从零打造美国铀供应链的公司

The US did first-generation technology to win World War II and stay dominant in the Cold War. We were over 90% of supply globally. It was nicknamed “Megatons to Megawatts.” We imported Russian warheads, blended them down, and ran our reactors on it. Congress passed a ban on imports of Russian uranium. That ban goes into effect January 1, 2028, so our first mission is to produce HALEU before the end of the decade for all these reactors.

Scott Nolan leads General Matter, an advanced uranium enrichment company. His ultimate goal is to slash energy costs so dramatically that entirely new capabilities become cheap and routine.

Cheers.

Cheers.

General Matter. What are you guys trying to do?

1. The Fuel Supply Bottleneck

We are enriching uranium. When we started in mid-2023, not many people were talking about uranium or nuclear energy. I’d been doing investing at Founders Fund for 10 years and investing in some of the reactor companies. I asked them, “What’s the hardest thing you have to do?” I thought it was licensing, obviously. They said, “No, it wasn’t licensing. It’s actually getting the fuel. We have to get all of our fuel from Russia. It’s very hard to import.”

So I said, “Why don’t the US companies that make all the fuel for the existing fleet of reactors—94 reactors on the grid, 20% of US power—make the fuel that you need?” They said, “What US companies?”

This is both new nuclear companies, like startups, and existing nuclear plants in the US. Both require your enriched uranium, and both have to source it from Russian companies?

Russian or European.

Okay. But non-US companies.

Non-US. Let’s run through the big picture on where fuel comes from. Reactors, like any engine or anything producing power, consume something: nuclear fuel. Nuclear fuel is made in five steps. First, you get uranium out of the ground. Then you turn it into a gas and separate the gas to enrich it, meaning you get more density of U-235, the fissile material. Then you turn it back into a solid and make your fuel pellets. Five steps. All reactors in the US run on this stack for making nuclear fuel.

The more modern reactors—SMRs and advanced reactors—will use fuel enriched to a higher level, just under 20%, around 19.75%. Most of the reactors on the grid today run at 3–5%. The big difference that unlocks is that, with low-enriched uranium at 3–5%, you generally want a much bigger reactor vessel, a core. If you go up to 19%, 19.75%, or 20%, it unlocks a much smaller core that lets you build a factory-built reactor.

A lot of what you’re hearing about with advanced reactors is a combination of safety, with enhanced passive safety that doesn’t exist in the existing fleet, and a form factor that could be built modularly in a factory, bringing the cost down.

When people talk about small modular reactors, the modularity means you can mass-produce them in a factory rather than needing to build them custom on a site, and you have a bunch of shipping containers that you hook up.

Exactly, more or less. We’re trying to minimize the construction aspect of the project.

Yes. Diablo Canyon—where do they get their enriched uranium?

From existing suppliers. That’s a combination of Russia and Europe. In the US, we still import about a quarter from Russia. That’s going to keep going until 2028. The rest comes from European providers: France and a European consortium called URENCO.

Does the US mine uranium?

We do, in small quantities.

The problem you’re trying to address is that uranium is mined in the US or Canada, then shipped across an ocean to be enriched, and ultimately shipped back as fuel.

More or less, yeah. Most of US supply comes from Kazakhstan and Canada, and where it’s turned into a gas varies.

Got it. I presume if you talk to a nuclear startup that wants to get uranium enriched to 19%, they agree with you on the geopolitics of this. It would be valuable to onshore this, but they probably have more acute problems you’re solving beyond the fact that it would be good to reshore the supply chain.

For them, it’s mostly about availability. As they think about their very first deployments and then scaling up in the early 2030s to tens of deployments, it’s all about where they’re going to get the enriched uranium: 19.75% enriched uranium, also called HALEU, or high-assay low-enriched uranium. That’s all about availability. That was the first problem I heard of: only Russia makes this commercially today. How are we going to import it?

In mid-2024, Congress passed a ban on imports of Russian uranium. That ban goes into effect January 1, 2028. Right now, they’re still importing under a waiver, but on January 1, 2028, that ends. So all these companies are wondering, “Who’s going to produce HALEU?” Our first mission as a provider is to produce HALEU before the end of the decade for all these reactors. Part two is bringing the cost down. That’ll be the other big lever we work on.

How is this regulated? Can you give us the map of enriched uranium? Where does weapons-grade uranium start to happen? And who do you have to work with as a regulator for all this?

If you’re doing commercial sales of nuclear power or fuel, that’s going to go through the NRC, the Nuclear Regulatory Commission. In the US, that’s your regulator. If you’re selling to the government, there are potential Department of Energy paths. If you’re an R&D reactor, you could go through a DOE path, but it’s primarily the NRC for commercial operations.

How have they been to deal with?

Great. Like I said, I think the investment community, the startup community, and the whole tech community have historically just heard, “The NRC is really difficult,” or something.

The NRC has a bad brand.

Yeah. The blame is often placed on the NRC. I actually don’t think it’s deserved. They’ve been great to work with. Ultimately, they care about two things: safety and security. There’s an environmental consideration as well. All their licensing focuses on those things.

That was the answer we got from the SMR company I knew. We asked them, “Hey, how hard is it to get the NRC licensing?” They said, “It’s pretty straightforward. You do the safety analysis, you do the security analysis, you show low environmental impact, and it’s a straightforward process. It’s a lot of paperwork and analysis, but it’s doable.”

