Yes, Nuclear. How Much? How Soon?

Here’s the transcript of a discussion I had at Southern Methodist University with Jim Burke, the CEO of Vistra, and Ray Rothrock, a longtime nuclear investor and board member at Centrus Energy.

Photo: Lorin Bryce

There’s no shortage of cheerleading about the future of nuclear energy in the US. That’s why I was pleased to sit down on April 9 at SMU’s Energy Outlook 2026 with Jim Burke and Ray Rothrock. (Sincere thanks to my friends at SMU, Xavier Tison, and Bruce Bullock at the Maguire Energy Institute for inviting me to participate in the symposium.) As you will see, Burke and Rothrock are extremely knowledgeable about the electricity business and both are very sober about the future prospects for nuclear energy in the US.

Here’s a transcript of our discussion. It’s about 7,000 words. And for an extra bonus, at the end of the panel, Ray and I did an a capella rendition of Reddy Kilowatt’s theme song.

This post is free for all readers. Enjoy.

Participants:

• Robert Bryce

• Jim Burke, CEO of Vistra Corp., one of the largest independent power producers in the US. He’s also the vice-chair of the Nuclear Energy Institute.

• Ray Rothrock, CEO of FiftySix Investments, a seed fund focused on cyber, energy, and climate entrepreneurs, and a board member at Centrus Energy, which supplies nuclear fuel to civilian reactors.

Bryce: We’re here to talk about nuclear, and. I don’t have to tell you all that nuclear is one of the hottest things going right now. Our caption for this session is: “Yes, nuclear – how much, how soon?” The Trump administration has set a goal of 4x nuclear, or 400 gigawatts by 2050. I’m going to put you on the spot right away.

Possible? Yes? No? If not, why not? Jim?

Burke: I think it’s possible. I think it’s really challenging. I don’t think we have a track record of adding 5 or 10 gigawatts successfully, realistically. So to get to that number, a lot of assumptions have to be made, Robert. I know you describe yourself as an optimist on nuclear, and of course we’ve leaned into nuclear as a company, but there’s a lot of complexity to that solution. We need a lot of things to line up, probably differently than they have so far, to be able to make it. But that’s a long horizon. In truth, it’s harder to hit the 2030 goals, because you have to be well underway already to hit some of these uprate challenges that we’re all working through.

Bryce: Let’s come back to uprates, because I want to talk about those, and also about how, as an independent power producer, you have a different view on the market.

Ray, 4x by 2050? Yes? No? Maybe?

Rothrock: No, it’s not possible. We built about 100 reactors between 1961 and 1985 – roughly 25 years. To do 400 now by 2050, and keep the 100 we have operating and build another 300, I just don’t see how it’s possible. You can build 75 or 80, I think. But 400? No.

Bryce: Okay, I’ll come back to that. Jim, you’ve been in the power business now for a good long while — 25 years? Vistra’s been on a rocket-ship ride. I just looked before we came up here: the price-to-earnings ratio of Vistra is 74. You’re trading like a tech company.

Burke: Well, there’s GAAP accounting and mark-to-market…

Bryce: Careful now. My first book was on Enron, so we don’t want to talk about mark-to-market. I’ll leave that one. But what’s driven that? What’s driven your valuation? Your stock has roughly doubled in the last 24 months. Explain why that is, because I want to lay the groundwork for what’s happening in the power market, and you’ve had a front-row seat. What’s changed? Is it just data centers and AI?

Burke: No, it’s not just that. Let me rewind a little bit, because this looks like a very recent phenomenon, but the foundation was laid earlier. In 2018, we made a very big decision to acquire Dynegy. At that time, that doubled the size of the company. It was primarily a combined-cycle gas turbine fleet. Our company, Vistra, its heritage is TXU. We were only in the state of Texas up to that point. Texas is the 10th-largest power market in the world, but we wanted to expand. At that time, you could pick up combined-cycle gas turbine facilities at, say, 400 to 500 dollars per kilowatt, which meant they were out of favor.

From 2018 to today, you’re seeing the price of not only existing gas assets but new gas assets rise considerably. So we’re in a position where, if someone wanted to replace all those assets, the cost would be far higher than what we paid. That has nothing to do, at that point, with just what’s happening with AI.

We’re building too. We’re building gas. We’re doing some of the nuclear uprates. But I think the market is realizing dispatchable power has more value. Ten years ago, there was a view that wind, solar, and batteries—which obviously weren’t as big then as they are now—could run the electric system. I think folks are revisiting those assumptions.

