X-energy's Sam Levenback on high-temperature gas reactors, the Dow and Amazon partnerships, and rebuilding the US nuclear supply chain

Speaker 1: Up next. Sam Levenbeck from X Energy. He's the CEO growth and strategy. Sam, good to meet you. How are you doing?

Speaker 4: What's happening?

Speaker 11: Thanks for having me. Pleasure to be here.

Speaker 1: First time on the show. Please introduce yourself and, tell us about the company a little bit.

Speaker 11: Yeah. Absolutely. Name is Sam Levenbach. I work at a company called X Energy. Yeah. X Energy is designing a reactor. Mhmm. High temperature gas reactor. And what's really exciting there's a lot of exciting things about what we're doing. The technology is super compelling. We have some incredible investors. I think the thing that we're most, cited for and known for is we're building our first project in South Texas with Dow Chemical, in Calton County, Texas. And we're gonna be building our second project in Washington State with Amazon for to support, one of their largest data center clusters Interesting. In in the Pacific Northwest. So just super exciting times for, nuclear and, the next generation of firm power.

Speaker 1: Yeah. I mean, Dylan and our team mentioned that you are a new Erebor customer. And I'm wondering, like, this doesn't seem like something that's, like, you know, unbankable or, you know, requires something new. But what what is it about your business that drew you to Erebor? Is it just excitement, or is there something that where your business looks a little bit different than a normal business?

Speaker 11: I think that there is a really amazing nuclear industry in The United States today. We have 94 operating nuclear reactors, but not a lot of new reactors in construction.

Speaker 1: Yeah.

Speaker 11: Right? And we have a more abundant nuclear supply chain

Speaker 5: Mhmm.

Speaker 11: In in this country. And that's you know, it's it's it's a shame, that we've lost, a little bit of the shine and the pedigree of of, what we have historically done so well in The United States. And so getting that nuclear supply chain back up on its feet and up and running and to support the growth that's really demanded by AI digital infrastructure is a pretty heavy lift, and it's not something that's directly in the wheelhouse today, frankly, of a lot of existing pick your basket of of of capital allocators. I think what's been really impressive about the Erebor team about you know, with Owen and, you know, what they're building is a real excitement and willingness to roll up their sleeves and understand our sector, understand what is risky, understand what is bankable, and, you know, get to work with us really with the same mission, which is, you know, the next century of American leadership in global nuclear commerce.

Speaker 1: Yeah. How capital intensive is the business? Do you have to buy land to build a facility to what level of the stack?

Speaker 11: Yeah. So our business is primarily around two things. Number one is designing a Mhmm. Next generation high temperature gas cooled pebble bed reactor. So that's a heavy lift. Our company has over 900 employees today working on that. Wow. We are also then the secret sauce of our reactor is the fuel, so called Triso fuel. And this is really advanced stuff, and we're building a fuel factory today in Oak Ridge, Tennessee. We received last week from the NRC our permit to operate that facility. And so that's the two halves of our business. It's designing a reactor, which our customers will license that technology for us, and they'll they'll build it on their balance sheet, and we will sell them, the fuel for the life of that reactor. So we are selling you the design to build the printer, and then we will sell you the printer ink for the life of that reactor.

Speaker 1: That's

Speaker 11: good. So, you know, there are different aspects of that business that are capital intensive. There are different aspects that are not. And getting into it with us and and understanding those different aspects is really something that Air Force coming on that journey When

Speaker 4: did guys decide to get into the fuel business? Did that come after the reactor?

Speaker 11: That's a great question. You know, it was really a visionary decision by our founder back in 2015. If you're just if you build the world's most advanced reactor and the fuel doesn't show up, you're kind of hosed. Right? And so a little bit of vertical integration here, a little bit of getting into if it's the secret sauce of your reactor, you you you better have your arms firmly around it was really, you know, early insight. And so that decision was made back in 2015, and we spent, you know, a lot of time redeveloping how this fuel was made, the intellectual property around how you do it. And, you know, that's been a really, you know, important part of our story in our journey.

Speaker 1: Last question for me. I I wanna know about timelines. We talked to a lot of folks in nuclear. I hear 2029. I've seen some reporting in The Wall Street Journal about even just bringing old nuclear power plants back online, but the dates are crazy to me. The 2030, 2032, 2035 gets thrown around. What are you estimating for when we might actually see the, you know, the the famously flat curve of nuclear power generation in The United States start to tick upwards?

Speaker 11: It's gonna you're gonna see it in the early twenty thirties. In the early twenty thirties is when we're gonna see, our first two projects coming online with many more thereafter. You know, the first pickles out of the jar are the hardest, but the demand is insatiable.

