largely unpredictable and driven by countless chaotic individual actions. We are involuntary instruments of larger forces we cannot comprehend, nor ever dream of leading. “Such is the inevitable fate of all men of action,” Tolstoy writes, “The higher they stand in the human hierarchy, the less free they are.”

But here in the dining room at the Inn, far from the academic chatter, we were examining the role individuals and teams play in shaping the destiny of a battle, a war, and a nation.

There is an old tradition in the U.S. military, going back to the aftermath of the Civil War, where younger commanders were teamed with Civil War veterans to tour a battlefield. It was called a staff ride because they would ride horses, get off at significant locations, stand around, and have discussions over what took place and why certain leaders may have made the decisions that they made. They discussed “What if?” questions to understand how they might improve their performance on the field. The past was a guide to the future. For the next two days, ignoring Harvard historians and Tolstoy, we would be doing the same.

There would be four sections to each stop on the battlefield. The first bit would be led by Colonel Rob Abbott, a retired Marine and licensed Gettysburg tour guide. He’d explain what had happened in the battle up until that point and the overall strategic context. Abbott has a wry sense of humor and would salt his narrative with colorful anecdotes. The second section, led by J.D. Baker, would relate in greater detail the leaders, their personalities, and the key decision they faced in that moment. Baker is also a retired Marine and the former director of the Marines’ scout-sniper program. He is a real cut up who turns it up to eleven in his animated delivery and comedic role playing. The next section would open up into a group discussion, both about how leadership could have made better decisions in this moment, and how we might apply the same principle in our own work. During the last section of a stop, Willink, the guides, and the other SEALs would offer their insights and takeaways in a roundup.

Willink closed out the night by informing us he’d be in the hotel parking lot at 4:30am for a workout of burpees–squats, push-ups, and jumps. We were free to join. It was our choice. Buses would depart for the battlefield at 6:30am.

July 1, 1863

I did not make it to the 4:30am workout. Breakfast at the hotel was a soggy array of waffles, eggs, and sausages. Waiting for a weak toaster to brown my bagel on its third time through, I sipped tepid coffee. Last night, for no apparent reason, I was appointed leader of my table, now called a fire team, which meant that I had to make sure five strangers were always accounted for and on the bus. Every time one of the SEALS asked me for a headcount, I froze a bit, worried I had missed someone.

The first stop on our battlefield review was to McPherson’s Ridge, a small hill half a mile west of Gettysburg from which John Buford, a Union cavalry officer, commanded his forces on the morning of July 1, 1863. The sun was starting to rise as our bus made its way to the ridge, and the statues placed near the roads were dark silhouettes in the early light. We disembarked, walked across a wet field, and formed a half circle around Willink, Abbott, Baker and the others.

“Listen up,” Willink said. We were starting our first discussion.

After peeling off from the battlelines of Virginia, Lee had concealed the movement of his army behind parallel mountain ranges and ridges, running north by northeast, slanting through Virginia, across the Potomac River, through Maryland, and up into Pennsylvania. His warpath was close to the route I had taken driving from Dulles International Airport, the key difference being that Lee hid his army behind the Allegheny Mountains further to the West (and what today takes 90 minutes was a many-day affair on foot).

The Union Army couldn’t see beyond the mountains and was at a complete loss as to Lee’s whereabouts. The Grey Fox had invaded the North, and no one really knew where he was.

Lincoln wanted to crush Lee, of course, but his primary strategic objective was to defend Baltimore, with its vital port, and Washington. The Army of the Potomac, therefore, had to position itself between Lee and the Capital. Adding to the chaos was the fact that Lincoln had sacked General Hooker and replaced him with General Meade, knowing full well that Lee had already invaded and that a battle was imminent. The Army of the Potomac began its march north, but when and where they would meet Lee was only a guess.

A guess, that is, until Union cavalryman John Buford’s brigade spotted and identified three southern infantry corps coming down through the mountain pass west of Gettysburg. After three weeks, Lee had concentrated his forces and had turned east through the mountains at last. Buford was staring down the tip of the spear of a 72,000 man army. Some 50,000 men would be coming down the road in front of him. And to his right, from the northern road, another 20,000 or so Confederates were approaching. Standing on McPherson Ridge, we were looking out from the same vantage point as Buford did that morning of July 1st, 1863.

The previous evening, Buford had rushed a note to General Meade: “We need help now.” Meade and the bulk of the Northern Army were 14 miles away. The closest army corps of a few thousand men was five miles away and led by John Reynolds. It would take time for the army to get to Buford. His staff recalled later they “had never seen him so apprehensive, so uneasy about a situation as he was at this time.”

As the night went on, Buford had a difficult decision to make. Generals on both sides were desperate to fight battles on grounds of their choosing. Terrain and elevation were often decisive factors. Even invading armies preferred a defensive battle, if they could manage it through maneuvers, where they could pick a position on the high ground and wait for an attack. In fact, Meade had devised a plan to try to lure Lee down to Pipe Creek and Parr’s Ridge, and make a stand about 15 miles south of Gettysburg, where he thought the terrain was most to their advantage.

Buford saw that to surrender the high ground just south of Gettysburg—the now famous topography of Little Round Top and Cemetery Ridge—would be a massive strategic blunder. But the odds were not in his favor. His two brigades, just 2,950 men, would have to hold off tens of thousands of Confederates until reinforcements led by Reynolds and then Meade could arrive and secure the heights. He’d have to hold them off for at least three hours in the morning of July 1st.