We did other acquisitions since then as well, but there’s no doubt that the AI tailwind has been a pretty recent phenomenon. I’d say the foundation for what we’ve done was laid earlier.

On the valuation piece, we aren’t valued off P/E. That literally can move a tremendous amount because of accounting. Folks view us through the lens of: are you delivering a cash return when you invest? Ultimately, cash is king. That’s how we’re valued. That’s a much more stable valuation metric. Our earnings on a cash basis have been rising through time as power markets have tightened.

Bryce: So you were buying at $500 per kilowatt, and today to build it is $3,000 per kilowatt?

Burke: About $2,500 to $3,000, all-in cost, for a new combined-cycle gas turbine. That’s correct.

Bryce: Ray, you’ve been around investing for a long time. You’ve seen what’s happened with Vistra and more broadly in the market. How do you think about this as a long-time investor with Venrock and in the context of what’s happening with nuclear and power?

Rothrock: I’d say two big things have happened. First, until Trump came to office the second time, nuclear always played second fiddle. There were a lot of headlines that prevented it from getting a real launch, even though companies were starting and people like you were investing in startups.

What’s happened is that executive orders changed the game totally. They basically rewrote the book. Some of the ideas that came out of that are things many of us have been talking about for a decade—like opening up federal lands and facilities to allow new nuclear projects and related industrial activity.

Second, there’s been a huge surge in demand from things like data centers. A significant portion of that investment—call it 25 percent of the total—is essentially power demand. If you take a big AI and data-center build-out, hundreds of billions of dollars, 25 percent of that is power. That’s demand that companies like Vistra and others have to supply.

So now there’s a forward demand signal. That’s affecting stock prices. These companies are going to be the providers of that power. So what you’re seeing is really about supply and demand more than about some abstract price-earnings ratio. This uplift you see in the projections—it’s very real. There’s no way to meet it only with gas and solar and batteries. Those help in the meantime, but in the long haul, you’re going to need nuclear and other long-duration, firm resources as well. That’s what I think is affecting the outlook for nuclear power.

Bryce: Let’s go back to nuclear. Ray, what are the key bottlenecks? We’ve talked many times about regulations, capital, and fuel. Force-rank those, if you can. Is there one more than another that’s slowing the build-out of new nuclear?

Rothrock: Before last year, it was regulation. NuScale’s design certification took 61 months to get through the NRC. Sixty-one months for a pressurized water reactor. You’ve got to be kidding me. We’ve used pressurized water reactors for 60 years, and NuScale had to spend hundreds of millions of dollars to get through the process.

That’s going to change. The rules now say they’ve got to get a permit through in two years. The NRC now has a five-person commission that’s all-in on advanced nuclear, and the chairman is strongly pro-nuclear. I’ve met him; he’s said, “I’m not going to be a roadblock.” So that regulatory barrier is starting to come down.

The second bottleneck was capital. These plants are expensive, as we all know. Here’s a little rule I learned when I was a nuclear engineering student at Texas: about 80 percent of the cost of any capital plant—whether it’s a power plant, a gas plant, or a big industrial facility—is steel, copper, and concrete. About 80 percent. The rest is engineering, labor, and soft costs.

With regulatory uncertainty and no clear demand signal, investors said, “We’re going to wait.” That was a big factor. These projects also take a long time. If you look at the first generation of large reactors, it took around seven years on average to build them. Some took far longer; the plant I worked on ultimately took around 14 years from conception to completion. Vogtle is a recent example—decades from initial planning to completion.

We have to get those timelines down. One of the tailwinds now is that the NRC is changing, Wall Street is falling in love with nuclear again, you’re seeing big valuations for companies like Oklo and others, and there’s a lot of activity from new developers. Capital has started to show up. So as the regulator gets out of the way and the money shows up, the remaining bottleneck is physical: the supply chain and labor. You need hundreds of thousands of people trained in nuclear—not just operations, but construction, welding, quality assurance, everything. That’s a real constraint.

Bryce: Jim, how do you see it? You’re building gas plants. What’s preventing you, as the head of a company like Vistra, from saying, “We’re going to pull the trigger on new nuclear”? What are the bottlenecks from your vantage point?

Burke: What Ray hit on has been massively important in getting people excited about nuclear. But the conversations we’re in are framed by how our markets work.

