Meta just unveiled the biggest-ever corporate deal for nuclear power. It’s a sprawling set of contracts for both existing plants and next-generation reactors that totals 6.6 gigawatts.
Just a few years ago, the conversation in the U.S. was about which nuclear plants were going to shut down next. Now, some of the world’s largest technology companies are trying to lock them up under long-term contracts, while building new ones.
But critics argue that parts of Meta’s deal don’t add new capacity fast enough — possibly pushing electricity prices even higher in an already-tight market.
And that concern is suddenly political. This week, President Trump said tech companies need to pay their own way when it comes to electricity, signaling just how central data centers are to the national debate over affordability.
This week, we have a breakdown of Meta’s nuclear push. We’ll look at what it means for power markets, how it compares to what the rest of the hyperscalers are doing, and whether this moment actually changes the future of advanced nuclear.
Credits: Co-hosted by Stephen Lacey, Jigar Shah, and Caroline Golin. Produced and edited by Stephen Lacey. Original music and engineering by Sean Marquand.
With resilience now a leading driver of grid investments, Latitude Media and The Ad Hoc Group are hosting the Power Resilience Forum in Houston, Texas on January 21-23, 2026. Utilities, regulators, innovators, and investors will all be in the room — talking about how to keep the grid running in this new era of heatwaves, wildfires, and storms. Register today here!
Transcript
Stephen Lacey: Did you guys ever make characters for yourself in the metaverse?
Caroline Golin: No. I don’t believe so.
Jigar Shah: Hell no. I stopped innovating in that whole universe in 2005. Well, whenever I joined LinkedIn, which I think maybe was like 2006. So as soon as that happened, I don’t think I did anything new ever again. It really served me well to be a curmudgeon.
Stephen Lacey: From Latitude Media, this is Open Circuit. Meta just unveiled the biggest ever corporate deal for nuclear power. It’s this sprawling set of contracts for both existing plants and next generation reactors. In total, Meta says it’s lining up 6.6 gigawatts of nuclear by 2035. And it’s a big change from just a few years ago when the conversation was about which nuclear plants were going to shut down next. Now some of the world’s largest tech companies are trying to lock them up and build new ones. But this isn’t a simple comeback story. Critics argue that parts of Meta’s deal don’t actually add up to new capacity fast enough and could push electricity prices even higher in an already tight market. And that concern is suddenly political. This week, President Trump said tech companies need to pay their own way when it comes to electricity, signaling just how central data centers are to the national debate over affordability.
So this week, a breakdown of Meta’s nuclear push, what it means for power markets, how it compares to what the rest of the hyperscalers are doing, and whether this moment actually changes the future of advanced nuclear.
Welcome to the show. I’m Stephen Lacey, the executive editor at Latitude Media. With me are my co-host as always. Jigar Shah is a clean energy investor. He’s co-managing partner of Multiplier, Hey Jigar.
Jigar Shah: Hello.
Stephen Lacey: Caroline Golin is the CEO of Envision Energy Advisors. She worked at Google for nearly a decade shaping global energy strategy. Hey, Caroline.
Caroline Golin: Good morning all.
Stephen Lacey: Before we get started, a reminder that you can now watch this show in full on YouTube. Make sure to subscribe to Latitude Media’s channel on YouTube for full episodes and clips. And of course, you can get the audio version, which many of our listeners are listening to right now anywhere you get your regular podcasts.
Okay. Before we jump in, I think we need to start with a piece of news this week about what President Trump said on Truth Social. In a statement with characteristically odd phrasing and capitalizations, he said he is now working with Microsoft and other tech companies to make sure that the power costs of data centers don’t get passed on to consumers. And right around Trump made that announcement, Microsoft VP Brad Smith released a blog post outlining the ways that the company is going to secure rate structures that don’t pass costs onto consumers among other things. And it was a multi-point plan. So I just want to talk about what this represents in this moment. Firstly, Jigar, what’s your read on this? It feels very much a reaction to the affordability question about the impact of data centers on rising prices. How do you read this announcement?
Jigar Shah: Well, I think that there should be an AI tool to help President Trump fact check his Truths. But maybe. I’d say Microsoft has been saying and doing this for a long time. They had positive water goals for a long time. Brad Smith was integral to getting the Constellation TMI contract signed when everyone was getting cold feed over at Microsoft. Melanie Nakagawa, I think has been doing an extraordinary job on her side and the sustainability side. So I think this is just a restatement of what Microsoft has long said that its policies are. So I don’t know that anything new was invented here. Now, if Meta came out with this, I would be shocked. And so we’ll see whether Meta says the same things. And our good friend Nate just left Tract and went over to Meta to run energy there. So he’s got great credentials and Tract, I think did a lot of great work.
I think the big thing that concerns me, which Caroline and I, like I had talked about on this podcast when she was a guest, is that the data center companies don’t often develop their own data centers. So they buy powered land. So for instance, I talked to the Jewel folks in Utah this week. They have all the water rights in the world from an alfalfa farm that that used to be before it turned into a data center. And so when I asked them about water, they said, “We have no intention of saving water and being water positive at this place.” But it’s 2.9 gigawatts of powered land. I’m pretty sure every one of the hyperscalers is bidding on that project. And if they win that project, I don’t know that they go back and redesign the entire data center to make it water positive to figure out … Remember that project is off grid. It’s running on gargantuan numbers of cat solar turbines and so I don’t know. If they were doing it all themselves, they probably would honor all those promises, but sometimes you go into a deal that the developer has already cut.
Stephen Lacey: Caroline, onto you about what’s new here. It does seem like Microsoft is saying, “Hey, we won’t negotiate overly favorable rates when we come in and plan a data center.” I don’t know if the rate structure stuff seems new to you. And then there seems to be an indication that they will attempt to bring their own capacity for new data center developments. What feels different?
Caroline Golin: Yeah. I have to agree with Jigar on this. This is all stuff Microsoft has been doing. It’s all stuff Google was doing. I don’t think we ever thought we needed to be explicit about it. In fact, quite the opposite. I think when we made an intention to engage in the community to mitigate rising residential customer costs, we said to ourselves, “Let’s not toot our own horn because then it will just look like a PR play and let’s just do this because it’s the right thing to do.” And so we didn’t make a big deal about it, but Microsoft has been doing these types of efforts. Google’s been doing these types of efforts for years. And I have to say, the concept that we have to be explicit about we’re going to go in and pay our fair share is actually a question of what are the regulators doing in that state?