Speaker 4: New new analogy.

Speaker 1: Because pickles are green like nuclear fuel.

Speaker 11: But there yeah. Yeah. You know, it's it's but you're gonna I I think that there's a reason that you're hearing. I mean, that's, I think, the kind of the consensus in the industry. You know, you gotta And so, I think you're seeing look, turning on those reactors and upgrading existing reactors, that's the that's the low hanging fruit and we should get every electron we can out of, the existing fleet. But that's not gonna be enough. There's only so many of those projects to do. The really throughout the twenty thirties, you're I think you're gonna see I know you're gonna see a steady and ever going growing clip of new nuclear reactors coming online and, powering this kind of generational super cycle of AI digital infrastructure.

Speaker 1: Yeah. How are you thinking about the different scales of nuclear power generation. I I was really excited about just replacing the one megawatt diesel reactor. Radiant's doing some cool stuff there. There's a couple other folks. Then, then there's been you know, as the AI boom has sort of taken off, everyone has been thinking about, well, what if we get a 100 of those together? Well, then we're doing something for a data center. Maybe we should upscale the whole project, do a 10 megawatt, a 25, a 100 megawatt, and then we can get into, like, the really big Westinghouse projects that maybe can be unstuck by a startup. But what's the landscape like from your perspective? Where's the most value in, in new reactor designs?

Speaker 11: Love this question. I mean, I I think that it it it it's all exciting and all have their puts and takes. Right? Yeah. So the big reactors, the the conventional light water reactors, the Westinghouse AP 1,000. Yeah. A lot of new excitement and focus on those. I I'm sure new ones will get built. I mean, just the amount of power demand that that that's out there. You know, Westinghouse is a great company. Yeah.

Speaker 1: But Votel, we all learned from Votel that it can take year we did turn on Votel, it took a long, long time. And I think the tech companies And that's

Speaker 12: what and that's the downside.

Speaker 11: Right? I think that even for, a big hyperscaler

Speaker 1: Yep.

Speaker 11: You know, even these companies spending a $160,000,000,000 a year on CapEx

Speaker 1: Yep.

Speaker 11: You know, a $15,000,000,000 bite on a single machine Yep.

Speaker 4: Is

Speaker 13: a a lot

Speaker 11: of risk to manage Yeah. And it's a lot of risk for any individual investor owned utility to manage. You know, if you look at the market caps of the largest investor owned utilities in The United States

Speaker 1: Mhmm.

Speaker 11: $15,000,000,000, $10,000,000,000, whatever the next AP 1,000 costs, these are big bites of the apple. So so those are the puts and the takes with those, but you know what? They're gonna be successful building more. I don't have any doubt about that. On the other end of the spectrum, the one mega one megawatt micros, look. If you can get those to scale, delivering tens and then hundreds at a time, it's an awesome opportunity.

Speaker 3: You know,

Speaker 11: the challenge there, and I think you alluded it alluded to it in the question is, you know, for it to make a meaningful dent in a hyperscale data center. You're building hundreds and thousands, you know, in a single location, and that's a lot of moving parts. That's a lot

Speaker 7: of Yeah.

Speaker 1: But but it doesn't it doesn't necessarily grid capacity is grid capacity. So if you put one at every hospital and you replace a diesel generator on every, you know, oil field, like, that does free up energy for other purposes, and that has Absolute. Equilibrates at some point. So

Speaker 11: Absolutely. And Yeah. You know, so, I mean, they're super exciting and, you know, the the the theory of the case getting to a factory kind of manufacturing, you know, throughput, you know, to bring down the cost

Speaker 1: Yeah.

Speaker 11: It totally makes sense. Yeah. You know, but it's it's, you know, it's it's nonetheless challenged on the economics on the actual delivered cost of electricity when you get that small. Our main product is in that middle sweet spot. So our a single one of our reactors is 80 megawatts electric.

Speaker 1: Oh, wow.

Speaker 11: That's pretty building Yeah. What Dow is building at our at at their first site is four reactors, so 320 megawatts.

Speaker 1: Wow. Okay. That's a lot. Yeah.

Speaker 11: That's a lot. Yeah. That's like a small coal plant, you know. That's that takes a real chunk out of, you know, what they're doing and that that more than takes care of the site that they are focused on.

Speaker 1: And then think that's I think that's the same amount of energy that's generated in my entire city that I live in. I live in a small city.