Let me set the stage. Roughly half the country is in effectively competitive wholesale power markets. The other half is fully regulated. So you have two overarching market models. Everything we do is in competitive markets. We take different fuels—it doesn’t matter whether it’s sun, gas, coal, or nuclear—and convert them into the same finished product: electricity. The question is: is the market going to pay you for that finished product, or is a specific customer going to pay you?

Fortunately, for uprates at our three nuclear sites in PJM, Meta has said they’re willing to pay for the uprates. That’s a very clear example of a hyperscaler connecting the dots to a specific form of generation—and an expensive one like nuclear.

Bryce: Sorry to interrupt. So Meta has effectively written you a check to cover the cost of adding about 150 megawatts or…?

Burke: We’re going to add just over 400 megawatts, and we’ll do that over about an eight-year period. It takes time because you have to sequence the uprates with the regularly scheduled refueling outages at these sites.

They’re not paying with a check up front. They’re paying for the output over time. So yes, we have a long-term power purchase agreement for the additional output—for 20 years. The real question, and I referenced an earlier panel on this, is: does a specific customer pay, or is society in general instructed to pay?

Because if you get a quote from an EPC for a new nuclear project, the big question is: if there is a cost overrun, who bears that risk? As a competitive company, our shareholders would struggle to accept the kind of unbounded risk that’s been demonstrated in previous projects.

Uprates are different. There have been about 170 uprates across the country. It’s a fairly well-understood process. New builds have been much more unpredictable. So even if you gave me a permit today and I had the land and the technology picked, we’d still be talking about who wears what risk in bringing that plant to market.

In regulated markets, where you saw the two new units built at Vogtle in Georgia, there was a rate base. Since we don’t operate with a rate base, we have to be more careful about the risk we assume.

Bryce: “A bit more aware?” Nicely phrased. So you’re doing 400 megawatts of uprates. Give me a sense of the cost per megawatt, and how that compares to building new gas or new nuclear.

Burke: As a publicly traded company—my head of IR is sitting here in the audience—that specific cost is confidential. I can tell you it’s more cost-effective than building new nuclear, but it’s still a multi-billion-dollar project, Robert.

Bryce: I’m impressed. Is it more than building new gas?

Burke: It depends on where the latest quote for gas is. It’d be in the ballpark, but I’d say it’s probably still on the higher end of where the gas quotes are.

Because it’s a carbon-free product, and in this case Meta is willing to pay something for the carbon-free attribute. What I keep hearing from customers—and I’m sure anyone in this space hears the same thing—is that there are customers willing to pay a premium, but it needs to be reasonable. That’s when you start looking at what the cost of carbon would have to be to equalize the economics for new nuclear versus new gas. There’s still a pretty big gap to bridge.

But uprates are cost-effective, particularly if you have a customer willing to sign the PPA now.

Bryce: So you mentioned Vogtle already, and I’m glad you did. Under the current construct, can new nuclear work in unregulated markets?

Burke: Under the current construct, I think it would be very tough—unless the customer is willing to take a lot of the risk. Vogtle is part of Southern Company, and Georgia is a regulated market. They also got a big loan guarantee from the federal government.

We could probably get a lot of support from the federal government—low-interest loans to help finance the project, expedited permitting. We could get a lot of support. But even if the financing were in place and a permit were on our desk today, the ultimate risk is still an eight- to ten-year build cycle. What’s the ultimate cost, and who bears that risk?

So in a competitive market, you would need a collaboration with large customers who are willing to pay for that. There’s been discussion about multiple companies like ours joining with other developers to put together a coalition, so there’s a cumulative learning curve we can all benefit from. That’s a complex negotiation.

Remember, this is a market that partly deregulated because of nuclear cost overruns. I entered the industry after deregulation; I came from a consumer products background, so I never worked in this industry when it had a rate base. What the large industrial customers were saying then was: “Gas is cheap. I’m paying for cost overruns on nuclear. I have a better alternative that’s better, faster, cheaper. Give me a chance to choose that.”

Nuclear, under that framing, has not demonstrated that it’s faster or cheaper. “Better” is a debate around reliability, fuel security, and carbon-free attributes. I’d argue that’s in the eye of the beholder. That’s the challenge we have with any new nuclear build—regulated or competitive.

Bryce: Let’s talk about speed for a minute. You’re building new gas. How many gigawatts of new gas are you building now?

Burke: We’ve announced a West Texas plant. It’s going to be about a one-billion-dollar investment.