Can you give us a bit of a map of nuclear reactor designs? There are the traditional designs installed in plants operating today, and then you were talking about these newer startups pursuing SMR-type approaches. There are a bunch of different technical approaches there. Can you just ground us in what everyone’s trying to do? Presumably, some don’t use uranium as a fuel, so they’re outside the scope for you but may be relevant to this discussion.

2. How Reactor Designs Differ

Most of the US grid, virtually all of it, runs on light water reactors. These are reactors with uranium enriched to low-enriched uranium, or LEU, levels of 3–5%. That goes into the core. Your moderator and your coolant are water, just regular water. That’s most reactors on the grid. You end up with a pressurized steam line and a large dome covering the reactor. That’s part of the large construction project that leads to billions in cost per reactor.

Is the large dome covering the—

That’s a containment dome.

I see.

In the case of an accident or loss of cooling, that’s meant to contain any radiological material.

I see.

The reactor is not actually that large. It’s a much larger dome covering the reactor.

Why does the containment dome need to be so big?

Mostly for the most extreme accident scenarios. It’s about making sure the thickness and internal volumes are there.

I see. It’s small on the inside; it’s just very thick.

Smaller.

That’s the traditional light water reactor design. How about the startups?

Nearly all of them are focused on going modular, bringing the size and core size down. Let’s go to stranded demand centers, serve those demand centers, and go containerized or even smaller. You’re going to be using HALEU in that very small format. Again, that’s uranium enriched to 19.75%, with a smaller core and smaller format. Now we’re talking about fuel formats, but they often go hand in hand with the reactor design.

Why do those two things go together?

Which pieces?

The smaller core and highly enriched uranium.

To unlock a smaller core design that could be built in a factory, you want your fuel to be higher octane, so to speak, with a higher enrichment level. The reason you care about your fuel being at a higher enrichment level is that, at some point, as it’s consumed—as it reacts—that enrichment level will drop. At some point, you can think of it as the power going out, the reactor fizzling out.

If you have a stationary 1-gigawatt reactor, you’re running uranium enriched to 3–5%. You’re probably refueling every 18 or 24 months. That’s acceptable. You have a very big reactor, so your core will stay critical for quite a while, and refueling every 18 or 24 months isn’t such a big deal.

If you have a very small core, think of it as needing higher-octane fuel. As it burns up, it will pass through acceptable power production and down to not even being critical more quickly.

Some of these reactors also want to have 5-year or 10-year refueling cycles because they’re somewhere remote, not somewhere central. It’s a combination: you want to unlock a smaller core that will remain critical for a longer period.

That makes sense. Can you talk about the passive safety advantages of highly enriched uranium that you mentioned?

Not highly, but HALEU.

Sorry. Yeah.

The enrichment level itself does not lead to safety advantages, but reactor formats do. Some of these reactors are passively safe. A classically well-known version of this was molten salt reactors. The molten salt has thermal properties that cause it to self-regulate. If somehow it were overheated, you could put a plug in the bottom of the reservoir. The plug would melt, everything would drain out, and criticality would stop. That’s the idea of a molten salt reactor.

You have other ones, like helium-cooled reactors, where the helium can’t be activated; it cannot become radioactive. So even if you had a coolant release, you’re not releasing any radioactive material. In general, these newer designs have either passive safety features or modern architectures around active safety.

You can design some self-correcting mechanism into the reactor, like a fuse plug type of thing, where the act of starting to run away from you causes the safety mechanisms to kick in.

Right. Well stated.

The AI scale-up and power-up has been powered significantly by turbines and other forms of electricity that you can bring online very quickly, generators in some cases. Has it affected your market at all, or really the market for your customers? Or are the timelines for nuclear power too long?

3. Nuclear Power and AI Demand

I think no, and no. It has not impacted us specifically. A lot of the topics in nuclear today are all about SMRs and advanced nuclear. That’s an important segment. I think it’s going to grow tremendously over the next decade, but it’s still a small market today. As an enricher, most of the contracting volume that we do is with the existing fleet. It’s for low-enriched uranium, with the utilities.

As people think about how they’re powering data centers in the next couple of years, we think some of that will be SMRs as they contract for them. But even if it’s all natural gas, that’s fine, because in the next few years, most of our demand is from existing fleet utility customers.

Now, where does that go in 5 years? I think there will be more and more push to move off natural gas and move to true baseload, and I think that’ll be nuclear. Then the really big question is: Does that happen through more gigawatt-scale reactors, or does it happen through more SMRs, and what’s that balance? If you take a big step back and ask, “What type of energy do we want to power these data centers?” I think there’s going to be a lot of pressure for it to be safe, clean, and baseload.

The thing that people forget is that nuclear is by far the cleanest and safest baseload. It’s actually safer than wind power per terawatt-hour.

Sorry, how do people get hurt with wind power? Just the maintenance of the turbines?

Yeah. If you look at all downstream mortality, it’s more dangerous than nuclear. Then, if you look at all downstream CO2 emissions, nuclear is actually the lowest, lower than wind or solar. If you think those things matter, and that baseload matters, then it will have to be nuclear. The one thing nuclear has not done a good job of, compared with everything else, is cost. I think that’s actually what it all comes down to. The question of whether we deploy lots of gigawatt-scale reactors or handle the scaling through SMRs comes down to who is actually deploying more quickly and cheaply on a per-megawatt scaling basis.