Because I know for a fact, at least when I was leading this work at Google, and I know the colleagues that were leading this work at Microsoft. And even to degree to Meta and Amazon, I don’t ever remember us going in and saying, “All right, how can you give us a really good deal and make sure that residential customer costs go through the roof so that we can build a data center here?” That’s just not the posture any of us took. I think what often happens is you have a gun to your head to sign an ESA because you can’t build fast enough and you sign whatever the standard industrial tariff is and hope that in the next two years while you’re energizing your data center, you’re going to figure out a way to improve that standard tariff. And if embedded in that standard tariff is a regulatory structure that says residential customers are going to pay more proportionately than industrial customers are, then that’s the product of it.
And I think what you’re seeing now is that hyperscalers and data center developers in general are going to have to go in and say, “Please fundamentally redo your industrial tariff for us, which could cause rate cases, could cause special dockets so that we can be sure that the way you set up your regulatory paradigm doesn’t favor us.” And I can’t really say that that should be the responsibility of a customer. It should probably be the responsibility of a utility and a regulatory staff to do that preeminently.
Stephen Lacey: Well, to some degree, this sounds like just the way politics are done now under the Trump administration. So President Trump sees a problem, wants to look tough on it, someone within a company takes what they’re already doing, makes it look like it was President Trump’s idea, and then it’s a win for everybody. Does it feel like that’s what’s happening here?
Caroline Golin: I just can’t imagine that Microsoft was jumping up and down to be the center of the affordability question. I think they probably got swept up in this and they were like, “Yeah, well, this is what we’re doing and were asked to make that more explicit.” I don’t think anyone is jumping up and down to be the center of this right now.
Jigar Shah: Right. But the problem with this entire conversation is what happens, as Caroline suggested, when the regulator is lazy. In Maryland, Microsoft offered to pay for all of the transmission costs that were coming through. And the Maryland regulator said, “It takes two weeks for me just to approve the PJM transmission upgrade schedule and peanut butter it across all of the rate classes in Maryland. It takes me nine months to create a new docket and to do what you just suggested. So I’m going to be lazy and just do the thing that I’ve done for 20 years.” So we upgraded the transmission largely because of data centers in Frederick, Maryland, and the data center company didn’t pay for it. It was like everybody in the state of Maryland that paid for it. So like, I don’t know, are they going to go back to the state of Maryland and now say, “No, no, no. We actually need to go through the nine-month process and what if Governor Moore doesn’t want to do it?”
Caroline Golin: Yeah. And the nine-month process is another way of slowing down investment. So there are trade-offs there.
Stephen Lacey: Does it feel like you’re absolving responsibility a little bit though, Caroline? Like to say, “Oh, it’s really the regulators. It’s not fully our responsibility.”
Caroline Golin: No. Listen, there wasn’t a single … And I can only speak … Microsoft’s one of the better teams. Not as good as the Google team, but one of the better teams. And I know their leadership really well and we would go into dockets together. Microsoft and Google worked together hand in hand on regulatory intervention all over the country. And I never saw a posture, at least out of Microsoft, that was like, “We just don’t care.” I think the posture is, I don’t know how to go back and tell my leadership, we have to slow down building this data center for a year in order for regulators to redo a process so that we can ensure that our costs go up to build that data center. To be honest, that’s a hard conversation to have. It’s just a hard conversation to have.
What I think needs to happen, and I think it is happening, and you’re seeing this in a lot of what happened in Ohio last year, and throughout AEP’s territory in general forced this and Dominion for that matter, is the utilities are getting ahead of this. They’re getting ahead of it and saying, “If you want to connect here and you’re a large load over 200 megawatts, basically meaning you’re a data center, then you have to take a minimum contract. You have to take certain risk.” And that can be exploitive on the other side, and we can talk about that one day, but I think the utilities are getting ahead of this. And it is going to be the utility’s responsibility. And to Jigar’s point about powered land developers, once utilities figure out how to map their entire system and figure out how to offer different parcels of land based on speed or carbon or reliability, that whole powered land issue goes away. But the utilities are behind on this and … You know.
Jigar Shah: I wish there was a tool.
Caroline Golin: There probably is.
Jigar Shah: Maybe it started with AI that they could use to do this next week instead of the oil and gas industry using it every day, but the utility is not using it, but whatever.
Caroline Golin: And I think they’ll get there. I just think that it hockey sticked and no one was prepared. And thinking even about Meta’s deal in Louisiana last year that got a ton of flack for the natural gas plant, all of us were talking to Entergy and Entergy was saying the same thing to all of us, which is like, if you want to get online quickly, we’re going to build a natural gas plant and we’re going to rate base it. And it doesn’t matter what other types of technologies you want to do, this is what we’re going to do because that’s the process they knew and that’s what they felt comfortable with. Is that right or wrong? We can debate the morality of that.
Jigar Shah: No, no, no. I will say that is wrong to be clear.
Caroline Golin: But then what do you do?
Jigar Shah: I’m not debating any morality of it. The fact that they would put in the most expensive technology they could possibly do that increases rates the most possible and not do anything with modern technology, which is Entergy’s point of view in life, is wrong. Yes.
Caroline Golin: Okay. But what I’m saying is that is it right or wrong for the customer to then go, okay, I don’t know. How much expertise, if you’re asking the customer to develop and finance advanced technologies, if you’re asking them to develop and finance grid technologies and grid solutions, and then you’re asking them to develop and streamline regulatory changes. We’re seeding a lot of power to these customers. I took it. I ran with it. It was fun, but it’s a lot of power and a responsibility you’re placing on them.
Stephen Lacey: I just have one final question on this. I think everybody in corporate America is afraid of when they will get a target on their back. And while President Trump has been extraordinarily bullish on the expansion of data centers, it’s very clear that the administration is trying to react to affordability concerns and put some distance between him and data centers or look like he’s doing something about rising costs. If you think about a company like Google thinking about what is the administration going to say next, do you think other companies are going to follow Microsoft’s announcements? Is this a domino that is going to start to fall for all the other large tech companies in terms of messaging?