Buford was a hard man and, by many accounts, arguably the best cavalryman in the whole army. Nevertheless, he had little direct guidance from headquarters. He was only in Gettysburg on a scouting mission. His orders were to “cover and protect the front, and communicate all information of the enemy rapidly and surely.” So what was Buford to do? Retreat and let the Confederates seize Gettysburg?  Or fight and skirmish? His best hope was to delay them. If he couldn't, they’d take the town and claim the high ground from which it would be lethal to dislodge them.

Willink split us into our fire teams to discuss Buford’s decision. If Lee took the high ground, he could flank the Army of the Potomac, and the road to Baltimore and its port would be wide open. (J.D. Baker made this point for us by pointing his whole arm towards Baltimore like a frantic human semaphore, as if to emphasize that Baltimore lay somewhere just beyond the town of Gettysburg.) Buford had to have confidence that his improvised tactical plan would move up from the front lines to the top of the chain of command.

Think about that, one of the SEALs marveled: the top general had a plan, and a subordinate had to have the confidence that he could change the general’s mind to adopt a new course of action.

This, we learned, was one of the first principles of combat for the SEALs, what Willink called decentralized command.  No soul has a full grasp of everything happening across the battlefield; the fog of war and distance blind all. Leaders who are remote, listening to radios, and staring at maps can’t monitor the actions of multiple units in a chaotic environment. Decentralized command pushes choice downwards, to the line’s edge, where fighters can shift tactics with new information.

Reciprocal trust is key. Those on the frontline have to trust that their senior leaders have made good strategic plans. And then those leaders in turn must place great trust in their subordinates, empowering them to take improvised tactical action. For that to work, subordinates need to know why, and understand the mission’s North Star. The opposite is to micromanage without explanation, to dictate every X on the map from on high: a general, hunched over his desk, trying to puppeteer every soldier, every regiment like a marionettist, yanking strings till they snap. It doesn’t work.

“Keep that in mind,” Willink said. “Meade had a good plan. Directly from the president. Protect these areas. And then he goes, okay, I heard from my front line troops that there’s a better option. Can you confirm there’s a better option? Nope. But it seems like a better option, I’m going to trust my guy. Let’s go. How do you think that made these guys here on McPherson Ridge feel? How do you think that made Reynolds feel? The boss listens…how much is that worth?”

As luck would have it, Meade and Reynolds were longtime friends who respected and trusted each other. Both respected Buford by reputation. One staff officer noted, “Buford and Reynolds were soldiers of the same order, and each found in the other just the qualities that were most needed to perfect and complete the task entrusted to them.” Buford sent a note up the chain of command saying he would “entertain” the enemy until Reynolds could reach the battle. The entire battleplan of the Union army shifted at that moment. Now Buford just had to pull off a miracle.

Buford devised a defense in depth, i.e. one with multiple layers. The intelligence Buford had collected suggested that the Confederates were massed in a small town down the road our bus had come in on, the Chambersburg Pike. From the elevated terrain of McPherson’s ridge, we could see the road going to the mountains in the west. Buford scattered his men in arcs from where we were standing — behind posts and rail fences, spaced in intervals like a series of ripples widening out to the west and north.

His men were cavalry, not infantry. They kept their horses a short distance behind their posts; a quarter of the men were stationed in the rear to hold the others’ horses. Buford wanted them to act like infantry in order to fool the southern commanders into thinking that there were more men than there actually were. Their orders were to fire their carbines at the enemy as they approached to vex and slow them down. The order was not to hold their ground. When the Southerners were close enough, a picket of men should peel off, hightail it on horses back to a new post, and delay the Confederate advance further.

Sometime just after 6am, two privates from the 8th Illinois Cavalry, Thomas B. Kelly and James O. Hall, saw a prodigious cloud of dust rising from the road ahead, about three miles down. Something massive was moving their way. They sent word back to a commanding officer. Lieutenant Marcellus E. Jones came to check it out. Eventually they could make out a Confederate flag and the men carrying it in the lead column. Jones asked one of the soldiers for his carbine, which had a range of about 800 yards. He propped it up on a rail fence and aimed at a soldier on horseback near the flag. At about 7:30am, Jones fired the first shot in the Battle of Gettysburg.

The fighting that morning was intense, and the Confederates drove Buford’s line further and further back. But the defense in depth had stalled long enough for Reynolds’ reinforcements to arrive.

The general consensus is that the South won the first day of the battle. Over the rest of the day, their forces would overwhelm the Union Army, push them back, and ultimately seize the town of Gettysburg. Lee himself declared that his opponents “were entirely routed.” When Reynolds arrived on the scene that morning, he had ordered reserves of infantry and artillery to hold Cemetery Hill, just south of the town that Lee now held. Shortly after giving these orders, Reynolds would be shot and killed, a Minié ball right through his neck, but the high ground had been secured.

Below, Lee would soon learn the price of his error.

Little Round Top, July 2nd, 1863

I slept through the 4:30am burpee workout again. With another half-toasted bagel and headcount, we were off on the bus at 6:30am to visit our first two locations in the morning, the Peach Orchard and Little Round Top. The fight for Little Round Top was the hinge in the three day battle, a rock that stood against wave after crashing wave of attack. By defending this otherwise unremarkable hill, a few Northern brigades blocked Lee’s army from flanking the entire line of the Army of the Potomac.