Bryce: And that’s 1,200 megawatts?

Burke: It’s 860 megawatts. So in your calculus, that project is public. It was attractively priced because we got into the turbine queue early enough, and we’re building at an existing Permian Basin power plant site. We’re tripling the output of that plant. Load growth in West Texas has been pretty strong. We anticipate that to continue.

We do not have a long-term contract for that plant, Robert. Our investors have effectively said, “We understand what you’re doing.” It’s a billion-dollar investment; West Texas is a load pocket; we like it. We’re also converting two coal plants, one in Texas and one in Ohio, from coal to gas. It’s still the same boiler structure, not combined-cycle, but these are good peaking plants. They’re plants the grid relies on for fewer hours of the year, but they’re important. All told, that’s going to be about 2,500 megawatts that we’ll have converted or online in just a couple of years.

Bryce: On the West Texas project, what county is that in?

Burke: It’s in Monahans… Let’s just say West Texas. [laughter]

Bryce: How long will it take you to build that?

Burke: It’ll be online in the spring of 2028.

Bryce: So two years?

Burke: About two years of major construction. We started some initial site work last year, but all of the heavy work is still in front of us. It’s a relatively straightforward project.

Bryce: I want to go back to capital, because both of you have talked about it. I moderated a nuclear panel at ARPA–E in San Diego a couple of days ago. Amy Roma, a regulatory lawyer who used to be at the NRC, said that about 60 percent of the all-in cost for nuclear plants today is capital cost—financing and the “time value” of money.

The way I heard that was: because the plants take so long, that capital component is one of the big obstacles. Is that your understanding, Jim? How do you see that?

Burke: If 60 percent is capital cost, that means the other 40 percent—soft costs, labor, engineering, contingencies—is where your overrun risk shows up. That 40 percent isn’t fixed capital; it can grow. That’s the unbounded part of these projects, as we see it.

Rothrock: At Vogtle, roughly 40 percent of the total cost ended up being interest during construction. Forty percent. And that project ended up costing on the order of 30-plus billion dollars.

Bryce: Ray, you’re on the board at Centrus. Centrus is an enrichment company and has been doing uranium enrichment for a long time, with operations in Ohio and elsewhere. I mentioned Amy Roma. She said one of the key roadblocks for new nuclear is enrichment—the fuel side of the challenge. Do you see it the same way? If so, why?

Rothrock: I completely see it the same way. Centrus is a US-traded company that came out of the original uranium enrichment activities the US government started decades ago. We effectively gave up our domestic enrichment capacity in the 1980s and 1990s, shut down plants in the US, and went overseas—to Russia, to Europe, to other suppliers.

Today, a very large fraction of the world’s enrichment capacity is Russian. I’ve seen numbers around 40 percent of US enrichment supply coming from Russia, directly or indirectly. That’s a strategic vulnerability. The US now has very little domestic enrichment capability. The current fleet—93 or 94 reactors—mostly has license extensions to operate for 60 or even 80 years. That’s fantastic, but it also means we have a long-term, locked-in demand for low-enriched uranium.

As new plants come on, we’ll create new demand. That demand really kicks in five, six, seven years after you start a project, because you don’t need fuel until construction is well along. Even if you put a plant in the ground today, you’re not going to demand uranium for five or six years. At Centrus, working with the government and others, we’re building a multi-billion-dollar enrichment plant in Ohio. It will provide just a fraction of what the US needs, but it’s a start. Other projects will be needed.

Bryce: The US oil and gas industry has grown effectively, in some cases in spite of government involvement, largely because of private ownership of mineral rights and what I call “the rights, the rednecks, and the rocks”—that unique mix of private ownership and technology.

To build that enrichment capacity and the broader nuclear supply chain, is the government going to have to be more involved throughout the nuclear value chain? Are we going to have to be more like Russia, South Korea, or France to get a big increase in nuclear capacity in the US?

Rothrock: I think the answer is yes, at least in terms of strategic direction. This is an important export situation, and this is sensitive material. The US government has to be deeply involved. I don’t necessarily believe they have to provide all the capital. They can provide loans and some subsidies, but private capital and Wall Street need to be part of it too. The enrichment equipment and facilities are not cheap. A lot of people and firms need to participate.

We’re trying to make this a truly American operation, which includes government support, but also private investment and a robust domestic supply chain.