Caroline Golin: I think the issue that I see is that I don’t know what the accountability actually means in this conversation. To really reform what’s going on … And I think it’s very disingenuous for any governor to stand up and say, costs aren’t going to go up. Costs are going to go up. And I think it’s disingenuous to say that it’s not. I think the question is, to what degree are we going to empower residential and commercial customers against controlling those costs? To what degree are we going to reform markets and regulatory structures to ensure that as costs go up, long-term benefits also go up across the system and they can balance themselves out over the next couple of years and insulate that risk. My fear is that there’s a lot of platitudes about things that companies are already doing and have already been doing. Ben Townsend over at Google has had one of the strongest climate neutral water platforms for five years, and we still got grief over everything. And so I think a lot of this is stuff that the biggest names are already doing.
The question is, are all those third party developers going to do this? Are the financing houses that are standardizing the way we build data centers increasingly to a needle point, are they going to require these types of things? The more and more that third parties become a critical part in raising and deploying the capital of digital infrastructure, which they are at a clip as they should be, if they aren’t held to a certain standard of build, it becomes very, very difficult for these to have any types of consequences long term.
Jigar Shah: I think maybe just to pull a thread that you started, Caroline, earlier in the conversation, I think that the hyperscalers were not bragging about a lot of the stuff that they were doing and they were basically keeping a lot of it under the radar screen because they didn’t want to get accused of whatever, doing it just for the PR. But my prediction for this year is they’re going to spend a crap load of money. I don’t think it’s in response to Donald Trump.
Caroline Golin: Oh, I agree. Yeah.
Jigar Shah: I think it’s in response to the fact that they got their ass handed to them in Indiana. And so I think a lot of people are saying like, “We are tired of being misunderstood and we’re going to spend a billion dollars in marketing and comms to make sure that people understand what’s true and what’s not true.” Well,
Caroline Golin: Maybe not even just in marketing and comms. My hope is that a lot of these programs that the hyperscalers developed, which were good programs, many of which I developed and led at Google, they had a meager budget. I’m hoping that that budget gets juiced and it should. And the governors should require it. And I think that that’s a great thing. And I think that once we start actually having a real conversation about, well, what’s actually best for reducing residential and commercial costs? We’re going to lead to a technology solution that looks very different from what we’re deploying right now, and I’m hoping we get there. And if we have to get there through the altruistic whims of large tech companies, then let’s do it.
Stephen Lacey: All right. Let’s move on to nuclear. There’s so much happening in this space right now. Over the last couple of years, Microsoft, Google, Amazon, and Meta have collectively committed to around eight gigawatts of nuclear power. This is coming through a mix of strategies that range from refurbishing aging plants to restarting shuttered reactors, to backing an entirely new generation of technology. That marks a really sharp turn from where things stood not long ago when the US nuclear fleet was defined more by retirements than expansion. And now Meta has stepped up in a big way with a portfolio approach that puts a very large number on the table. The goal is to power Meta’s multi-gigawatt Prometheus cluster, which is the company’s answer to the OpenAI, Oracle, SoftBank, Stargate. So let’s start with Meta and then talk about where this fits into the bigger picture. Caroline, disentangle this for us. Meta says it’s lined up 6.6 gigawatts of nuclear, but that number bundles together a lot of very different things. So what are the components of that portfolio?
Caroline Golin: Yeah. So we’re basically looking at three approaches here. One of which is a restart and an upright with Vistra, which I would say is the safest and cleanest way of getting nuclear power on the grid. I’ll also say that these were plants that I think just a couple of years ago were slated for retirement and now because of Meta’s investment will likely go on towards expansion, which is a really cool aspect of this deal.
I think it’s been in the works for a while. It’s a 20 year PPA. It’s pretty cut and dry. It’s something that I think … Generally every one of the big tech companies is going to do. Everyone’s going to do an uprate. Everyone’s probably going to do the next part of this deal, which is an investment in SMR.
Two components, which I’m sure we’ll get into. I think one probably has more merit than the other. So we’ve got to deal with TerraPower, which, it’s been Bill Gates’ company and a middle of the road SMR company that has found its way through the DOE with fits and starts, but it looks like it’s gaining its feet and I think is bound to get its … It’s got its construction approval. And I think it’s about to get build approval from the NRC. And then you have another deal with Oklo who has been a very prominent player with the administration and has seemingly become the darling of the DOE despite having some real financial struggles last year. Both of them, all of it will service in PJM. The two SMR deals are designed strategically, I think, to one serve Prometheus, which I believe is the Oklo. And then TerraPower is a TBD deal, which is actually very similar to what we did with Kairos, which you’re led to believe that that’s about a co-location deal.
Where is Meta going to build new data centers and want to co-locate with an SMR? Whereas the Oklo deal and the Vistra deal are to service Ohio growth, which last I checked, Prometheus is, what would you say, two gigawatt, is that right?
Stephen Lacey: Yeah. Yep.
Caroline Golin: Two gigawatt. The interesting thing about the Prometheus build for me … And we can touch on this a little bit. Is that for such a long time, we thought the long pole in the tent was access to power and the speed of which we could access power in digital infrastructure. In many cases, the longer pole in the tent is how quickly you can build a shell and utilize that power. So what Meta has done in Ohio, instead of building a traditional shell, which takes two to four years, which means the ramp schedule takes three to six years for whatever you’ve contracted for, we’re going to build these crazy tents, which cuts down the build schedule to 12 to 24 months, which cuts the ramp schedule down from 18 months to three years. So where a utility was historically looking at a large data center build and saying, “Okay, you’re going to grow into that load over seven years, which can be a major frustrating factor of servicing a data center.” Now they’re looking at this going, “You may really grow into this load over the next two to three years,” which has massive implications for how quickly can you build the transmission, which this is an area of Ohio, which suffers a lot from transmission capacity and what are the solutions that we’ll need both today and in the six years when you’re at full speed?
So we should get into the details and I’m sure Jigar has a lot of opinions on all of it, but this is a portfolio deal that, in my opinion, services three things. One, the reality that we want to keep building AP 1000s, we want to continue to build out the existing nuclear fleet that we have. And so Meta’s doing their part on that. Two, they’re looking for a co-location play with an SMR partner, and that’s where TerraPower comes in. And three, they are responding to political pressures and working with the administration, and that’s where Oklo comes in.
Stephen Lacey: Jigar, what’s your read? How big of a deal is this?
Jigar Shah: Well, I think just Caroline first, I feel seen, so thank you.
Caroline Golin: You’re welcome.