Victory would constrain Lee’s options for the next day. With no way to flank the Union, Lee was left with a thrust straight at the heart of the Union line, Pickett’s doomed charge into the maw of Northern guns firing down from the high ground. Winning the fight for Little Round Top was critical for the Northern victory at Gettysburg. It is the high water mark from which the South would recede forever. But it was almost lost.

We gathered on the highest ridge at Little Round Top, in a semi-circle around Willink, Abbott, Baker, and the other SEALs. It had started to rain. Some pulled out umbrellas. I put on a rain poncho some young technologists had given me, a world away in California. It says “Dyson Sphere Maintenance Crew” across the back. A Dyson Sphere is a sci-fi dream to harness all the power of the sun. I can’t think of a more remote concept from what I was looking down upon.

In the fields below us, in an area called the Devil’s Den, and near it the Wheat Field, by the afternoon of July 2nd, the bullets had cut the wheat down and painted it red. There were 6,000 dead or wounded in the Wheat Field that day and they were packed in so tight soldiers later said you could walk from one end of the field to the other without ever touching the ground. The farmers had fled the area, but their hogs stuck around, and had started eating on the dead and wounded men. Soldiers wrote in their journals that they stayed up all night with swords and bayonets trying to keep the hogs at bay in the dark.

The slaughter on both sides that day was horrific. The action, sharp and at close quarters. Human masses would teem, move, get destroyed, and then crop up again. The most common image used by soldiers in their diaries is of a scythe reaping a harvest of death. Bursting shells, firefights, clouds of flash and thunder, volleys, canisters exploding into a hailstorm of shot, men screaming, sweeping enfilades fired down lines of men, killing five or seven soldiers at a time. It was a mowing.

From where we stood, the elevated ridge lines clearly provided superior shooting positions, rocks for cover, and ravines for exit. Abbott had set the scene for us, and now Baker was reflecting on how it came to this, the storm of violent chaos below and the desperate attempt by the North to protect its exposed side on this hill.

“We are dealing with the repercussions of one individual who decides that he doesn’t like his position on the Union line. He disobeys orders six times. Six times. Paddy O’Rorke, the 20th Maine, Pennsylvania and New York–they’re all struggling on Little Round Top because of one individual,” Baker said. “It only takes one to make sense of where we are.”

That man was a character named Daniel Sickles. Sickles was a notorious figure. A former congressman from New York, and a New York City power broker, Sickles shot and killed his 15-year-old wife’s lover, and was acquitted on the murder charge, making his case the first successful use of the temporary insanity defense in U.S. history.

Sickles commanded the Third Corps, and he didn’t exactly follow orders well. He was not a soldier by training; he had not attended West Point, nor had he served in the military prior to the war. He had earned his position as a general through political influence. At the end of the first day of fighting, thanks to Buford’s and Reynolds’ delaying action, the Union had formed a defensive line on the high ground in the shape of a fish hook. That hook began south of downtown Gettysburg at Culp’s Hill, curved northwest to Cemetery Hill, then turned south, running two miles along Cemetery Ridge all the way to Little Round Top. Meade had positioned Sickles towards the end of Cemetery Ridge near the bottom of the fish hook, just north of Little Round Top.

For reasons unknown, on the morning of July 2nd, Sickles decided to position his men out on a limb, extended far from where he should be in a peach orchard, nearly a mile away from the main defensive line on Cemetery Ridge. Sickles didn’t even notify headquarters he did this. (One major who served with Sickles later said he and Sickles both agreed the peach orchard was more advantageous for meeting the anticipated Southern attack. Sickles claimed his artillery would be better situated on the flatter ground. Another account reported that Sickles thought his men were more vulnerable on the ridge thanks to Buford’s cavalry being relieved and no one replacing them. And yet another account stresses that Sickles was a friend of Joe Hooker, the man General Meade replaced, and that Sickles, a Democrat, resented Lincoln’s choice, and that he despised the cliquish and cautious West Pointers).

At any rate, Sickles ignored Meade’s orders to return to the defensive line on the ridge six times. After the fifth act of insubordination, Meade mounted on his horse and rode out to the peach orchard to deliver the order himself, telling Sickles to return to his position on Cemetery Ridge. But Sickles refused to leave. Meade’s aide had never seen Meade so angry. Sickles’ corps was exposed, isolated, and on flat ground. But by this point, it was already too late. Sickle’s men were soon hit by the first wave of Confederate attack.

Meade then had to order thousands of men out to the fields and down from the high ground to bail Sickles out. These were the men who fought and died in great numbers in the Wheat Field. And because Sickles had moved his corps to the peach orchard, he had left Little Round Top defenseless. Save for a few men waving flags to communicate, the hill was now unprotected, exposed, and inviting an attack.

Willink used this opportunity to raise a difficult problem for leadership. What do you do when the people below you do not follow orders? We broke up into our fire teams, and he had us run a role playing scenario where one of us would play Meade and the other Sickles, trying to convince one another.

The people who played Sickles asked, wasn’t Sickles like Buford? Shouldn’t we see decentralized command as a virtue? It worked the day before. What was different here? Maybe Sickles had discovered some new information about the topography that Meade and the others were unaware of.

Willink jumped in with his perspective during the recap. “There is absolutely going to be risk with decentralized command. And so you, as a leader, have to pay attention,” he offered. Compared to Buford, who communicated clearly and often, Sickles was an unresponsive scoundrel.