Isn’t the true competitor for nuclear “solar plus batteries”? Both are coming down a very aggressive cost curve, so you could just be cost-competed by solar and batteries in a lot of use cases.

I believe the solar cost curve. I haven’t seen data that’s as compelling on the battery cost curve. That could be coming, but if you want a data center that’s five-nines, you need a lot of storage. I’d actually delete one thing, the battery piece, and say the competitor for nuclear is actually pure solar: solar in sun-synchronous orbit for data centers, if we’re talking about data centers. I do think that’s the one other angle that’s pretty smart and could possibly win. If we’re talking about power on the ground, I think nuclear still wins, whether that’s for data centers, industry, or whatever it is.

How will cost get solved, or will it get solved?

It will get solved. The way it gets solved depends on what you think the cause of the cost issues in nuclear is. This is one of those overdetermined problems: “Why haven’t we scaled nuclear?” I think most people just say safety. They look back to the ’70s and say, “There was a movie, The China Syndrome. You had Three Mile Island. You had Chernobyl. Clearly, nuclear was going to stop, and we wouldn’t scale it.”

I do think the scaling piece is actually one of the proximate causes. We never hit economies of scale. We weren’t doing it quickly enough. We never built out the supply chains. The workforces would shift around from project to project. Vogtle was way over budget and way over schedule. But at a time of historically low interest rates, 20% of Vogtle’s cost overruns were interest rate overruns, interest payment overruns. You realize that the overnight cost of a plant, when they take 10 years to build, actually doesn’t matter as much as interest rates and how long you’re actually taking to build these things.

If you think about the big reactors, I think the way we bring the cost down dramatically is just to build them quicker and build more of them. I think that goes a very long way. If you look at the period when things really slowed down and that downward spiral began for nuclear, it was the ’70s. You had oil price shocks, you had oil prices going very low, and you had interest rates that peaked, I think, in 1981 on mortgages—18% mortgages at that point. How do you finance a huge infrastructure project when your cost of capital is well north of 10%, way north of that?

Because the classic old story of what happened is this very accident-based mood around things like Three Mile Island and Chernobyl, and you basically reject that as an explanation. You don’t think it was that we saw these near misses and accidents and said, “Screw that. We’re going to strangle nuclear with overregulation,” and everything like that. You think it was much more of a “What the hell happened in 1971?” type story, where we lost the ability to build large infrastructure projects generally, and nuclear plants are an example of a large infrastructure project?

Yeah. Maybe the safety concerns put a slight pause on it, but I think that’s often blamed as the only thing. Maybe that kicked off this cycle to some extent, but I think it was probably inevitable anyway when your product is not cost-competitive. I think fundamentally that’s the thing in nuclear. It’s the cleanest baseload, the safest baseload, not the cheapest baseload. You have to be the cheapest.

I also think, for people in general, if presented with, “Here’s a new technology. Would you like this to power your grid? You’ve maybe heard some safety issues with it that are probably confusing about which side to believe, and then also your energy won’t be cheaper,” if we’re talking about the 1970s or 1980s, when global warming was not really a topic and emissions weren’t as front and center, why would you say yes to, “I want a more expensive electricity bill”? I think that put us here. I think the way it gets solved is by building more.

The other side of it, if we want to talk about the cost structure being overdetermined, is that you have to ask yourself, “Why did the cost get high? Why wasn’t there more downward pressure on it?” I think there just wasn’t a strong enough incentive in the ecosystem to push prices down. You had utilities that are more like cost-plus organizations, and they’re generally going to outsource construction of the facility. You end up with a dynamic that was, in some ways, similar to the launch industry, with no very strong economic pressure to reduce prices.

It just stopped being a market and started to become a place where people rent-extract in all their local corners.

Maybe not intentionally, but I think there was an absence of strong pressure to keep prices low, push them down, and compete them down. I think that happens in a lot of industries. It happened in space launch as well.

It feels like, in general, the nuclear mood is this very interesting story of extremes, where there was crazy, unbridled optimism early on. Remember the Radium watches, where people were making watches with radium-226 for the glow-in-the-dark dials, and the idea of energy too cheap to meter, and the Jetsons future? It was really ebullient in its optimism. Then there was this crazy pessimism, where nuclear became public enemy number one and the enemy of environmentalists, which is ironic given you’re saying it is provably the safest and lowest-carbon format.

It was a big debate within the Sierra Club, actually. There was about half of the Sierra Club—

There was a pro-nuclear side of the Sierra Club?

Very pro-nuclear, and it was divided on this around the time of Ansel Adams taking photos for the Sierra Club. I remember this as something that created some huge rift. The argument was basically, “We can try to have less energy consumption as a society, and that might be the best thing for the environment, but it seems unlikely. Therefore, the alternative is more coal and other things we're doing. Maybe solar can work, maybe wind can work, but those are more impactful to the environment visually and in terms of acreage. Nuclear may be the most environmental source of energy if you think we're going to consume it.” That was a big internal argument, even amongst the environmental movement around that time.

I hadn't realized there was a rift there, but ultimately one side won out.

Yeah.

Where I was going was, it feels like we are now getting to a better spot where, at least amongst the youth, opinions are very striated by age, but younger people are much more likely to be pro-nuclear. Do you think we're getting to a better spot, generally, in terms of public perception?