Jigar Shah: Look, I think the notion that Vistra or anybody else would say that this is additional load or additional power plants, that these power plants were going to get shut down is ridiculous, right?
Stephen Lacey: Yeah. Why is that? Explain.
Jigar Shah: Explain. Remember that these plants were going to get shut down when we had rampant overcapacity in the 2018, 2019 timeframe. Ever since the Ukraine conflict, we haven’t had that issue, right? Wholesale power prices are high enough that these nuclear power plants are in the money, but on top of that, we passed a bunch of tax credits in the Inflation Reduction Act, which were maintained in the OBBBA that keeps all of these plants online. So these are ridiculously profitable plants, right? Just to be crystal clear, no one was shutting these plants down and if they were, it’s because they were incompetent. And so the notion that you’re now going to go back to them and say, “I’m going to take credit for the fact that my 20-year contract is keeping your plant open.” No, not buying it.
Now the real question is that you do have 10 gigawatts or so of uprates that we can do to existing power plants. And what that means is that the technology today is way better than the technology that was installed like 15, 20, 25 years ago. So basically you’re in a place where you can get 10, 12, 14% more power out of the same footprint by upgrading some of the components. A few of those uprates are in the money at five cents a kilowatt-hour. Those are all slated to get done and they have been doing those since 2010. Many of them don’t pencil at five cents a kilowatt-hour and need more like 10 or 11 cents a kilowatt-hour. Those are very important to get wrapped up into a deal with someone like Meta. So like Meta signs a contract at seven cents a kilowatt-hour instead of five cents a kilowatt-hour, right? That extra two cents a kilowatt-hour is being paid on all kilowatt-hours and can then allow Vistra to afford an uprate that’s seemingly out of the money across the entire contract. So kudos to Meta to doing that because I don’t think those uprates would have happened, but for the 20-year contract.
Caroline Golin: No.
Jigar Shah: So they should get full credit for the uprates, but I don’t think they should get credit for keeping the plants running because we did a lot of stuff in the IRA and the OBBBA to do that.
The next piece of this now is what do you do to support new reactors. So we went through an extraordinary struggle to build the Vogtle nuclear plants in Georgia and train 10,000 plus people on how to build those plants. Obviously we didn’t have the next plants operating, so all 10,000 of those people are three sheets to the wind, like building data centers, making a lot of money, doing whatever it is they do, and they are no longer accessible to anybody who wants to build a new plant.
So now the question becomes, which of the reactor companies do you support. I think people don’t really understand that reactor companies don’t build anything. So like X-energy is not good at building anything, TerraPower’s terrible at building stuff, Oklo’s not building anything. Westinghouse famously went bankrupt because they’re not good at building anything. So all of these companies are technology companies. They design nuclear plants and they go through a very daunting NRC process, et cetera, to get approved. And the only two people who’ve gotten approved are New Scale and the AP 1000. No one else is approved in the United States, not TerraPower, not Oklo, not Radiant, not Aloe, not like … Pick a number. X-energy. None of these people have been approved. Right now they have filed documents. Hopefully they will be approved soon. GE Hitachi, BWRX‑300, not approved in the United States. They’re only approved in Canada for the Ontario Power Group. So Holtec 300, they announced that they got a $400 million grant, not approved in the United States. They’re going through that process now.
So I just want to make sure that we’re crystal clear on what these reactor companies are. They are not construction agents. They have no idea how to finance their reactors. They don’t know how to do anything except design a cool thing and get it through the NRC. And so when Meta says, “I don’t know how to do anything, and so I’m going to find a partner to do something,” and then they say, “My partner is Oklo TerraPower.” That gives me no confidence whatsoever. Now, if they had said that we’re working with Bechtel, maybe. For instance, the Holtec 300, they have a global partnership with Hyundai. Hyundai is one of the most experienced builders of nuclear power plants in the world. That gives me some comfort that Hyundai has actually decided to step up and wants to support building those reactors in Michigan.
So I think part of my problem with all of this stuff in nuclear is that no one knows the ABCs of nuclear. They have no idea what the difference between a large reactor, an SMR and a microreactor are. I just think that we’re in this weird spot where everybody is like, “President Trump has come to save us and we are going to build a bunch of nuclear reactors.” And then they say the dumbest thing I’ve ever heard out of their mouth. And I’m like, “Okay, you’re not serious.”
Stephen Lacey: Caroline, what do you make of the risks here? Those are some interesting points. And on top of what Jigar said, you also have a lack of a workforce, like the skilled engineers to help execute these plants. So what are the big risks along with the lack of permitting and approvals and the lack of development expertise?
Caroline Golin: So I think I’d step back and say, there’s an iterative process here that’s going on. Which is if you can show the administration that you have a deal, then I think the administration will work with you on getting you past other humps in the process, permitting, workforce, construction. And that’s probably how this is going to play out.
Stephen Lacey: Not much time left to do that though.
Caroline Golin: Not much time, but I think the way the NRC is evolving … And I know there’s been a lot of commentary on the NRC. As far as I can tell, it’s lost about 10% of its workforce, a lot of really great people, some of which I knew. But it’s just getting its footing on leadership. It’s just getting its footing on its role. I think the biggest risk I see … And this would go for any nuclear deal done, doesn’t matter if it’s Metas or anyone. Is that the biggest risk I see moving forward is that it’s unclear to me what is going to be systematically the same in the way we learned and developed nuclear or didn’t for the last several decades and systematically different working with the NRC moving forward and what risk comes along with that. A lot of this is politics, but let’s not pretend for one second that every single industry in this country, energy or not, hasn’t played politics to advance its business goals. There are less implications with the politics of getting solar online than there are with getting nuclear online.
Jigar Shah: Can I unpack that?
Caroline Golin: And I think we just have to step back and say that. I’m saying that if you are a technology that is optimizing your political fervor to get yourself a deal, to get pressure to get a deal, to move forward on trying to build something new, great. That’s happened time and time again in this country. If we don’t have the same protocols and we don’t have the same discipline around building nuclear, there are bigger ramifications for that industry and also for the environment and for people if it’s really messed up. And so I think the question here … And I’m not saying it’s going to go one way or the other, but I think what we have to recognize is we’re in an administration that wants to see deals done. That’s what they want to see.
