The Exchanges

Q with this conversation, Rick and I are basically saying, hold my beer, because we're going to talk about fusion. Um, so Rick, let's get into this. So fusion, obviously really interesting, different technology. Tell us what fusion is. Tell us whether it's actually 40 years away and always will be, as the adage goes, and most importantly, why are we talking about this now at a data center AI conference?

A Yeah, thanks, Laura. Um, and I guess maybe just to set the stage a little bit, like, fusion is different than fission, so at its most fundamental level, uh, fusion is the combination of small atomic nuclei together, which then releases a bunch of energy. Fission is the opposite. It's splitting apart a big atomic nuclei, uh, and that releases energy. Uh, fusion is essentially what you see in the night sky when you look up. Those are stars. Those are fusion power plants all over the sky. Um, what we're trying to do is to bring that energy source, the source of 99.99% of all the energy you've ever experienced. Solar, wind, fossil fuels are all basically derivatives of our local fusion power plant, the sun. We're trying to bring that down to earth and put it in a bottle and then get some heat and electricity out of it. Um, so it's, why is it, you know, why is it relevant now? Why, why are we talking about it now? Because there's been a lot of advancements In fusion, people maybe haven't paid as much attention to what fusion's doing, but over the last 5060 years, its progress on its main criteria, like how much energy you get out versus in, has been faster than Moore's Law. It had a long way to go, but it's been very fast, and the reason that's been happening more recently now is we have a lot more compute and understanding of how do you control a plasma, which is a big soup of char…

AI assessment note: “fusion is the combination of small atomic nuclei together, which then releases a bunch”

Q We're gonna get into all the details around that. First and foremost, though, you all are effectively a startup. I don't know if that's what you'd call yourselves, but you've raised over two billion dollars, which is not typical for climate tech startups. So how have you been able to secure that level of investment?

A Uh, well, I mean, certainly we're going after markets as in the trillions, um, so whether it's the power generation equipment that could be deployed everywhere in the world, or it's the electricity that comes from it, those are big markets, um, but I would say we're also have taken a very thoughtful approach to what is our approach to fusion, and we've decided we're gonna take the most scientifically proven architecture, we've talked more about it, it's basically a magnetic donut called a tokamak, and we apply this key innovation, Much stronger magnets that allow us to shrink that way down so it's actually a device that you could actually build and finance and become a power plant. So just the approach is one thing. I think the other thing where we've been able to raise that, that amount of money has really been relentless execution on the most critical milestones for us to prove that this approach works. From our series A, that critical milestone was build that very, very powerful magnet, which people said you cannot do. There's not enough material in the world to build it. No one's ever done it. It's not going to work. We said, we'll do it in three years. Shame on us. We did it in three years and two months. And by the way, that's in the midst of COVID and supply chain disruptions, that proof of that magnet. Three months later, we raised 1.8 billion series B. And the reason i…

AI assessment note: “Three months later, we raised 1.8 billion series B. And the reason is because”

Q earth shattering for you to show net energy gain. And you're talking about a very close timeline. So get real specific here. Like, how are you going to do it How sure are you that you're gonna do it? Is this still like, you know, there's fundamental science, but you're ironing it out? Is it an engineering problem? What makes you so confident you're gonna be able to get there?

A Yeah, so as, as, uh, Lara, as you rightly point out, there's two ends of the spectrum of how you achieve fusion, and basically, um, fusion, what you need to achieve fusion, and more energy out than in, is basically you need three conditions. You need to be hot enough, and when I say hot enough, it's like really hot, like a hundred, a hundred fifty million degrees centigrade, so hot. Um, you need the fuel to be dense enough, so enough for the fuel around, and then you need to contain it enough, to hold it together enough. Lasers do this by just pounding a pellet so fast as nothing That it can do otherwise than just implode and fuse, right? Uh, magnetic confinement on the other end of the spectrum, what we do, is you actually hold that plasma for longer with a very strong magnet. The magnet makes the spirals that the, the ion, the atomic nuclei do very tight, so it keeps it together, so it will continue to fuse. So, that's the, the differences. Why are we confident in this? Um, our architecture, this goes back to our original, like, why do we raise so much money? Our approach is based off of the most kind of scientifically proven approach To get to a, kind of, a commercially relevant version that will get net energy. And that is this tokamak. So that is a magnetic donut. There have been about a 150, a 160 of these things that have been built over the last 5060 years. They've marc…

AI assessment note: “Our approach is based off of the most kind of scientifically proven approach”

Q much fun. So, you mentioned that there's a magnet that you all effectively need to build. A lot of folks were skeptical that you could build this magnet, period. Not only did you build it, but you had to create a supply chain around it in order to be able to do the things you're doing. So how did you do that as a company, effectively as a startup company?