Yeah, I think we are. You look not just at age demographics, but also politically. Ten years ago, nuclear was much less popular in general, and it was also a political issue. Now it's got bipartisan support. The Biden administration wanted to triple nuclear production in the US by 2050; the Trump administration wants to quadruple it. There's good alignment politically around nuclear now. I think it's because people who historically would have been more concerned with the environment are realizing it's the cleanest, lowest-CO2-emissions source, and it's actually extremely safe.

Yes. Who do you sell to? Who are your customers or prospective customers? I know you probably can't pick favorites, but what are a few of the most exciting approaches that you see amongst the startups?

If we're talking about specifics, let's just go smallest to biggest, in no particular order.

You're so diplomatic. Finish your Guinness.

The smallest would be Antares. That's going to be more for forward-deployed military, space, things like that. They have an incredible business in that market. Next size up would be Radiant: 1-megawatt, very good reactor design, containerized for remote applications, primarily in base resilience. Great business there.

Then you start getting up into the SMR size, and you have people like Aalo, Valar, or even Oklo. Then you go bigger. That would be serving the data center market, primarily. That's probably the first application and the one that's most urgent. Then you go larger than that, and you're looking at Oklo, X-energy, TerraPower—larger formats that can really compete on the grid in the near term. Not all of these are using HALEU today. Some might use low-enriched uranium, or LEU+, 6–10%, because that's available—

But all of them could. It's like putting 100-octane fuel in your car. It can take it.

I don't think all of them can. If you go big enough—those are all the advanced ones. I think none that I named... I'm not aware of any limitations of any that I named. I think all of those can run HALEU at some point. Some are choosing to run LEU or LEU+ today because that's what's available. If you go bigger on the grid today, those large reactors, those gigawatt-scale AP1000s, the Westinghouse design, cannot run HALEU, but they are looking at operating and running LEU+.

Got it.

Again, going back to what we were talking about before, you increase the enrichment level, and you don't have to refuel as often, or you can get a little bit more power out of it. So that's the landscape. Think large-format light-water reactors, really the ones people associate with Homer Simpson: that's low-enriched uranium. Everything else that's happening generally trends towards a higher enrichment level, just under 20% HALEU.

Is this relevant to submarines and all the non-power-plant applications?

No, submarines care even more about refueling cycles. When you refuel a submarine, you have to cut it open and remove the reactor, so as infrequent as possible is ideal there, and they'll run north of 90% enrichment.

Which is, again, getting into a different regulatory regime. You cannot, as a civilian company, use that.

Completely different, and that's why all of the new SMRs, all the advanced reactor designs, specifically choose HALEU, 19.75%, just under 20%, which is the cutoff for different regulation internationally and domestically.

I see. 20% is the cutoff, because beyond that, you're getting...

You're getting close.

Too close. That makes sense. Where are you guys in the development of the company? We're talking a lot about the market, but how many people? Do you have a bunch of uranium sitting in a warehouse somewhere? When do you enrich it? Just a description of where you are on the journey.

Big picture, we started at the end of 2023. Today, here we are, almost three years later. The team is just over 200 people, scaling pretty quickly. We have facilities in LA as headquarters and engineering. The facility that we're building is in Paducah, Kentucky, so western Kentucky. Everything we're doing at the company is aimed towards bringing that online by the end of the decade. That means licensing. We just submitted our NRC license last week. It means all engineering efforts. It means manufacturing, construction—pretty much every piece of this is targeting that end-of-decade timeline.

You're delivering enriched uranium by the end of the decade?

Yes. On the commercial side, all of our contracts are lined up against that date, because we know that's when the ban on Russian uranium imports starts, January 1, 2028. Utilities do have some inventory, and they do have some other contracts that can see them through for a little bit, but we want to be online as soon as we can after that to help plug that gap, which is about a quarter of US demand. Then all of the advanced reactors are more or less targeting first deployments in '28, '29 commercially, and then scaling up from there. That's about the right time to be online for the market, and it's about as fast as we can go.

You worked at SpaceX. Elon's companies have this green lights to Malibu concept with planning. The idea, as I understand it, is that it costs you time to plan conservatively a timeline that you can hit because you're building buffers into the schedule. Instead, you should plan based on a green lights to Malibu, what-if-all-the-traffic-lights-were-green basis. I'm curious what you think of this idea, generally. But then, second, is this end-of-the-decade an Elon-style green lights to Malibu, we'll-have-FSD-next-year-type timeline, or is it a timeline you guys are going to hit?

This is one we'll hit. That's our external timeline. We're speaking publicly.

I see. You have a secret internal timeline as well.

The secret internal timeline is to go as fast as we can. We do have an internal timeline. That schedule philosophy of green lights to Malibu comes out of this idea that if you give people more time, they will simply take more time. There is a minimum. And you can maybe go too quickly and squeeze all the time out of the schedule to the point where now you're doing things that are not in support of the correct architecture or a very long-term cost structure. That's always the critique.

That is the tension.

That's the tension. That's even the internal tension at any company trying to run a green lights to Malibu schedule: Number 1, do you have waste? Number 2, do you make the wrong architectural decisions? Then number 3, how do you manage people against this plan? If you're doing performance-based management and you give someone a very difficult plan that maybe is perfection, it's more difficult to judge performance against that.