They care less about policy or institutional consistency. And could that bleed into some concerns around safety and around environmental impacts? And that’s something that everyone needs to be thinking about. First and foremost, the utilities who desperately want to see this stuff happen. Because you have one big failed plant and that could undermine the entire industry. And this is an industry that I want to see and many of us want to see grow. And so that is the macro risk here with anything, with any of these deals. I think the bigger risk for Meta on the SMR side is that it just … The worst case scenario is there’s some real failures at it and you have environmental impacts and that you would hopefully avoid, or you’ve just thrown a lot of money at something that never gets built and you could have used that capital towards investing in something that could have been built.
Stephen Lacey: Sounds like the Metaverse.
Caroline Golin: That’s a risk. That’s a risk. Yeah. But that’s the risk. I think I see less risks clearly with Vistra, but with the SMR deal, those are risks. But that’s what happens when you are trying to spur a new technology, but that is the financial risk. I think on a … And I know we’ll get into the commentary around this. Meta’s trying to do what it’s been told to do, which is you need to invest in the solutions of the future so we don’t run out of power. So they’re investing in the solutions of the future and then they’re getting raked over the coals because they didn’t invest in capacity that would show up tomorrow. I think they are trying to invest in capacity that would show up tomorrow, and that’s why they’re working with a lot of midstream gas developers in Ohio right now, but they’ll likely get raked over the coals for investing in that capacity as well. So I think we have to be … I don’t want to attack Meta for saying you invested in an SMR that maybe I wouldn’t have invested in. They’re doing something. It’s their money. They’re not using rate payers money and they’re not using taxpayer’s money and that was the administration’s entire agenda.
Stephen Lacey: For sure. But I think it’s fair to point out that the risks and the limited time window for executing these kinds of political deals as well, but point taken. Jigar, you sounded like you wanted to jump in there a bit ago on Caroline’s point.
Jigar Shah: Look, I hear you, Caroline. I totally agree on the politics side. Oklo is a political animal if nothing else. The thing that bothers me is as somebody whose job it was to actually get into the weeds on 15 different reactor designs, and you and I talked about many of those 15 different reactor designs, the two that we were most confident would never work were Oklo and TerraPower, right?
Caroline Golin: Yeah.
Jigar Shah: TerraPower, because Bill Gates does not have their back and has never had their back. And the Rocky Mountain deal was flimsy at best. Rocky Mountain basically said, “If you build it and you take all the cost over on risk, we’ll pay this price for it at the end of the process. And if that’s 40 cents on the dollar, then suck it. We’ll pay 40 cents on the dollar.” That was the deal that Rocky Mountain Power agreed to. And Bill Gates was like, “Well, I’m not going to take that risk.” So clearly he showed his hand. Because Oklo’s never taken the NRC process seriously. So they fill out paperwork, they do stuff, they do this, they do that.
NRC took a historic step of actually rejecting them because generally speaking, you just pull your application when you know that that’s going to happen, and Oklo wanted them to do that. They have now figured out how to be a political darling, and they’re trying desperately to get their friend Chris Wright, to actually just give them a pass through the Department of Energy’s abilities to do a public good reactor on … Which is really for test reactors at National Labs. It’s not for commercial use reactors. And so you’re in this weird spot where the two actors that are easily crossed off the list … You would easily pick Radiant over Oklo. You’d easily pick Kairos over Oklo. You’d easily pick Aloe over Oklo. But no, let’s pick Oklo.
And then on TerraPower, you would easily pick X-energy where Dow has already made a commitment and Amazon’s already made a commitment. You’d easily pick the GE Hitachi, BWRX‑300 because they’ve already got an order for four of them in Ontario Power Group’s territory. You’d easily pick Holtec 300 because they’ve already chosen Hyundai. And so you know that they actually have a proven group that could build the reactor. So what am I supposed to glean from the fact that they picked the two people that I know for sure are going to fail?
Caroline Golin: Well, I can only speak from when we did our due diligence and we ended up picking a company as to SMR company to invest and it was Kairos for a couple of factors. The first was that we really were confident in their strategy in construction and deconstruction and the learning of building nuclear power. And we don’t have time to go into all of Kairos approach, but their modular approach and the way that by the time they had really built their first reactor, they had built it about 12 times. Their investment in supply chain and the culture of the people there, that goes a long way if you’re going to be in business with people for a long time. And their commitment to the community. I don’t know how many times I went to Tennessee with Jeff Lyash and we stood there in front of the community and talked about the benefits of SMRs and they got the community on board.
Stephen Lacey: And Jeff is the smartest guy on the planet.
Caroline Golin: Those were the reasons why we went with Kairos.
Jigar Shah: That used to run the Tennessee Valley Authority.
Caroline Golin: Oh yeah. Jeff Lyash, who used to run the Tennessee Valley Authority. Great guy. That’s the reason why we went with Kairos amongst many other things. I can’t say if Meta did that same due diligence. I can just say that I think that Meta made their decision for one of three reasons. Cost, political pressure, or flexibility in ownership. Those are, at the end of the day, are probably going to be the parts of this deal, parts of every deal that impact this. You have to remember when Google did the Kairos deal, it was way before all this political pressure came in. We were working on that because of our twenty-four seven commitment. We were working on that because we had said in 2020, we will invest in nuclear and we had to figure out how the hell we were going to do that. So we weren’t coming at it from a prism of extreme political pressure, extreme capacity crunch pressure. In fact, the first reason why we did Kairos was completely divorced from any capacity need. It wasn’t until our second iteration with them did we figure out, “Oh, we should scale this to a portfolio and think about co-location.” That wasn’t the initial goal.
So we were coming at it strictly from what technology do we think … Very similar to probably the way DOE was coming at it. What technology do we think is the most bankable, bettable, best for the grid, right? And culture came into that and the way that they handled the community came into that. Now, we have an insane amount of new pressures coming at us to invest in nuclear, not to mention utility pressure. AEP wants to do nuclear more than any other utility in the country. And so these are additional variables that Google didn’t have to deal with in the beginning when we first set up our team and started doing the due diligence and started thinking about this. And I cannot imagine that making that decision for the first time is any easier than it was.
And I don’t think … And this isn’t a knock on anyone. I don’t think that Meta internally has the workforce that maybe Microsoft … Microsoft was the first to invest and hire a bunch of nuclear experts that maybe Microsoft or Google invested in. And so they are relying on more of external pressure. If they fail, it’s their money and they failed. Jigar, the DOE didn’t have to give the money. And I know that there’s issues with that. And I know that there’s issues with like, well, they could spend their money on other things. Yes, they could, but there’s reasons why.