Bryce: So your answer is: for the US to grow substantially in nuclear, if we’re talking about more than 100 gigawatts, 200 gigawatts of capacity, we have to emulate the mercantilist approach that the Chinese and the Russians are using?

Jim, do you agree with that?

Burke: No.

Bryce: “No,” says the independent power producer. Okay, tell me why.

Burke: It depends on what problem we’re trying to solve.

If the only goal is “more nuclear,” then I think what Ray is saying is probably accurate. Even in regulated markets you’re seeing regulated utilities say they would struggle to take a large new nuclear project forward to their public utility commissions right now. So it’s not just a competitive-market challenge. Even regulated markets are saying this is a risk they’re not sure their shareholders or their regulators are willing to take on.

Could the federal government, through, say, the Department of Defense or another federal entity, become a buyer of some nuclear output, and thereby incentivize technologies and fuel production? Absolutely. But if you take a government-sponsored power plant of any form—this is true of subsidies in general—you create interactions between subsidized resources and unsubsidized resources. We’re a markets-based company, and you’re going to have unintended consequences when one resource is heavily subsidized and another is not. In the world of “better, faster, cheaper,” does the consumer pay for power through their power bill and their taxes, or just through their power bill? We’re seeing a lot more intervention in various technologies, and then we sometimes have to pay again to undo those interventions.

We also have an abundant natural gas system in this country that China doesn’t have. We have natural resources that can work alongside nuclear.

So again, if the only goal is nuclear, then a mercantilist, state-led approach might make sense. If the goal is a reliable, secure system that’s cost-effective, you have to ask whether that’s the best energy strategy.

Bryce: As independent power producers, you’re not necessarily going to welcome a lot of government intervention that puts a heavy thumb on the scale for a particular technology.

Burke: I think markets allocate capital. That’s the whole point of capital markets, and that’s what independent power producers do: we allocate capital.

The question is: if someone who doesn’t have the same kind of accountability to shareholders and stakeholders decides to deploy taxpayer money in a specific way, could that accelerate retirements of older plants that still have system value? The answer is probably yes. Then you’re potentially subsidizing new generation to dig a hole and fill a hole—retiring plants and then paying to replace the capacity at higher cost. We also don’t know the true demand profile yet. We’ve actually been more conservative than some of the load-growth assumptions being discussed.

We think Texas is growing fast. Historically it’s been 2 to 3% annual load growth. We think it could be 4 to 5%, maybe as high as 6% per year.

Bryce: Nationwide?

Burke: No, that’s Texas. PJM, which has been relatively flat historically, we think will grow in the 2 to 3% range.

So the question our industry—regulators, transmission and distribution utilities, and generators—has to answer is: what load forecast are we being authorized to build to On the T&D side, that’s rate-based. On our side, our shareholders are effectively underwriting our assumptions. If there’s a risk of overbuild, in competitive markets the shareholder takes that risk. In regulated markets, the remaining ratepayers take that risk, and the cost gets spread over a smaller base if demand doesn’t show up.

There are real consequences to using inflated load forecasts, and then allocating capital based on them. We need to be aware of that.

Bryce: I’ll let you finish that thought, then I want to get Ray back in.

Burke: Just to close that loop, how we meet load growth is going to matter. We need to be disciplined, because both underbuilding and overbuilding have serious consequences.

Bryce: Ray, one quick question before I go back to you. Jim, how many power plants does Vistra have?

Burke: Currently about 50, and we’re adding about 10 more with the pending Coherent/Constellation-type acquisitions and conversions.

Bryce: Total capacity?

Burke: Currently around 44,000 megawatts. We’ll be at about 50,000 megawatts—50 gigawatts—within a couple of years, depending on when our next deal closes.

Bryce: As Jim was talking, what struck me is that it’s easy to think about the US electric grid as one thing. In reality, it’s incredibly fragmented. There are over 3,300 electricity providers. There are more than 800 co-ops, around 2,000 municipally owned utilities, about 180 investor-owned utilities, and then federal entities like TVA and Bonneville. It’s an incredibly complicated sector.

We’ve been dividing it into “regulated” and “deregulated,” but behind that, the complexities are huge.

Burke: That’s right. For example, when we try to acquire assets, we’re now approaching about 5% of the nameplate capacity of dispatchable generation in the country. Take roughly 50 gigawatts divided by maybe 1,000 gigawatts of dispatchable capacity—not counting intermittent resources.