You have to have in your own mind, "What's a little bit of buffer? What's a reasonable thing?" It's often very useful. Green lights to Malibu is not only a scheduling tool; it's also a constraints tool. If we really try to dial this in and remove all buffer, what breaks? What's the thing that's the critical path that's preventing us from...

It's a bottleneck discovery tool.

I think it can be, when done properly. The critiques I raised can be dealt with in other ways if you're just extremely exacting about, "No, this is the architecture we're going to do." If leadership is very deep in the weeds on every engineering decision, they can say, "We have this schedule, but no one in the room thinks we should do this thing just to shave one day. Don't do that thing. Let's do it right. Let's take three days back. It's okay; we can make it up elsewhere."

What did you work on at SpaceX?

SpaceX, when I started, I was just an intern, in 2003.

You were there really early.

Yeah, pretty early. One of the first interns.

SpaceX was founded in '02?

'02. Very, very lucky to graduate at the right time in aerospace, when prior to SpaceX there wasn't too much interesting to do. I remember the first summer was working on test stands. We were getting ready for an engine to be able to be mounted vertically and put on a test stand.

First winter break, I was working on a surrogate for the engine. Then after I graduated, I worked on the engine systems: the Merlin engine, a bunch of components on there.

I didn’t design it myself. More experienced people were doing that, but I helped make sure the individual parts could withstand the thermal and structural loading.

Then after that, I worked on a series of engines across a few vehicles, including the Dragon capsule, which at the time was just cargo. I was part of the early team and got to work on a few subsystems there, so I was there from 2003 to 2007.

I’m always struck by how influential Elon’s management style and philosophy have been in US manufacturing, and Chinese manufacturing for that matter. Most of the companies founded later explicitly reference some of his tenets. What has been influential for you in terms of design philosophies?

I think what I draw most from the SpaceX experience is working on something that not everyone agrees is the most important thing. It overlaps with Peter’s philosophy.

Picking a flying car is not a 140-character type of problem.

Pick something that’s almost, in some ways… What industries haven’t lived up to their promise over the last 50 years? What are the actual bottlenecks to these things? What are the markets we know need fixing, even if it’s not a completely new greenfield thing, but a big lever on a much bigger industry?

At the time, in the early 2000s, everybody was talking about small satellites. How do you do smallsats? How do you do CubeSats? Then it turned out there was no good way for them to get to launch. When your satellite costs a billion dollars, $100 million for a launch isn’t so bad. When it costs $1 billion, now you really need to think about it.

It’s almost the same thing with a small reactor, an SMR, a small modular reactor. Now you’re no longer doing a multibillion-dollar construction project, but you need the fuel. In fact, the fuel cost matters proportionately more. For some of these SMRs, the total LCOE could be 50% fuel cost, including enrichment.

It’s that high?

In some cases, yeah. It depends on the exact design, but at today’s market prices, it can be.

LCOE is levelized cost of energy, but basically the energy cost, right?

Basically, it’s the cost of producing energy, factoring in both the upfront capital for the reactor and the operating cost. The fuel cost starts mattering a lot, which is the second part of what I appreciated in 2023. Not only did we need to remove this bottleneck, but for advanced nuclear to really scale on HALEU, we’d want to bring down the cost. The biggest piece we felt we could influence turned out to be enrichment as well.

It feels to me like one of the big challenges for you guys is the same challenge many hardware companies face: the dollars and time required until you’re profitable, or even have meaningful revenue to fundraise against. So many hardware companies have had good ideas and gone out of business because they couldn’t prove to investors that the idea was good before they ran out of money. Ultimately, there’s this fundraising risk. How do you burn down the fundraising risk that’s inherent in the company?

If you really look at the big picture, you’ve got equity funding, debt funding, and non-dilutive funding. In our case, when we began, we competed for a program that Congress authorized to help address this ban—

As part of the ban?

Exactly, and the lack of availability of fuel for advanced reactors.

We competed for that. We were down-selected under the Biden administration and awarded $900 million under the Trump administration earlier this year for capacity building. That helps us stand up the first facility and takes a lot of that cost off the table. It’s a big lever for getting past some of the challenges companies often have in getting their first project online.

That’s great. How do you spin up on all this material? You’re not a nuclear physicist by background.

Luckily—

You are?

No, I’m definitely not, and luckily, I don’t need to be. Everyone thinks of enrichment, hears “nuclear,” and thinks, “Is your team half nuclear scientists?” The answer is that there are actually relatively few. Very few.

Just chemistry?

Not even chemistry. In our facility, there are no nuclear reactions. There aren’t even any chemical reactions. What you have is a separation of a gas. You still need chemists and nuclear scientists, but mostly for safety analysis, mostly for making sure you don’t have those things happen. No chemical reactions, no nuclear reactions.

That’s a good question because I don’t know much about enriching uranium, so this is good. It’s filling in some gaps for me. You mine uranium out of the ground—

As a solid.

Exactly, as a solid. It’s like a rock-looking thing. It’ll look like a rock to you. Then you boil it?

Then someone else has step two after it’s out of the ground. That’s called yellowcake, U₃O₈. Classically, it would be open-pit mining, and then you have a milling process to turn the ore into U₃O₈. Today, that’s done with in situ recovery, like enhanced techniques that involve pumping a fluid underground and collecting the uranium out of rocks. That comes out as U₃O₈. Step one: get the U₃O₈. That’s mining and milling. Step two: take the U₃O₈ and chemically convert it into a gas. That’s a separate chemical facility that does that work.