Another SEAL suggested that Meade might have taken ownership of the situation earlier, based on Sickles’ already demonstrated recalcitrance. Meade might have had a face to face meeting with Sickles in the Peach Orchard much earlier in the day than he did. He could have better explained to Sickles in the morning that he didn’t have a full understanding of the enemy’s position or movements. Whatever the case may be, it was on Meade to make sure Sickles understood the overall strategy and to train him up. Leadership is a skill, Willink and his colleagues would remind us.

Any principle can be taken too far. In the case of decentralization, it only works if the constituent parts — in this case, men — are aligned in purpose. Buford, Meade, and Reynolds had strong relationships. Sickles was an antagonistic outsider. And as Baker put it, “If you disobey an order, you better win.” Tragically, Sickles did not. Meade’s men paid the price.

A year later Sickles would testify against Meade before a Congressional committee, saying that his actions in the Peach Orchard set the Union up for victory at Gettysburg, and that Meade had been trying to retreat to Pipe Creek. In a darkly funny twist, Sickles was awarded the Medal of Honor in 1897.

Back on the ridge, Baker closed out our session by reading aloud from Elijah Hunt Rhodes’s journal, now a book called All for the Union. Rhodes started the war as a private; by the end, he was the colonel of his regiment. Rhodes and his fellow men marched thirty-four miles non-stop to join the battle at Gettysburg that day. The soldiers, though exhausted, quickly prepared to engage and entered the fray below us on this ridge.

“So as you’re looking over these fields, and you’re trying to imagine what it was like on July 2nd, 1863,” Baker said, before stopping for a moment. “You know, I was with some Marines and they questioned the training and education that I was giving, and they’re like, ‘What are you preparing your men for?’ And I would read them this chapter. This is what I’m preparing for. It’s why you get up at 4:45am in the morning. Because I’ve got to be ready to go this distance, to enter these fields, all for the Union.”

It made me question my decision to skip burpees with Willink at 4:30am.

***

One Alabama brigade started July 2nd on the other side of the mountains to the west in New Guilford. They were awakened at 3 o’clock in the morning and ordered to march thirty miles in 11 hours. As the Alabamians approached the front line, they sent a group to go fetch some water. But Union sharpshooters captured the twenty-two guys carrying all of the canteens. So they didn’t have any water. Then they lost four of their five colonels, one wounded, three down with heat stroke. The rest made it to Little Round Top, and they were going to unleash an assault upon the hill.

The Alabamians hit a wall. They kept probing, looking for a gap, trying to feel out where the utmost left flank of the Union was, because they wanted, in the words of Baker, to flank’em and spank’em.

The force they hit was put in place by a Union colonel named Strong Vincent and the soldiers manning the line were the 20th Maine, led by Joshua L. Chamberlain. The Union was lucky these 500 men were there. Their presence as a defensive line at this moment hung on the most unlikely chain of competent, daring men finding each other and taking action.

Here, again and again, they pushed back Alabama’s attacks; Alabama renewed their attacks. It was a bitter struggle of diminishing numbers, low to no ammunition, and hand to hand combat. On top of that, diarrhea, heatstroke, and hunger. The short distance between the men from Alabama and the men from Maine produced some of the most terrifying carnage, and also the most inspiring heroism of the war.

Our group made its way down from the high ridge, and walked down to the southern back side of Little Round Top, to the point where the 20th Maine held its ground. A slight incline, it was full of bushes, trees, stumps and large boulders, but there was still a clear view to the rising slope where the action took place in the late afternoon of July 2nd.  Abbott set the scene for us.

Meade had mounted his horse, and was on his way to get Sickles’ corps out of the Peach Orchard and repair his defensive line. Along the way, he and one of his staff, a man named Gouverneur Warren, noticed that Little Round Top was undefended. Meade told Warren to check it out and report back to him. He had to go rip into Sickles for not obeying orders. So Warren peeled off. When he got to the ridge, his anxiety deepened. There were no Union soldiers there. Rifle bullets zipped by and tore through the trees around him. If he didn’t get reinforcements quickly, the Confederates would take the hill.

Warren immediately sent handwritten notes off to rustle up some brigades to the hilltop. The unlikely, fragile chain began to form. He gave a note to a teenage soldier-courier, telling him to find a commanding officer named General Sykes. Then Sykes told one of his captains to tell General Barnes, and now that captain was riding around, trying to find General Barnes…but he couldn’t find him in the confusion. Suddenly Strong Vincent of the 20th Maine saw the captain looking lost and hopeless. He asked to see the note. Vincent was supposed to be heading to the Wheat Field, to help save Sickle’s corps. But now he was alerted that Little Round Top was vacant. He had to change his plans.

Taking action, not waiting for permission, Vincent marched his men to the top of the hill, and ordered other brigades to join him. It was the right thing to do, but his action was so quick that he made one mistake that would come back to hurt the brigades that defended Little Round Top. Vincent forgot to tell his logs chief, the man in charge of supplies, to redirect their supplies to Little Round Top, instead of the Wheat Field. The brigades that fought on the hilltop had limited ammunition, limited water, and limited food.

Once on the hilltop, Vincent began to place the regiments in an arc stretched the line all the way around the hill to its rocky western face where we had been. The left most point was the extreme end of the Union line. It would become the point of maximum conflict. To Colonel Joshua L. Chamberlain of the 20th Maine, Vincent said, “I place you here! This is the left of the Union line. You understand? You are to hold this ground at all costs!”