Jigar Shah: I don’t care about the money. I actually want to win. I actually want the United States to be able to do big things again. I want us to be able to build nuclear power. And like picking two reactors that you know are going to fail is not helpful in getting us to that level. And you know and I know that AAP does not want to build TerraPower Oklo. AP wants to build AP 1000s. the President of the United States said in his executive order that I want 10 AP 1000s. If they were really playing ball, they would have just said, “We signed a deal with AAP to build two AP 1000s and we’re going to be the full off-take agreement on those.”
Stephen Lacey: I want to get to the question of how we build more AP-1000s, but I first just want to tackle one last thing related to the meta deal, which was the criticism from people like Jesse Jenkins of Princeton University, who I think was one of the most vocal critics who basically said, “Look, a huge part of this is contracting for existing nuclear power and this is problematic because you bring in city scale demand, but don’t add a bunch of new supply. Prices are going to rise in a tight market like PJM.” What do you make of that critique? Is it a fair characterization?
Jigar Shah: Look, I love Jesse, I do, but his academic background doesn’t lead him to comment on commercial deals. I find that he often will substitute his academic modeling work for commercial deals, and I find that problematic. I think if we’re talking about an academic-
Stephen Lacey: Well, it makes sense, right? Like that you lock up that supply, you’re going to need to bring in more capacity. It’s likely going to be gas generation.
Jigar Shah: So I think it’s important to note to what Caroline said last time, the deal that they signed with Vistra is not the same deal that Amazon put forward with Talen. So like Vistra is going to sell that capacity into the market and then like Meta’s going to buy it from the market, which is the right way to do it. Now, you can say academically that maybe what we need to raise the bar to is that instead of only 15% of this capacity being new, which is the uprates, that 100% of all the data center capacity over time needs to be incrementally new. And so then they should sign contracts with Dana Guernsey at Voltus to do some BYOC contracts. They should sign a contract with this group over here, with Rewiring America to replace a bunch of people’s old air conditioning systems with heat pumps. Whatever it is that you want to say academically, we can have an academic conversation.
But I think that like one of my problems in this moment is that I don’t understand exactly what he is leading folks to believe. Because I think, his new for-profit company, Firma, is saying the capacity is going to come from batteries and then the net incremental energy is going to come from new solar and wind farms, right, assuming. So those will be 24 by seven … No, they won’t be matched. There’ll be 365 days a year like netted out, and then there’ll be capacity I think that comes from batteries. And then there’s a bunch of power flow modeling to show that this is possible and whatever.
But the thing that bothers me about the criticisms here is that it doesn’t lead to a solution to the fact that people who work and live in the PJM are paying $16 billion a year in new capacity payments. And that part of it’s divorced, this criticism is divorced from like a foolproof plan that Mikey Sheryl and Abigail Spanberger can put forward. The governors of New Jersey and Virginia to cut the cost of the capacity auction in half.
Stephen Lacey: Yeah. I guess, Caroline, I want to get your reaction, but like essentially what he’s saying is Meta needs to bring more new capacity to what they’re doing here and if they don’t, inevitably prices are going to rise. What do you think about what Jigar said and just like that basic argument?
Caroline Golin: I think that everyone should be following Jigar’s line of thinking and trying to develop a stronger, more decentralized portfolio of capacity solutions. I think Meta’s investment here is thinking long term and it’s a much more portfolio play for what they believe will be more than a two gigawatt need. I think some of the wording is probably their comms department could have been maybe a little clearer on it. I think they are working on near term capacity. I think they’re just not promoting it because they’re working with Williams to lay new pipeline. They’re not going to promote the fact that they’re going to build onsite natural gas because then no one’s going to like that. So what they’re going to promote are the things that they’re really proud of and think people will like and think will be good for the market. I don’t disagree at all with what Jigar is saying. I think the problem is that these companies are making decisions in a very short timeline with very limited options. I don’t know anyone out there who says, “I can get you two gigs of clean energy with the transmission capacity availability into Ohio, which is a major issue in three years.” I don’t know anyone who can do that right now.
Jigar Shah: I got a guy.
Caroline Golin: You got a guy. Well, I remember looking at our Ohio growth strategy and a lot of the problem was transmission capacity more so than it was being able to access generation in PJM, which even lends more credence to what Jigar is talking about, which is like, great, improve transmission capacity by putting more distributed opportunities on the distribution system. Absolutely. And I’ve been on that train for a very long time. But I think we just have to say, you can rake Meta over the coals for this, but you should probably look at everything they’re doing before you do that. And then if you want to rake them over the coals for building onsite gas, okay, go ahead and do that. But I think it’s a little disingenuous approach. This was not an investment to say, this is going to show up on day one and this is going to service everything we’re doing with Prometheus. This is an investment in the future. This is an investment in fungibility for where we’re going to co-locate potentially with SMRs. And we’re hoping that it takes care of future growth.
I just think it’s tackling a different problem. And I tend to agree, we have a lot of academic arguments and part of all of our sustainability goals and our platforms around these sustainability goals is they’re all academic. There’s not a single sustainability goal out there, including Google’s that was actually tied to dispatch. It’s tied to our own load profile, which is not dispatch. So I think we take a lot of academic arguments and we superimpose them on what is a fairly stringent process to build power in this country right now. And there’s not a lot of people who know how to build power in this country and there’s no one who wants to build power in a merchant setting. So you can’t just plug it into a computer program and spit it out in a term sheet. It just doesn’t happen that way.
Jigar Shah: That’s next week. That’s the next week.
Caroline Golin: That’s next week. We’ll figure that out.
Jigar Shah: OpenAI.
Caroline Golin: Yeah.
Stephen Lacey: I want to tie this in a little bow for our viewers and listeners and just put this in the context of everything that’s happening in the industry right now. So all the hyperscalers are circling nuclear. Microsoft came out of the gate with the Constellation partnership to restart Three Mile Island. Then Amazon made ways by acquiring the Talen data center campus in Pennsylvania. We talked about that last week and the kerfuffle it started about nuclear and behind the meter resources. Google had this multi-pronged approach, this partnership with Kairos you mentioned, and later the restart of the Dwayne Arnold nuclear plant in Iowa. I’m just trying to figure out … And now all of a sudden this large meta series of deals. I want to understand what this means on balance. Are these partnerships really accelerating timelines faster than they otherwise would have? What kind of market signal are we seeing that is truly moving the market?