I see. That’s a chemical process to convert it into gas.

Then it would arrive at our facility as a gas that we can separate.

You just spin it in a centrifuge. What else is in there? What are you separating it from?

There are a few techniques one might use to do this. Where we’re doing it in Kentucky, we’re at the site of the last place the US did enrichment, in western Kentucky, in Paducah. There was a facility that was shut down in 2013 on a DOE site. That was a gaseous diffusion process. It was the first technique that was pioneered. The Europeans use gas centrifuges; other countries have used aerodynamic processes. There are even aqueous processes, as well as laser enrichment. So there are at least five techniques that are classically used, and our technique is one that we’ll unveil in the first half of next year, when we’ll talk about it.

It’s gaseous diffusion, and so you…

That’s what was done in Paducah.

I see.

That’s not necessarily what we’re doing. We’re building a fresh facility in Paducah, on the same site, in the last community that did enrichment in the US.

If it’s simple, why do you need to be in Paducah, where the last version was? Did they leave a bunch of machines behind, and you get to say, “Wow, there’s all this free stuff here”? Or is there expertise among local employees?

Yep.

Why Paducah?

Our search for Paducah lasted over a year. We went to over 10 states and looked at over 1,000 plots of land. The answer in the end was the obvious one: go to where the community most recently did this. You have a combination of two things. We always felt like the community was the most important part. We’re building a facility that’s going to last decades. We want to make sure the community we’re building in actually welcomes us there.

Paducah was extremely familiar with enrichment, comfortable with it, understood it to be a positive thing for the community, and had the talent. So it was this combination of community support for what we’re building and the talent base there above anything else. The DOE site had other benefits. By building on a DOE site, we were able to have some shared services. So out of the gates, there was a little bit less to stand up, and there was already a piece of land there that’s 100 acres, just big enough for us to build a first facility, with DOE security around it. There were lots of different benefits.

Presumably, it’s simpler from a permitting point of view because you’re working with DOE rules, and not county and municipality rules?

Could be. We’re primarily doing commercial sales out of this facility, so we will be NRC-regulated.

But federal, and not state and local?

The biggest benefit was that the site had been heavily characterized environmentally. The environmental work required to get to a point where we knew what the environmental impacts would be was quicker than it would have been on a completely uncharacterized, separate piece of land.

That makes sense. Again, going back to enrichment, what are you separating the uranium from when you’re enriching uranium?

From itself. You have two isotopes.

It’s the different isotopes. I see.

Uranium out of the ground is 0.711% U-235. That’s what you want. That’s the number when we talk about 3–5 or 19.75. It’s 1%—

It exists in the world as 0.7%, and then you’re enriching it to a number beyond that.

Exactly. The rest is primarily U-238, which is not the fissile material you want.

If the isotopes have different masses, why can’t you just spin it around in a centrifuge and have it separate?

Many people do.

I see.

Many people do. Or you can do an osmosis-type process, which is gaseous diffusion. Or you might split the gas using an aerodynamic process. There are a lot of options.

It does sound quite simple when you put it that way.

When you put it that way.

Why is it hard?

The hard part about most of nuclear, I think, is that nuclear is perceived as this incredibly complicated thing.

I think fusion is pretty difficult. That is really new physics. That’s new science. Fission has been around for a long time. If we just talk about reactors, it goes back to economics more than anything. To me, with all the advanced reactors versus gigawatt-scale reactors, the most interesting thing is not just the technology. It’s really: What is the cost implication going to be?

4. Competing on Enrichment Cost

For us, enrichment has been done before. A number of countries do it. But the real challenge is that we don’t only want to bring it back in the US; we want to make it globally competitive. We want the US capability to be truly commercial, able to compete with Europe, Russia, and China over time. So it’s all about cost. I think the hard thing is not simply doing enrichment; it’s doing it in a way that can be scalable and highly cost-competitive. That’s the North Star.

You don’t worry that this is a bad market somehow? The fact that there isn’t a US producer suggests there’s something a bit broken about the market and the buyers.

No, I don’t think so. I think there’s a path dependency in how we got here. We can talk through that, but—

No, please tell me the path to entry.

I think the history of it was that the US developed first-generation technology to win World War II and stay dominant in the Cold War. We were over 90% of the global supply. Ultimately, that approach, that specific design and technology, was not cost-competitive with the European and Russian approaches. We shut those down following the Cold War.

The Berlin Wall fell. The US entered a trade agreement with Russia. It was nicknamed Megatons to Megawatts. We imported Russian warheads, blended them down, ran our reactors on them, and started to trade with Russia that way. We’d been trading with Europe since the ’70s. Once we felt like the Cold War was over, there was less reason to keep these plants running in the US.

It was basically a glut moment in the US?

Why are we running these more expensive plants when we’re now in a world of free trade and can trade for this? At the same time, there were efforts inside the US to develop new technologies, but they didn’t really hit the cost structure required to compete internationally. They’re still in development to some extent, trying to get on that curve. So I think it’s simple: We used to do it, then we relied on free trade, and only more recently has this seemed like an issue. The issue is really around HALEU availability for advanced reactors and low-enriched uranium availability for the larger reactors, given the ban on Russian imports.