Chamberlain was a professor of rhetoric at Bowdoin College in Maine. Major Ellis Spear, a former student of Chamberlain’s who served under him, described Chamberlain as “a gentleman and a scholar, and although he was also without military knowledge or experience, he was a man of such intelligence and urbanity and kindliness of feeling that he exerted a useful influence.”

Bowdoin College thought Chamberlain was departing to Europe to conduct research, on paid leave no less. Teachers were exempt from mandatory military service. Instead, Chamberlain snuck off and joined the Union Army. And now he was holding the line, fending off hit after hit from a ferocious Southern attack. After hours of the onslaught, his brigade had nearly run out of ammunition. His line had thinned. Wounded men lay around screaming in agony. The dead scattered everywhere.   

“I want to give you a sense of Chamberlain, in his memories, of what he did here at Little Round Top,” Baker said to us. “Chamberlain is right here where you are standing. He’s got this young officer who comes up to him and says, ‘Something’s happening down there out in front that we can’t see.’ So Chamberlain runs back, and he mounts a great rock.”

Baker acted out Chamberlain’s action. He moved from where we were standing in our semi-circle back to a rock no one had really noticed before.

“This is the rock that Chamberlain mounted. So Chamberlain rises up on the rock and sees something going on down there. He’s elevated himself, and he’s up here.” Then Baker began to read aloud from Chamberlain’s war memoir, Bayonet! Forward, a section that is his report written to his superiors three days after the battle:

“Mounting a large rock, I was able to see a considerable body of the enemy moving by the flank  in rear of their line engaged, and passing from the direction of the foot of Great Round Top through the valley toward the front of my left…We opened a brisk fire at close range, which was so sudden and effective that they soon fell back among the rocks and low trees in the valley, only to burst forth again with a shout, and rapidly advanced, firing as they came. They pushed up to within a dozen yards of us before the terrible effectiveness of our fire compelled them to break and take shelter. They renewed the assault on our whole front, and for an hour the fighting was severe. Squads of the enemy broke through our line in several places, and the fight was literally hand-to-hand. The edge of the fight rolled forward and backward like a wave. The dead and wounded were now in front and then in our rear. Forced from our position, we desperately recovered it, and pushed the enemy down to the foot of the slope.”

Baker continued to read out loud: “It did not seem possible to withstand another shock like this now coming on. Our loss had been severe. One-half of my left wing had fallen and a third of my regiment lay just behind us dead or badly wounded. At this moment my anxiety was increased by a great roar of musketry in my rear, on the farther or northerly slope of Little Round Top, apparently on the flank of the regular brigade, which was in support of Hazlett’s battery on the crest behind us. The bullets from this attack struck into my left rear, and I feared the enemy might have nearly surrounded the Little Round Top, and only a desperate chance was left for us. My ammunition was soon exhausted. My men were firing their last shot…It was imperative to strike before we were struck by this overwhelming force in a hand-to-hand fight, which we could not probably have withstood or survived. At that crisis, I ordered the bayonet. The word was enough. It ran like fire along the line from man to man, and rose into a shout, with which they sprang forward upon the enemy, now not thirty yards away. The effect was surprising; many of the enemy’s first line threw down their arms and surrendered. An officer fired his pistol at my head with one hand while he handed me his sword with the other.”

The bayonet charge of the 20th Maine was the turning point of the battle and possibly the war. That could be a claim taken too far, but I am prepared to defend it. Academic historians in our anti-heroic age tend to downplay the importance of Little Round Top, largely I think because in the decades after Gettysburg, it took on the stature of a legend. It was just too heroic to be true. It was too self-serving.

Now granted, it is true that Warren, Vincent, and others deserve their fair share of credit; and that Chamberlain was his own best promoter in the decades that followed (he was a gifted writer). And it also must be said that there were other critical fights taking place that day, first in the center of the Union’s line and then at the top of the fish hook at Culp’s Hill. And there was another day of battle ahead. I can grant all of this, but the counterfactuals these historians present to make the case just aren’t convincing.

It’s hard to imagine a scenario where the Confederates take Little Round Top, roll up the Union left flank, cut them off from their supply chain, and flank the rest of the defensive line, and yet somehow, we are then to believe this wouldn’t matter to the outcome of the battle as a whole. In his Gettysburg: The Last Invasion, Princeton professor Allen C. Guelzo suggests that reinforcing Little Round Top may have weakened Sickles position in the Peach Orchard and thus caused the Union more trouble, which one look at the topography could have refuted. The best the Union could do there was limit their losses. Even less credible are Southern officers who later wrote they had already decided to retreat before Chamberlain and his men charged. There is no getting around the fact that screaming Yankees on the run had put fear into their hearts. Those Yankees chased the Confederates all the way beyond the next hill over. Some joked they could have kept going all the way to Richmond.

“If we had been five minutes later [to Little Round Top],” Charles Salter, a private from a Michigan regiment posted near Chamberlain, wrote, “the enemy would have gained the ridge we were on, and turned our left flank, and it would have been very hard to drive them from it.” This was a private with no fame to gain, not a politician with an election to win or a proud loser looking to save face.

After Baker stepped down from the rock, I had to stand on it for myself, to see what Chamberlain saw. There is a presence, I must admit, a standing-thereness to the rock, a heightened silence, an edge to something for which there are no words. Chamberlain had started the day with 500 men under his command. At the end of the day, only 198 were left. Some years later, a group of the veteran Confederate soldiers who fought that day wanted to place a monument on Little Round Top to honor the Alabama regiments. They sent a note to Chamberlain, in effect asking for his blessing. And Chamberlain’s response was, “Why would they want to put a monument on Little Round Top? They never got there.