A Yeah, it's a great question. I mean, when we saw this material, like, this is great. We could build a magnet out of it. And then it's like, well, how much is available? Like, uh, you know, a few, you know, maybe a hundred meters of it. And for our device, we use thousands of kilometers of this stuff, basically. Um, so how do you build that up? Well, we work very closely with the producers of this tape. Um, I actually have some tape. Your listeners won't see it, but anyone who wants to see it, come up afterwards. Um, it looks like a cassette tape. It's basically mostly steel, and there's a one layer in there. Well, it's like basically one micron thick, uh, where you can send about 2000 amps across it with zero loss. Um, so amazing. Miracle material. Nobel Prize in 1986, but not produced in any volume until like the 2010. So we're only just starting to get this stuff. So how do we build up that supply chain? Well, We work very closely with the supply chain. We, we work with them to identify what are the holdups, um, that would prevent them from going to more scale. Originally, it was like, well, we're not producing it at the level or the quality that, you know, we think you need. So we actually, in some of our critical technologies, like understanding what is the quality of the tape, and therefore, where can we put it in our magnet so that we could actually take lower quality tap…

AI assessment note: “We work with them to identify what are the holdups... we help them increase their yield.”

Q which is so cool. Um, yes, and as you say, there are challenges, and, uh, and you described a great one for us. This is cool how you kind of solve this copper plate problem. Have there been other moments like that? Like, what are some of the other things that have popped up that have been major challenges in the context of building this first-of-a-kind facility that you're building?

A The obvious one is we can go down some technical routes, like, there, and, but I'm also going to go beyond technical, but for one technical one is, like, even just the architecture of the magnet that we're building, um, you know, no one had ever built the In large scale, the kind of magnet that we're building is non-insulated, non-twisted. It's basically layering the tape right next to each other. Um, when, it's interesting, when a, when a magnet is running at cryogenic temperatures, copper, like, actually becomes an insulator, like, because it's not superconducting. So everything's not superconducting, it becomes an insulator. So you can do things, like, you can make a metal thing that actually is an insulator for this magnet. But we came up with a, a new way of doing this in a large And there was a real question within the company. Do we do that? Or do we do what people had done with high temperature superconductors, which is layer tape in this cable, twist it, and then you have current running through the cable. There were reasons to do both. Like we, the, you know, the team wrote like a hundred page report on which one should we do long, hard technical arguments over the course of like months, but then it turned into like over the course of a week, like we've got to make the decision. This was a risky decision. This was a bet the company decision, but There was faith that w…

AI assessment note: “one technical one is, like, even just the architecture of the magnet that we're building”

Q So how should the folks in this room who are thinking about AI data centers think about fusion as a potential electricity source?

A Uh, should think about it, like, it won't solve your problem right now, we just talked about that, but it is the potential long-term solution that can power the AI transition, and it is, in essence, uh, You know, we didn't talk about all the benefits of fusion, but one of them is, is energy security. It is like the perpetual forever energy security source. If you have a fusion power plant, you own your energy. Uh, the fuel that goes into a fusion, there's a longer answer. The fuel that goes into your, your fusion power plant is basically the heavy water. It's like the, the hydrogen with one extra neutron deuterium. Uh, one out of roughly every 6000 hydrogens is like a deuterium. So it's basically filtered seawater. We make our other fuel, tritium, so to fuel a 400 megawatt power plant would use, like, gas canisters of deuterium that fill up less than half this room, and then you shut the door. It's fueled for life. There's no fuel input into the plant beyond that, uh, and so it, it looks very much like a natural gas plant, but no fuel coming in and no emissions coming out, and you, when you think about an island nation, like, Tamasic is one of our investors from Singapore, like, What does it mean to actually own your own energy, to not have to import energy? It, it's a, it's an interesting geopolitical question.

AI assessment note: “it won't solve your problem right now... but it is the potential long-term solution”

Q I know Bob himself is one of the most humble, collaborative, incredible humans on the planet, so I'm sure that helps. But also, is there something about, like, the technology moat that you all have? Like, what is it that makes What makes you so confident that you're able to do what so many startups don't and go out and be really obvious and open and collaborative in your approach?

A Well, I think it's, I mean, part of it's actually to like Bob and others in the, our founders, it's part of the DNA of the company in some sense. Um, you know, this, people join the company because it's like, it's, people say it's a moonshot. I say it's a star shot. Actually, it's, it's like, it's actually such an incredible vision of what this company could do. Uh, fusion writ large is, Like the ultimate power source. It, you know, we like to say it's not just the next energy solution. It's the final one. Like what, what's better than a fusion power plant, a better fusion power plant. And hopefully there's a lot of different architectures that will work. So there's that element to it. Like everyone's super motivated and excited about the mission. Um, what I would say, it's also the mindsets of the people in the company and working together very collaboratively solve problems. Like there's problems. Like one of your questions were like some of the challenges we've seen in building spark. Like, We see challenges, like, every day. Some of them turn into, like, company existential questions. Like, oh my god, for building this big magnet, how are we gonna solve this one? Like, there's one where, like, we had to put, like, a copper cap, uh, on top of the magnet, um, just to cover up the tape. And, uh, and we realized, like, unless you make that electrically homogenous, um, then if y…

AI assessment note: “Like one of your questions were like some of the challenges we've seen”