If you tried to start a company like this 10 years ago, you might say, “What’s the market?” The utilities are fully contracted. There are no gaps. Why would they leave gaps? They’re going to make sure they have a reliable source. Then there isn’t a really big advanced reactor market you could sell into. So, both on the LEU side and the HALEU side, it would have been much more challenging to get to some critical mass that could support a new entrant. Today, that window is open.

Classically, cost advantages in manufacturing come from scale, which is one common factor, or simplicity in design. It feels like the market is what the market is. It’ll grow, hopefully, but it’s not like you guys are going to come out of the gates targeting 10x or 100x larger scale than the incumbents. Also, when you say simplicity, as you described, there are a few approaches that everyone knows here, so you don’t get to have a radically simpler process. How do you plan to have a cost advantage versus all the incumbents?

I’d say you missed one more. There are economies of scale, there’s simplicity, and then the third, honestly, is decades of technology improvement that no one’s captured. The last time anyone, to our knowledge, reinvented enrichment—at least if you look at the incumbents being Russia, China, and two European firms that use the same underlying technology—the last time anyone’s done a clean-sheet redesign was decades ago. If you think of all the progress in technology over the last few decades, there’s a lot you can do in addition to economies of scale and simplification.

Meanwhile, there’s no public-domain technology in this space from since the ’50s or ’60s. So you look at those few decades of technology that you could leverage. Then you do a clean-sheet redesign, doing something very novel, and you can end up somewhere pretty different.

I want to try something on. It feels like sometimes you have a trend that sweeps an industry and takes place at many companies at once. In the US, in the ’70s and ’80s, this happened with lean manufacturing. A lot of the success of private equity firms at that time, or early Danaher, came from all these US firms that had not embraced lean manufacturing. You can buy a manufacturing firm and adopt the lean stuff, like better inventory management and everything like that. You can make the business much better because it turns out this is a managerial technology that had not been adopted by a whole bunch of firms.

Maybe there are people doing this today with AI, applying it to services industries.

Exactly. That would be another example. It feels like this may be another trend you’re a part of in hardware. There’s stasis in companies: there was stasis in launch until SpaceX came along and targeted much lower-cost launch. There was stasis among the primes, and then Anduril, and now a whole bunch of other neo-primes are coming along.

Similarly, you’re identifying stasis in the existing companies that enrich uranium. You can very easily… Does that resonate with you as a trend where something hasn’t been adopted across a whole set of companies, and entrepreneurs are coming along, finding those weak spots, and building better competitors?

It seems like it, although each of those three examples is at least a decade apart, I’d say, on average. SpaceX, Anduril, and us would be a decade apart on average. The only way those opportunities were found was by people trying to go deep in the space. Nobody really thought space launch was the thing until Elon tried to launch something and realized it was the cost of the launch. Trae at Founders Fund was looking for a defense company and couldn’t find someone taking this holistic approach. Then we only discovered this because I knew many or virtually all of the advanced reactor producers, and they had the same issue.

What is the thing that all of the incumbents miss? Why do they become slow and stodgy? Presumably, this is not a problem for you yet. But it will be a problem when you’re successful, and you’ll also need to avoid becoming slow and stodgy.

You definitely want to avoid that.

How do you avoid becoming ULA or Lockheed Martin?

I think all those industries—not to single them out—were cost-plus industries. That was my experience. Before SpaceX, I worked as an intern at one of the aerospace incumbents.

Which one?

Boeing. I worked on a military program. It was cost-plus. There was a big project, and there was a goal of getting online. At some point while I was there, it felt like more and more people were working on the project, but it didn’t feel like we were going faster. You realize that’s one tool: put more people on the project. If there’s not strong downward pressure on cost, your incentives just aren’t aligned with moving fast and doing it in a lean way. I think these industries that end up getting stagnant often have that same characteristic.

Now, people assume that once a company gets to a certain size, it becomes that way. I think it doesn’t have to at all. I think the best companies that get large, even to tens of thousands of people, always have that new project, that new thing that’s borderline existential, where if they mess it up, they’re going to be in a much worse place than if they’d never tried it. It almost risks the business. I think that’s the thing that keeps a company young: You just have to keep doing that next project and not just say, “I have this nice business. I’m just going to earn my economic rents and do nothing else ever again.” You have to have a vision for the future of what you want this industry to be.

What were your other takeaways from Boeing?

Really, the main takeaway I walked away with was that the cost-plus structure was not the right structure to motivate moving fast. That was just empirically evident from working at both.

It sounds like you more broadly think it’s a completely broken incentive system.

I think the whole point of cost-plus is to make sure people can work on projects whose costs are very uncertain. We have a very urgent project; the government needs you to do XYZ. We need some of the suppliers to come forward and actually say that they’ll try to do this, and it’s for everyone’s good.

It’s a wartime, not a peacetime, incentive system.

I think so, or it’s something where it’s very urgent, has never been done before, and it’s very uncertain what the cost will be. Once you get to equilibrium, I don’t think that structure makes sense, at least not to motivate the right behaviors. I think the bigger thing I walked away with after Boeing, even from just being an intern at SpaceX when it was under 50 people, was that at the end of the first week or month, I felt pretty certain SpaceX would eventually win, even though it was at most 1/1,000th the size. It was just because of the speed of execution.