When we try to get a little larger, a lot of antitrust and market-power questions arise. We’re roughly a $50 billion market-cap company. If an AP1000 unit costs roughly $10 billion per reactor, how does a $50 billion market-cap company underwrite a $20 billion two-unit build? Most people believe two units at a site are smarter than one, because of learning and economies of scale.

The short answer is: you don’t. You can’t responsibly take that kind of balance-sheet risk as a single company under current rules.

If no one is allowed to get bigger—because of antitrust constraints—then who fills that void? That’s why SMRs and smaller reactors are on the table. They’re smaller bets. They may or may not be the cheapest bets, but they’re not “break the company” bets. That’s part of the challenge.

Rothrock: That’s a very good point. An AP1000 is a $10 billion project. There are only a handful of companies in America’s power business that can even contemplate that kind of risk.

Which brings us back to how much government involvement is it going to take. I’m not saying government should write all the checks, but we’re going to need someone to take risk capital. That could be partly companies like Vistra, partly large customers, and I think the government, so far, doesn’t want to be the check-writer of last resort. That’s how it’s been positioned to us.

I tend to agree that there ought to be skin in the game. Risk capital should be rationed in a methodical way, not just socialized. There are groups in Washington lobbying for taxpayer-backed insurance schemes to backstop nuclear construction risk. I find that interesting and important, but I’d rather see some of that flipped to incentives: bonuses for building on time and on budget. The Korean EPCs use that model. Incentives can be powerful. And risking your own capital is a powerful incentive by itself.

Burke: I’m vice chair of the Nuclear Energy Institute, so I’m involved with NEI. It’s our job to make sure nuclear is first and foremost safe and reliable, and our goal is to grow nuclear.

At the same time, this country is blessed with many different ways to make electricity. We should keep that in mind as we set public policy because there can be unintended consequences if we pick winners too aggressively.

I want to see all the dozens of SMR technologies and large-scale designs compete. Don’t just pick a winner and repeat it. Let them compete. May the best technology win. That’s how fracking evolved; that’s how many consumer innovations evolved. I think the same can happen in nuclear.

Bryce: Let me follow up on that. On our Substack project, “SMR Intelligence US,” we’re now tracking 41 different US-based companies working on small modular reactors, with 47 different designs. TerraPower, I’d argue, is one of the horses to beat right now. Speaking of the NRC, they recently got a construction permit for their Natrium advanced reactor in Kemmerer, Wyoming.

Ray, I’ll put you on the spot: with 41 different SMR companies—gas-cooled, sodium-cooled, light water, and more—if you had to handicap which reactor chemistry and what size is most likely to win in the market that Jim is describing, what would you pick?

Rothrock: At the end of the day, it will be about economics, reliability, and how well the technology fits into the existing supply chain. The US grid is about 1,200 gigawatts. One percent of that is 12 gigawatts, five% is 60 gigawatts. Those are the kinds of numbers we’re talking about if you want SMRs to be material.

We have a lot of experience with small reactors already. Early commercial plants like Yankee Rowe were relatively small by today’s standards. The Navy has been operating small reactors on ships and submarines for decades. So “small” isn’t new; what’s new is trying to do it economically at scale in the civilian market.

If I had to pick a category with an advantage, I’d say water-cooled, light-water SMRs—technologies that use LEU in fairly conventional fuel forms. There’s a lot of inherent labor, experience, and supply-chain capability around water-cooled reactors. People understand them, from operators to regulators to fabricators. That’s a big advantage.

Companies like Kairos and TerraPower have wonderful ideas with molten salts and advanced coolants. These concepts were tested back in the 1950s and 1960s at the national reactor testing stations, and now we’re applying new materials and methods to try to make them cheaper and more reliable. I hope they succeed.

But at the end of the day, economics will rule. Gas-cooled reactors have been tried—we had a commercial one at Fort St. Vrain in Colorado. It worked for about a dozen years but had issues, including gas leaks and complexity. Maybe the new players like X-energy and others will overcome those problems. I hope they do.

One more point: fuel. We’re also seeing a shift from traditional rods and bundles to TRISO fuel—coated fuel particles packed into pebbles or other forms. That’s very interesting, but not yet fully proven at commercial scale. You’re introducing new operational, construction, and fuel-supply risks. Will people be willing to finance that? That remains to be seen, and it will likely narrow the field to a few proven approaches.