Jigar Shah: I think it’s important to note that there’s a disagreement. And I just want to make sure that we’re just crystal clear. For those people who have worked in nuclear power for decades, the only confident way of moving forward are AP 1000s and lightwater reactors. So that’s the Holtec 300, that’s a GE Hitachi, BWRX 300. Like simple, straightforward, built it before. We know how to provide regular uranium, not HALEU, like high assay, like enriched uranium, which we don’t make yet, but we’re hoping to. And so I just think it’s important for people who are serious, you basically just build a bunch more lightwater reactors. Then you’ve got a bunch of people in mostly Silicon Valley, but also Seattle who basically are like, “We want to do advanced reactors. We want to do cool stuff.” And like the TerraPower reactor, for instance, with sodium, remember this is liquid sodium. This is not like simple stuff to play with. Never been successfully deployed in the history of the world for an extended period of time. Do I think that the TerraPower people are amazing? Yes. Do I think that their stuff is like well engineered and well thought through? Yes. But do I want to set up the entire strategy that the Biden administration put forward on tripling our entire nuclear fleet on this technology? Hells no. No. Of course not. Right?
Caroline Golin: Well, then you should call Trump and tell him to start investing in nuclear himself.
Jigar Shah: Lord Almighty, he just merged True Social into a fusion company girl.
Stephen Lacey: Yes, that’s true.
Caroline Golin: That’s true. We didn’t even talk about that.
Jigar Shah: The brother’s been doing a lot. You can’t expect him to do everything.
Caroline Golin: And we also didn’t talk about fusion at all. Google has a fusion deal too.
Jigar Shah: I know. I know. And so does Microsoft. But I want to make sure that we’re just crystal clear on the fact that if you’re serious and you want to copy China, which is building 30 nuclear reactors right now, you build lightwater reactors, whether it’s the AP 1400 from Korea or the 1000 megawatt reactor from like Westinghouse. Then you’ve got the SMR play, which are generally lightwater reactors and those will get done. Holtec is on track, I think, to getting theirs done with like Hyundai globally and they’ll go public sometime this year. They’ve already publicly announced that and GE Hitachi has got their four reactors getting built in Ontario. Everything else is an advanced reactor. And while I think we should be investing heavily and learning more about it, USNC is building their one megawatt reactor at the University of Illinois campus. I think it’s awesome. All of this stuff is awesome. But if we need serious power and we want to build at the rate that China’s building, just like 30 reactors under construction right now. There’s one technologies set that works and everything else is like a 2035 consideration.
Caroline Golin: Okay. There is so much more complexity to that. And okay, let’s start with the fact that the reason why there’s reticence to build AP 1000 is because there is only two companies in the country who have the balance sheet to take on that risk, Southern Company being one of them, Southern Company who saw its stock depressed for many years because of its choice to continue and build Vogtle and is reticent to do that again. And the other one at this point, potentially being Westinghouse, and we’ll see. We’ll have to have a whole nother conversation about that.
Jigar Shah: Oh, about the Brookfield two-step?
Caroline Golin: Yes. Yes. But the cost to build AP 1000 is so high. It is a question of competency around where you’re going to place your capital. We had this conversation briefly in the concept of flexibility. If you are a Google, a Meta, an Oracle … Look, I named them.
Jigar Shah: Oh, that is never going to happen. I don’t think Oracle’s going to do anything interesting on these principles that Trump has demanded they’re going to do.
Caroline Golin: But if you’re any of these, the highest and best use of your capital is going to be to deploy chips to train them to win the AI race. And there is not a single rational actor who would say otherwise. The second highest and best use of your capital is to get rid of operational hurdles. One of them being energy, fine. But it is not, and I don’t think any hyperscaler thinks this, their responsibility to fix or to ignite the nuclear industry in this country. They are happy to be a premium paying partner. They are happy to do their fair share in capital deployment. One, because Jigar’s right, they think it’s cool and they think it should happen. They understand the physics and they understand the climate benefits and they’re behind it but also because they see the future of, do we lay a bunch of pipelines? Do we bet on a massive change in the regulatory structure where all of a sudden distributed energy gets accredited and aggregated and flexibility becomes a thing and we figure out how to co-locate and we figure out how to do hybrid technologies? You could bet on that. And we should. And it is our responsibility as the entire industry to give growth that option.
But if you are looking in 2024, that is an unclear option. So your options are, I’m laying pipeline and I’m investing in natural gas or I’m building an alternative so I’m no longer choosing between one or two options anymore. But I’m not doing it at the cost of my balance sheet and I’m not doing it because I think I should be the lender of last resort. So if the administration wants deals done, we’ll pay more, we’ll pay twice what we could pay on the wholesale market for a deal, but we’re not going to be the lender of last resort. And that’s the reason why it is going to be hard to get a lot of momentum around AP 1000s. I do not disagree in terms of the technology perspective, but from a financial perspective, that’s the reason why it’s not going to happen.
So then you go to SMRs and the thing you can do with SMRs is you can place them where you want them. And we go back to my comment last week around why did Google acquire Intersect? It’s because they want fungibility across their portfolio of where to put power. They no longer want transmission capacity, they no longer want market signals or regulatory structures that says, actually you can’t have capacity value for this, even though we know it’s generating when you need it to be in their way on why they can’t build a data center. That’s the play for the SMRs more than anything. And we need all of it. And eventually if we get fusion, we don’t need anything.
Stephen Lacey: Well, we’ve been dancing around this question of the Trump administration’s policies on AP 1000s. And last fall, they made this announcement centered on Westinghouse to support the construction of $80 billion and a handful of AP 1000s. And it looked like the federal government stepping in to jumpstart the development of these reactors, but the deal was really murky and hard to understand and non-committal. Can you guys summarize like what exactly the administration announced, what they were attempting to do and why it raised more questions than answers?
Caroline Golin: Go ahead, Jigar. I’ll let you take this one.
Jigar Shah: I can tell you what I think, but I have talked to over 20 people about this because I’m deeply involved in the V.C. Summer restart and nobody knows to be clear. Nobody knows, so like everyone-
Stephen Lacey: So the plan was to build 10 new reactors, that’s what they said they wanted to do.