November 19, 1863

The significance of Gettysburg was not clear right away to the warring parties, nor even to Lincoln himself. He was distraught that Lee had gathered his broken forces and had escaped back to Virginia. Gettysburg was Lee’s Waterloo, and Meade had let Napoleon go. Lincoln felt it should have meant the end of the entire war. He wrote a letter to Meade excoriating him, but in the end, decided not to send it. Instead, he left it in his desk drawer for history to discover.

Three weeks after the battle, the dead were still scattered throughout the hills and fields. David Wills, a local lawyer and leading light of Gettysburg, wrote to the governor of Pennsylvania that “In many instances arms and legs and sometimes heads protrude and my attention has been directed to several places where the hogs were actually rooting out the bodies and devouring them.”

There were more dead than there were inhabitants of Gettysburg. It was an open question what to do with all of the limbs, corpses, and other remains. The governor appointed Wills as his agent to establish a national cemetery for the Union dead. Confederate bodies were thrown in unmarked mass graves and moved to the South years later. Wills secured land for the cemetery, and arranged a dedication ceremony.

Wills invited a man who was considered the greatest public speaker of the time, Edward Everett, the former president of Harvard University, to provide an “oration.” Originally, Wills wanted the event to take place in October, before the frost came, which would make it difficult to inter the bodies, but Everett objected , saying that he would need more time to compose a speech worthy of the occasion. So the event was pushed to November 19th, 1863.

Everett composed a flowery speech that was two hours long, an epic which he recited from memory. One Gettysburg resident said she thought Everett’s speech would never end. In the words of Garry Wills, a historian and scholar of rhetoric, Everett’s speech, as ornate and polished as it was, “was made obsolete within a half-hour of the time it was spoken.” When Everett had finished, music played, and then President Lincoln stood up and spoke fewer than three-hundred words in around three minutes. He had finished writing and revising them the night before.

It is the nature of poetry to endure in the mind. Joseph Gilbert, a journalist for the Associated Press, was there to report on Lincoln’s speech. Lincoln stepped to the front of the platform, his hands clasped, and began to speak. Gilbert was taking notes, but as he listened to the words, he became mesmerized, and his pen came to a full stop. “Fascinated by his intense earnestness and depth of feeling, I unconsciously stopped taking notes and looked up at him,” he wrote.

Plato says that the two aims of a funeral speech are to extol the dead and to exhort the living — “Laud the dead and lead the survivors.” Lincoln did both, but he didn’t just instruct us to strive forward. He challenged the crowd to remember the meaning of the past. And as with most challenges, there is a right way of responding. The future will judge whether we responded well. Lincoln dared the crowd then, and dares us now to ask whether we have made ourselves worthy of the sacrifices of the past.

A couple of months before my Gettysburg visit, I met a French sculptor named Missor who told me he feared we were returning to a pagan sense of time, when time was thought to circle back on itself, like a millstone turning in place. He told me we needed to remind people–even shock them–through sculpture that history could be linear. History could rise to greater heights, potentially to the stars, he said, but we had to see time as something we carve, like a stone, to pass on to those who come after us.

“Non! Mankind’s ultimate act is not to forget,” he said, as only a fiery Gallic artist could. “He only becomes a singular being if he remembers.” He then elaborated the point: “Mon ami, in ‘Les Misérables,’ the hero gives his life for a promise he made to someone who is dead. I find it so beautiful, and also, so civilizational… to keep a promise to someone who is not here anymore, who cannot check on our promise. For me, that is the definition of humanity, what separates us from the beasts. That is what I mean by memory.”

Beneath the statues in Gettysburg, I remembered my friend, and the fear that the past would indefinitely repeat itself. Lincoln forcefully rejected this, and said that a future dedicated to the propositions of freedom could be different. The statues reject this, too, I think.

“Some people say these monuments are over the top, in your face, we won the war monuments,” Abbott told us at one point. “But that’s not what these are at all. These monuments are here to heal the veterans. Gettysburg truly is a place of healing. Come out here someday and be quiet and just pay attention, and you’ll see old guys coming up and touching the monuments. Because our veterans today, we don’t have these monuments. Our veterans today, a lot of them will come here to remember the people that they lost.”

To close our time together, Willink led us to a hill just outside the ground of the cemetery. “We get to walk around out here, we get to talk about these leadership decisions, and learn those lessons, and then you walk through that graveyard. And that graveyard represents a fraction — a fraction — of the people who sacrificed their lives for the Union.”

He pulled out a piece of paper and read aloud an entry from an unknown soldier’s journal depicting the horrors of the Gettysburg battlefield: rotting bodies, “some with faces bloated and blackened beyond recognition, some with glassy eyes staring up at the summer sun and others with faces downward and clenched hands filled with grass or earth which told of the agony of their last moments.”

Willink reflected on what it feels like to see this horror in war and then to return to your home. “If a human being can do this to another human being, then we must be doomed. It’s so easy to go down that path. But if we can detach, if we can take a step back, something else reveals itself on the battlefield. The will to live. The will to fight for freedom. The will to fight for something bigger than ourselves.”

Willink then recited the Gettysburg Address. There wasn’t a dry eye in our group.