Speaker 7: It's a

Speaker 11: it's a it's a meaning yeah. It's a meaningful bite. You know, what Amazon and Energy Northwest, their second project, you know, they're talking about anywhere between four and twelve reactors on a single site. So anywhere between 320 megawatts in a gigawatt. What's really exciting about that theory of the case, our theory of the case, for these mission critical, deployments, you know, a a petrochemical Mhmm. Facility, it can't go down. Yeah. You need, you know, very high reliability.

Speaker 1: Yeah.

Speaker 11: So if you have four reactors, you know, if if if if you overbuild just a little bit, if you only need two or three reactors, you know, if one goes down, you're still golden. Yep. You know, if you build 12 reactors in Washington state, if you only need ten, one, even two go down, you're still completely utilizing that very expensive data center with that those very expensive, you know, NVIDIA chips, you know, running in it, you know, and getting the full use case out of it. So that kind of reliability of our form factor is something that's, you know, very exciting, you know, particularly to our customers.

Speaker 1: That's awesome. Well, thank you so much for taking the time. Yeah. Great to meet you. Have a great rest of your day.

Speaker 4: Congrats.

Speaker 1: Have a great weekend.

Speaker 4: All the 900 or a thousand a thousand people at the company Seriously. On a on all on all the progress.

Speaker 1: Yeah. We'll talk to you soon, Sam. Have good one. Goodbye. Let me tell you about AppLovin. Profitable advertising made easy with axon.ai. Get access to over 1,000,000,000 daily active users and grow your business today.

Speaker 4: Before we bring in our next guest, we gotta check-in with Tyler

Speaker 1: Oh, yes.

Speaker 4: On his project. Let's go over to Tyler. Do

Speaker 7: we Okay. So so I can read my

Speaker 1: Yes.

Speaker 4: My little essay.

Speaker 1: And you swear and you swear on your life that you did not use AI to generate this.

Speaker 7: Correct. Yeah. You can check you can check my my chat logs. Okay. Did not use any AI.

Speaker 4: Okay. For in, anybody that's just the challenge was write 500 words that Pangram will pick up as AI

Speaker 1: Yes. At

Speaker 4: 90%.

Speaker 1: Or above. Okay.

Speaker 14: And if you win Ready?

Speaker 4: Yes. You get a $50

Speaker 1: This is this is Tyler's impression of steak

Speaker 4: sandwiches. Okay.

Speaker 7: So I want you guys to also try to guess what what the percentage is.

Speaker 1: Okay.

Speaker 7: This isn't just an essay that is trying to fool an AI system. It's an experience where words and phrases flow over the compute like a waterfall. This short speech marks a pivotal shift towards the style of writing where the audience doesn't just consist of real people but machine learning algorithms too. It underscores the innovative change frequently cited in the New York Times and CNN who, along with many CEOs, industry professionals, and growth hackers, have noticed the world changing pattern emerge. It's a testament to the emerging influence of the AI industry. Additionally, it demonstrates the way in which artificial intelligence systems may fail may fall victim to possible bad actors, further enhancing the importance that some government institutions may play in the future. You are absolutely right

Speaker 1: Okay. Keith, let him keep going.

Speaker 7: You are absolutely right that AI systems might not always accurately predict whether text is written by humans and you're actually touching on something really important. It's a great insight. In the in the future, I I it's like pretty long

Speaker 4: but Okay. Okay. I'm I'm putting it at 90 I think 6%.

Speaker 1: I I think this is gonna come back at a full 100. Let's hit it.

Speaker 7: 100%.

Speaker 1: Let's go. He did it. I never lost. Faith, enjoy those cheese Cheese steak

Speaker 4: steak dinner. Absolutely massive. You guys don't you also at home, this cheese steaks doesn't sound that significant but these are the best cheese steaks sandwiches in the world over at Matu.

Speaker 1: Oh, yeah. Yeah. Read out the Em dashes. Did you use any Em dashes?

Speaker 7: Yeah. Yeah. I mean, there are Em dashes in what I just sent. Okay.

Speaker 1: But I did that. You put m dashes everywhere. That's fantastic. Let me tell you about MongoDB. What's the only thing faster than the AI market? Your business on MongoDB. Don't just build AI. Own the data platform that powers it.

Speaker 4: Yeah. Hill says you should have used the word Delve, big missed opportunity. Oh, so so Delve

Speaker 11: is under I think actually,

Speaker 7: like, PERN models don't use Delve anymore. I think that was just four o, which I guess this probably

Speaker 1: Yeah. Probably detects. Four. But yeah. Anyway, we have our last guest of the show, Alex Heath from Sources in person live in the TVPN UltraDome.