The thing I walked away with, especially in the early years, with that as a more recent memory while I was at Founders Fund, was these small teams.

Even though they seem completely outmatched, by moving quickly with a good group of people, you can still beat incumbents.

It gave you that ability to be optimistic about the odds of a five-person startup.

Bet on the small guy. If they have a vision, if they have some insights about how they want to do it differently, especially if they're motivated by something more than just, “Let's stand up a small business here.” No, let's actually change things long-term.

The three examples I used were SpaceX, Anduril, and General Matter. I hadn't realized until you said it, in formulating those examples, that all three are deeply tied to Founders Fund. What is it about Founders Fund and the culture that makes it a good breeding ground for these sorts of companies?

Maybe the one thing: SpaceX was started before Founders Fund. I don't think Founders Fund can take too much credit—

Founders Fund was an early high-conviction backer.

First outside investor, at a time when very few others wanted to do it. I think the—

I'm going to go with deeply tied.

Yeah. SpaceX gets the credit. Elon gets the credit. You could say that it's something coming out of the whole zero-to-one mindset, and these frameworks of working on something that's important that other people don't realize.

What important truth do you believe that most of your peers would disagree with you on?

Don't go after a trillion-dollar market on day 1. Focus on a small market that really needs a new solution, and satisfy those customers. Make them love your product, own that market, and go from there. Those, I think, are common to those companies.

Probably even more than that is Founders Fund does not programmatically start companies, and it doesn't want to either. Everyone there is selected for wanting to be an investor, not hoping that they can someday be an entrepreneur. I think this leads to a mindset of, “Don't start companies. Don't do it programmatically.” It's almost like that airplane with all the war wounds on it, which makes you think, “What about all the companies that didn't get started out of Founders Fund?”

I think the bar when you're starting something, at least for me, was always that being an investor is a great job. It's a great career. It's awesome in so many ways. To actually go start something, it had better be extremely important. You have to believe no one else is doing it, which is why, in both Anduril and our case, there was a long search for someone doing it. Because if there was, it wouldn't be worth starting a new company.

Then finally, the one that's seemingly the most difficult is that, in my case, it should be something where my skill set actually contributes to making it more likely. Being extremely important, having no one doing it, and actually lining up with my skills is such a small Venn diagram that it's almost like the activation energy is so high to go do it. It has to be extremely compelling.

Then most of the risk, I think, ends up being on execution. You end up with these forgotten opportunities that were overlooked for a long time, but that are actually really central to an industry taking off. It just so happens no one did them for a long time because there wasn't the convergence of market opportunity, and often it was just extremely hard.

How much conviction did you get on the idea before you took the jump and went and started the company?

I would say, extreme conviction.

What are the intermediate milestones for you guys as you approach production in Paducah by the end of the decade? Presumably, that is not the target on the whiteboard in the office. You have short-term targets.

Nothing in between. If you think of the big workstreams to do that, at the end of it, you have a large facility, and it's filled with equipment, and the equipment is operating. If it's operating, that means it's licensed. One workstream is licensing. As I mentioned, late last month, we submitted our NRC license. That's one of the workstreams.

Another one is a big building that's operating. Let's work backwards. You have your license; that's how it's able to operate. But the building also has to exist, and so you have a large construction effort.

Typically, in the nuclear space, people outsource the construction of their facilities. In our case, we copied the Tesla playbook: build your own in-house construction EPC team.

What's EPC?

Engineering, procurement, construction. Design your buildings, order the steel, build it. We have our own GC on the construction side. We have our own procurement and engineering teams internally. So that building is being designed and built, and it'll exist in time to install hardware, commission it, and operate it by the end of the decade.

Then, of course, you need all of your equipment inside. That's ordering from your supply chain. It's manufacturing, and manufacturing final designs that hit the economics that you want. I think a lot of people think of nuclear, and they think it's all about core technology or licensing. For the reactors, it's actually about—if you're a big reactor, it's probably about a construction project. If you're a small one, it may be mostly about building your factory. For us, it's more that; it's more manufacturing.

Manufacturing is the thing that relatively few companies in the US have really proven they can do at scale. I think that'll be the really exciting part in a year or two.

Last question. We've talked about what you learned at SpaceX and Boeing and various places. What did you learn working at Founders Fund?

Avoid competition. That happens automatically if you work on something that's really important, that no one else is doing, and that you can uniquely be good at. Avoid competition—you achieve it that way, but it's also something that will inherently be true if you're solving a problem that's not solved.

What important truth do you believe, or what important problem is no one working on? I think that was actually the thing that led us here. It wasn't a desire to start a company, it wasn't anything specific like that. It was: What thing is actually holding back the nuclear industry? What is the bottleneck? What is the one thing that could prevent all the reactors from scaling, all the SMRs, or gigawatt-scale light-water reactors? It's fuel. A lot of the other pieces have been solved, and so it's just fuel.

For other future entrepreneurs, there's presumably equally big opportunities just staring them in the face.

There certainly are. I don't know what they are. I think it takes a lot of digging to see them.

Enriching.

Then enriching and turning into a gas. No, I think you only find out about these things if you follow some passion that you might have that not everyone's thinking about. It's usually not part of a trend, or at least it's ahead of the trend. If you're into something, and you think it has value, and you go deeper, and you work in that space, you'll see what the problems are. I think they become obvious to those who go deep enough.

Scott, thank you.

Thank you.