It's 2026, the 250th anniversary of the signing of the Declaration of Independence, so it’s a natural time to reflect. What is the state of our union? Are we living up to the expectations our founders set out for us?

If you only listened to the loudest voices in the legacy media, you'd think America was on the verge of collapse. Rolling Stone asked if we are "witnessing the fall of the American Empire." Francis Ford Coppola, one of the great directors of the twentieth century, was quoted in The Guardian as saying we are at the point where "we might lose our republic." In the wake of Charlie Kirk's assassination, Vox asked if America is "on the brink of civil war."

And the tone of the national conversation on social media is even worse. The subject of the chatter changes daily, but the gloomy tone persists: the terminally online, wildly overresponsible for what information Americans ingest, are not quite sure what will ultimately doom America, but the leading contenders include political violence, tribalism, insurrection, cancel culture, systemic inequality, climate change, drug overdoses, border control, gun violence, crashing fertility rates, the "loneliness epidemic," Chinese competition, restrictions on speech, wokeism, community collapse, rising housing costs, authoritarianism, socialism, and the national debt.

But if I had that conversation with a Founding Father, I'd start with this:

The Union survived, and there's an American flag on the moon.

NASA / Neil A. Armstrong - Apollo 11 Image Library

The nation fought a brutal civil war over the issue of slavery. Eleven of the 34 states seceded; 2% of all Americans died; the union's national debt rose 3300% in just five years; the President suspended habeas corpus. Slavery was abolished. And the Union survived.

The world fought a series of brutal, global wars. In both, Americans were drafted to fight across oceans. In the worst, a German dictatorship conquered continental Europe and left Britain alone, besieged, the last thing standing between the world and totalitarian dictatorship. America, by then the world's foremost industrial superpower, intervened and defeated the Germans. And the Union survived.

A massive social revolution extended America’s ideals to every American, whether white or black, male or female; America led the world through wave after wave of technological revolution that dramatically changed the way people live, work, and find meaning. And the Union survived.

There are 340 million Americans that live in 50 states coast-to-coast. Only 0.5% of infants die before age one, and the average American lives to be 78 years old. Less than 2% of Americans work in agriculture, compared to 90% in 1776, and the average American consumes 15 times more energy daily. Americans can now talk to almost any human on Earth, instantly, using a device that fits in their pocket. Flying machines — planes — can carry you anywhere on Earth in less than 24 hours. And even more powerful flying machines — rockets — can escape the surface of Earth and have brought Americans to the lunar surface. 

I do not think it would be difficult to explain to a Founder how incredible it is that the American union survived for 250 years, and that life in 2026 is awesome. I wrote a 350-page presentation explaining why, but here I want to share ten big highlights.

Judging from the discourse, the inverse — convincing you that things are great, today, actually — might be a more challenging sell. We are David Foster Wallace's fish in water: "There are these two young fish swimming along and they happen to meet an older fish swimming the other way, who nods at them and says ‘Morning, boys. How's the water?’ And the two young fish swim on for a bit, and then eventually one of them looks over at the other and goes ‘What the hell is water?’”

In the remainder of these pages, it is my goal to remind you that Americans today are floating in absolutely pristine, carefully-engineered, painstakingly-filtered, brilliantly blue water. The last 250 years of technological progress are often invisible, yet make us safer, healthier, more comfortable, more connected, and more productive — and a quick look back at what life looked like in 1776 will help bring that truth into focus.

Water

In 1776, water was scarce. There was no household running water; every drop had to be physically hauled from the nearest freshwater source. The average household consumed just a few gallons of cold, unpressurized water each day, and adults and children alike would at times drink home-brewed beer, hard cider, or whiskey instead, out of fear of disease. 

Those fears were warranted. Water sources were often brackish, muddy, or polluted, and in a pre-filtration, pre-germ-theory world, water was often a vector for diseases like dysentery or typhoid.

The “Conduit”, Boston’s 1652 supply, the first Water Works in the U.S.

Food

The food supply in Revolutionary America was more reliable by comparison. Producing food required a massive human burden, of course — nine out of every ten colonial Americans worked in agriculture. 

1853 Currier & Ives print of an American farmer plowing his field.

With no refrigeration, de-brining salted meats required hours of soaking in water. But most Americans were well-fed, at least during the summer. In the winter, fresh produce was scarce given the impossibility of long-distance food transportation, making scurvy a legitimate threat.

Shelter

Winter created far greater problems than scurvy in 1776, however, as insulation was effectively unknown. Houses were small, drafty, and entirely reliant on inefficient, ashy, smoky, smelly hearths for heating. You could only really get warm around the fire, or in bed with a bladder of hot water or a bed of coals; Thomas Jefferson famously couldn't write to complain about the cold in his estate, Monticello, because his inkwell had frozen solid.

Lower Swedish Cabin, Drexel Hill, Upper Darby Township, Pennsylvania, built ca. 1640–1650, may be one of the oldest log cabins in the United States.

The winter’s cold and the daily necessities of cooking and water purification made firewood a truly massive industry. Cutting, hauling, storing, and constantly managing firewood meant that it may have been  more than one quarter of the 1776 economy. (For comparison, 28% of America’s GDP in 2026 is roughly $8.7 trillion.) Nearly every colonial household consumed about forty cords of wood — 5,120 cubic feet, or around 70 tons, of stacked fire wood — annually.

Sights and Smells

With no incandescent lighting, nighttime was oppressively dark. You could read fireside or by candlelight, but both were expensive. George Washington famously spent $15,000 on candles each year (in today’s dollars).