Interestingly, I know of at least two startups that designed their reactors on paper, then went to the supply chain and asked, “Can you build this?” The answer was, “Sure—in five, six, seven, eight years after we qualify everything.” So they redesigned their reactors to fit what the supply chain can realistically deliver. That tells you something about how supply-chain reality will shape the winners.

Bryce: Let’s follow up on the fuel piece. Vistra operates Comanche Peak, a light-water reactor. You use low-enriched uranium in standard fuel bundles. That supply chain is well established.

Ray, can you walk us quickly through LEU, HALEU, TRISO, and the other varieties of fuel now being discussed for the SMR market? It sounds like if you’re saying “water moderated,” you’re also implicitly saying LEU as a fuel has an advantage over designs that need different fuel cocktails.

Rothrock: The world fleet today—roughly 460 reactors—operates on LEU: low-enriched uranium, typically about 3 to 5% enrichment, sometimes a bit higher. Natural uranium is about 0.7 percent U-235, so we have to enrich it. The US essentially invented enrichment during World War II, and we dominated it for decades.

That LEU supply chain is what feeds the existing reactors worldwide. It’s proven and well-understood. If you enrich higher, up to 19.9%—HALEU—you can make reactors physically smaller or achieve longer cycles. You don’t want to go beyond 20 percent because above that you’re into high-enriched uranium, with much tighter proliferation controls, and we have a strong policy against that in this country.

Getting from 5 percent to 19.9% is not cheap. It increases fuel cost, but it can shrink capital cost by enabling smaller cores and longer refueling intervals. Some SMR developers talk about four-, five-, or even seven-year fuel cycles.

However, refueling every 12 to 18 months in today’s light-water fleet isn’t such a bad thing. We know how to do it efficiently. Again, it comes back to economics: higher enrichment and more complex fuels might lower operating costs or capital costs in some dimensions but raise them in others. We will see.

TRISO fuel—tiny coated fuel particles—is another important development. It offers strong safety characteristics, but it requires a brand-new fuel manufacturing industry. You’re shifting from rods and bundles to fuel pebbles or compacts. That’s a huge change. So yes, designs that can use conventional LEU, conventional fuel fabrication, and existing supply chains have a meaningful advantage in terms of deployment risk and financing.

Bryce: I’m glad we have the nuclear engineer on this panel. I would not have given a concise answer to that. Let me turn to a few audience questions.

One asks: “Thoughts on thorium?”

Quick take, Ray—we’ve talked about all these other fuels. Thorium has long been discussed. Any thoughts?

Rothrock: We’ve been talking about thorium for decades. It has interesting properties, but the world has built a trillion-dollar infrastructure around uranium. To pivot that entire infrastructure to thorium would be a massive undertaking. I’m not saying it’s impossible, but I don’t see it as the near-term winner.

Bryce: Another question: “How many micro-nuclear plants are really operating and up and running in the US.?” If we counted the Navy, that would be several dozen, right?

Rothrock: Oh my gosh, yes—on the order of a hundred if you count all the ships and submarines. Then there are 20 or 30 TRIGA-type research reactors at universities and labs. So there are a lot of small reactors operating, just not in the commercial power sector.

Bryce: Another question: if the Navy can do it, why can’t we make that happen more broadly in the civilian market? You’ve talked about the enrichment part of that. The Navy has an integrated supply chain and they’re taken care of.

Rothrock: Right. The Navy’s reactors are fueled with much higher enrichment—on the order of 90 percent. They fuel a reactor for the life of the ship, because it’s very hard to change the reactor once it’s in the hull. You fuel it once and that’s it.

They also have very different operational requirements. If the ship has to move, you can’t wait three or four hours to fire up turbines. You have to go to full power within seconds. That’s why they use such high enrichment and very robust designs. It’s a completely different use case and policy framework.

Bryce: Jim, here’s a question for you: FERC is scrutinizing co-location deals right now. Do you think direct hyperscaler-to-plant contracts are good for the grid?

Burke: I’m thrilled to get that question, because co-location came up in the previous panel as well. Two things matter: speed to power, which is the primary objective for large customers, and system efficiency. There are two things we need to do.

First, we need to recognize the excess capacity that’s currently on the system for about 99 percent of the hours in a year. That’s today’s combined-cycle gas turbine fleet. Across the country, the combined-cycle fleet—regulated and competitive—runs at only about 50 to 55 percent utilization.

So when people say “we’re out of capacity,” I think that’s a misdirection. The assumption seems to be that to hook up a data center, someone has to go build a brand-new plant. That’s not where we are today.