Jigar Shah: Well, so that’s the executive order. So the executive order is 10 new reactors. Then you have the blatant blackmailing of the Japanese and the Koreans by the US government. So as part of that, the Koreans are putting $200 billion for real, like I talked to the guys into the United States. So they’ve got an entire investment group that they’ve set up in DC to invest in projects and United States for that commitment. And the Japanese have said the same. Then Lutnick comes out and says … The commerce secretary. That that money is like free money. And then the Japanese company is like, “No, no, no, that’s not free money. That’s market rate money.” So now like the commerce secretary has bet his entire future on getting this $85 billion fund to work, and no one knows how it’s going to work. It’s largely anchored by the Japanese, Brookfield would manage. Brookfield, on the other hand, is basically like, we just want to IPO Westinghouse. We want to get a deal done. We want to show that this thing is going to happen, and we want to IPO Westinghouse and make a crap load of money because we see where Oklo’s market cap is. And so Brookfield is just pushing to get V.C. Summer announced.
Now, they’re doing a bunch of work to figure out what it’s going to cost to restart V.C. Summer. They have no idea to be clear. So they’re going to think about … Hopefully they’ll know by the summer. That then becomes a real transaction that they then send to Lutnik and Lutnik says, “Okay. I’m going to get this deal done.” If he doesn’t get the deal done, we’ll see. But when I ask, I’m like, “Well, when there’s a cost overrun, who eats it?” The Japanese are like, “We’re not eating it.” So I think that that means the US Treasury is eating it, but I don’t exactly know under what rule the US Treasury could eat something. So the whole thing is very confusing to me, but I generally think the Brookfield people are smart. I generally think that the Westinghouse people are whatever, somewhere in the middle trying to get smart. And then I think the Camaco people are the most courageous and virtuous people I’ve ever met, and they own the other 50% of Westinghouse, and they don’t want to be a part of anything that looks like shenanigans, right? And so popcorn has been popped and I’m trying to figure this out.
Stephen Lacey: What are the red flags for you, Caroline? And are the hyperscalers getting pushed into being off takers for these plants?
Caroline Golin: Yeah. Yeah. But the hyperscalers are being pushed into being off takers of everything, literally every technology.
Stephen Lacey: So what are the red flags of this Westinghouse announcement?
Caroline Golin: I think the red flag is that no one really understands who’s in charge of pre FID, post FID, cost overrun. Is this going to be a straight ring-fenced asset to hyperscaler with an astronomical PPA? Are they somehow going to build a ton of transmission capacity out of these government lands so that they can get it to some market somewhere? Is this something where the administration is banking on if we build it, they will come, or if we build it, we will make sure they go there? And I think the bigger issue is that, to your point earlier, this is a 10-year process, seven at best probably. And what’s the administration’s plan in three years, two and a half years? I think there are a ton of red flags. I will also say, and I’ll probably say this a million times over this year, I don’t necessarily agree with the methods of this administration, but shit’s getting done. So I think we have to live in that paradigm and work with maybe an undisciplined and unruly, but fervent desire to do something.
And I guarantee every hyperscaler is getting pressured into off take. Some of them probably want it, but the rest of us … I say us. I’m not part of this group anymore. But the rest of them are probably scratching their heads on the same issues, which is like, well water, transmission, workforce, these are all issues on government land.
Stephen Lacey: Well, let me close out with the most generic question that I think a lot of people watching this have, which is, are we truly in the middle of a nuclear revival? What do you think?
Caroline Golin: Yeah.
Stephen Lacey: How strong?
Caroline Golin: I think if we see two of these SMR companies generate electrons, then we’re off to the races. I think if they all fail, then we will be so far behind China that we won’t attempt to catch up.
Stephen Lacey: Jigar, what do you think? How much?
Jigar Shah: Yeah. No. We’re definitely in the middle of a nuclear renaissance and it’s definitely going to happen for precisely the reason that Caroline just outlined. But I think that the part that I struggle with is how long we have to deal with all of the noise. It’s very clear that we’re going to use light water reactors, that these advanced reactors are not really going to be the answer for a long time. My sense is that you’ve got three players. You’ve got the AP 1000s with Westinghouse and Caroline described very accurately what the challenges are there. Then you’ve got the GE Hitachi one. We didn’t go through their challenges, but their challenges is that the reactor that OPG wants to build is not the actual final reactor that GE Hitachi wants to sell, which is exactly what we shouldn’t have done at Vogtel. I’m not sure why we’re doing it again with GE Hitachi, but they’re smarter than me, I guess.
And then the Holtec 300. And the only reason the Holtec 300 is in the running is because Kris Singh is by far the most successful nuclear entrepreneur in history and his company is going public this year and he’s 80 and he’s like, “Screw it. I’m going to bet my entire fortune on getting this done.” And he’s only working with tier one people like Hyundai. So he’s the only person who actually has half a brain cell to do this properly and the guts to see it through. I don’t exactly know where this is going to go and who’s going to be the winner, but it is most certainly not going to be an advanced reactor that overtakes the light water reactor during this next 10 years.
Stephen Lacey: Fascinating times. There’s so much going on in this space. I pledged to our audience that we are not just going to be a data center podcast, but my God, there’s so much interesting stuff happening. I learned a bunch. Thanks for clarifying some of the details of these deals. And Caroline Golin and Jigar Shah are my co-host. This was great. Caroline, good to see you.
Caroline Golin: Same. Same. Always.
Stephen Lacey: Jigar, we’ll see you next week.
Jigar Shah: My office will hopefully be painted and look better by next week, but we’ll see.
Caroline Golin: Rainbows and fairies and unicorns and stars.
Jigar Shah: I will do whatever it takes.
Stephen Lacey: Yeah. And we’re going to see slow incremental improvement with every episode.
Jigar Shah: Exactly.
Stephen Lacey: I look forward to seeing the construction project in the background.
Jigar Shah: Books I haven’t read. All sorts of things.
Stephen Lacey: I am Stephen Lacey, your co-host and executive editor. The show is edited by me, Sean Marquand and Anne Bailey. Of course, go to YouTube and subscribe to our channel for fresh episodes of Open Circuit. And of course, you can find the audio version anywhere you get your podcast plus transcripts and every episode are available at latitudemedia.com. And there you can subscribe to our newsletters, including our AI Energy Nexus newsletter, which runs through a lot of the stuff that we are talking about today. Thanks a lot for being here. We’ll catch you next week.