Washington's preferred candles — made from spermaceti (a waxy oil from the heads of sperm whales) or beeswax — were mostly odorless, but expensive. A single beeswax candle cost roughly a laborer's daily wage. As a result, most people could only afford to use tallow candles made from animal fat, which were smoky and putrid. Or they simply embraced the darkness, and looked up at a sky completely alive with stars. (This is one part of life in 1776 that I envy.)  

Even during the day, many people couldn't see clearly. Eyeglasses were expensive, yet primitive: Benjamin Franklin wouldn't invent bifocals until 1784. Not that you’d always want to see the sights of 1776 America. Colonial cities were, in a word, disgusting.

Engraving of Philadelphia from 1799. Second Street North from Market St. and Christ Church, created by William Birch

Indoor plumbing was nonexistent, so if the urge arose in the middle of the night, a colonist might simply defecate in a chamber pot and toss it out the window to be retrieved by a "night soil man." Streets that didn't smell of human excrement certainly smelled of horse manure (“emissions” meant something different for 18th century transportation) or rotting trash on the streets. Colonists at the time rarely took a full bath, and to survive in such an olfactory environment many colonists adopted the "nosegay," a small, fragrant bouquet worn to ward off offensive smells.

Medicine

In such a sanitation-free environment, diseases rampaged through the colonies with little resistance. Doctors were still in the "balancing humors" stage of medicine, hadn't yet discovered anesthesia, and did little to keep people alive: 40% of all children died before age five, and those who made it to age five could only expect to live to 60 years old.

George Washington died at age 67, likely of epiglottis, a bacterial infection of the throat that is survivable today.

In 1776, many now-treatable ailments were fatal. You could die of pneumonia, a burst appendix, sepsis from a minor cut, tetanus, the flu, or a simple fever. You could die of dysentery or scurvy. You could die of any number of now-vaccinated or eradicated diseases: smallpox, typhoid, malaria, cholera, tuberculosis, diphtheria, whooping cough, or yellow fever. The world was casually, routinely lethal. 

Comfort

Making a new shirt in 1776 took hundreds of hours of manual labor. Flax seeds needed to be grown, harvested, retted, broken, scutched, heckled, spun, and woven — so most people only owned one good set of clothes. Clothing was so valuable that some colonists listed it in their wills.

A New England kitchen, engraving in A Brief History of the United States (1885)

The world was not comfortable, and cosmetic improvements were unnecessary and therefore neglected. Dental hygiene, for instance, was uncommon. Most adults were missing teeth, and only the rich — including, famously, George Washington — could afford to replace them.

Energy and Work

The vast majority of "schooling" was just one-on-one apprenticeship to learn a trade, often starting as early as age 13. 

As noted, 90% of the population worked in agriculture, and farming was manual labor powered by humans and horses. A human can generate about 75 watts of work; a horse can generate 750. For comparison, a single gallon of gasoline can generate 33,000 watt-hours of energy.

Knowledge work like governance and trading were similarly inefficient, with low literacy rates, no standard timekeeping method, and exceptionally slow communications channels. News traveled at the speed of a horse — whether "the British are coming" or the news of the signing of the Declaration of Independence itself, which took 29 days to travel the 550 miles from Philadelphia to Charleston, South Carolina.

Paul Revere's ride, Illustration. 

Travel

It was rare for a colonist ever to travel more than a few dozen miles from home. A longer trip, say from Boston to New York, might take a week on horseback or a “miraculous” five days on "The Flying Machine" stagecoach. 

Cordrey, John; The London to Birmingham Stage Coach; Science Museum, London

Crossing the ocean was nearly unthinkable. Of course, colonists had made the treacherous, ten-week voyage to reach the American colonies — but most would make such a risky trip once in their lifetimes.

Flash forward to the modern day, and we are surrounded by invisible miracles.

I am writing this piece in the middle of winter, in a fully lit, 72-degree room at 8:38pm. Earlier today, I drove my electric car on a paved road to pick up a week's worth of groceries. We ate fresh vegetables for dinner (in winter!), cooked indoors using filtered, on-demand water on a natural gas flame, prepared by my mother-in-law who is visiting from her home 2,000 miles away in Georgia. My two children, born safely at clean hospitals, are sleeping comfortably in soft beds and warm clothing. My wife is talking to her friend who lives in Tennessee. I am watching a basketball game happening in Seattle on the right half of my screen, while typing this on the left. I'm wearing a shirt gifted to me by my employer, one of perhaps a dozen t-shirts I own. The room smells like vanilla thanks to a $5.43 candle. The candle is made of scented wax, not spermaceti.

America in 2026 is not perfect, but we have achieved an incredibly high standard of living. Of course, we still face profound challenges, but we have shown that material progress is eminently possible. That’s why I work to build and fund technology companies that promote the national interest: I am a proactive optimist who believes in the American Dream. I do not believe things automatically will get better, but I know that things can get better if we make them so.

If you brought a Founding Father to modern-day America, it would take a while to explain to them how far technology has come over the past 250 years. (Or slightly less time, if you show them my report, More Perfect.) But once he understood the everyday miracles of our technologically-infused lives, it may be even harder for him to understand the pessimism of the national conversation.

At some point, it might be necessary to explain that the term "Founder" is used much more liberally in 2026 than in previous generations, especially in one region of Northern California, but those are minor details.