By 2030 and beyond, we will need new plants. But in the interim, what needs to happen is we place data centers as close as possible to existing or new generation to minimize the amount of new transmission that has to be built. Put load and resource as close together as possible. If we do that, we’ll better utilize the existing combined-cycle fleet across the country. You could move average utilization from 50 percent to 70 percent and support on the order of 50 gigawatts of data-center load by 2030.

There’s one thing data centers have to do, and it’s important. In that one percent of the hours—the super peaks—they have to be willing to curtail some percentage of their load. Not 100 percent, but enough to relieve stress. They can slow compute, move workloads geographically, or use their backup generation. They’re bringing one-for-one backup generation already. Most of the time, that’s diesel or natural gas.

We’ve conflated “the grid occasionally gets tight” with “we can’t connect data centers.” The math does not support that. The lowest-cost way is better utilization and targeted curtailment, rather than assuming you must build a brand-new plant for every new data center immediately.

Bryce: So peak shaving and better utilization of existing capacity would be the better option for everyone?

Burke: From now until about 2030, yes. After that, if we actually see the load show up and persist, we’ll need new builds. I still think there’s some uncertainty. The hyperscalers we work with want the option to grow, but they don’t always want to make 20- or 30-year commitments to lock in a gigawatt at a very expensive price. Better utilization gives us time to see how demand really develops over the next two or three years.

We will need new builds, but we don’t need to stop connecting data centers today while we wait for new plants. That day is coming, but it’s not here yet.

Bryce: Bottom line, you’re saying we need better utilization of what we have.

Burke: 100%.

Bryce: At ARPA–E, Chris Wright, the Secretary of Energy, raised this very idea—that AI itself might be used to do better grid management and optimization for exactly this purpose.

We have about a minute and forty-five seconds left. Another audience question: “If you had to pick one country as a model for nuclear energy, which would it be? France, China, or another?”

Ray, one country for the US to follow—who would it be?

Rothrock: I’m not sure the US has to “follow” anyone. France actually followed us. They standardized around a small set of designs and built a lot of them. That’s the lesson: standardization and repetition work.

Bryce: So that’s a model: standardize around one or a few designs, build a bunch, and focus on repetition.

Jim?

Burke: I think people in the US would have to ask themselves how much state-owned, state-controlled, state-backed investment they really want. That’s the fundamental question.

Energy is one of those sectors where this question always comes up, just like in health care and other areas. I actually think the innovation and sharpness that our capital markets demand will make us the long-term winner.

In the short term, yes, you can push more projects through if you have strong central government backing like China does. But I don’t think that means they’re going to be the long-term “winner” in nuclear. I think this country can be.

Bryce: Blue-sky question to close. 2050 is too far away—I’ll be dead by then. So let’s say 2035. How much new nuclear capacity will be online in the US by 2035? I’m talking about new nuclear, not restarts at Palisades or Three Mile Island—new reactors. Ray?

Rothrock: If there are three reactors connected to the grid by then, that’ll be a great day. Three reactors of any size or type—I don’t care.

Bryce: Three reactors, so less than three gigawatts?

Rothrock: Three or four gigawatts, roughly.

Bryce: Jim?

Burke: And you’re excluding uprates—you’re talking about new units, either on new sites or new reactors on existing sites?

Bryce: Correct. New reactors.

Burke: I’d say certainly less than five gigawatts. I’m with Ray: probably in the two- to three-gigawatt range.

Bryce: I’ve said ten, so I’m going to stick with ten, because I’m an optimist.

Now, you all don’t know this, but we’re going to end with a song. Ray is actually a musician—a bass player—and he’s wearing his Reddy Kilowatt pin. If you don’t know Reddy Kilowatt, he was a cartoon character from the 1920s on, created when the industry was making more power than the public was using.

Utilities went on radio and then television with ads featuring Reddy to encourage women in particular to use more power in the home. Reddy was a cartoon character selling the promise of abundant electricity. So here we go:

“I wash and dry your clothes, play your radios,
I can heat your coffee pot.
I am always there, with lots of power to spare,
Because I’m Reddy… Kilowatt.
Remember, just plug in—I’m Reddy…”

[Audience laughter and applause.]

Jim, you can join in next time.

Ladies and gentlemen, please join me in thanking Jim Burke from Vistra and Ray Rothrock.

---

Before you go, please click that ♡ button, share, and subscribe. Thanks.