Helion CEO David Kirtley on why fusion finally works

David Kirtley is the co-founder and CEO of Helion Energy, the fusion company. Helion has raised more than $1.5 billion at a $15.5 billion valuation, and Sam Altman is its biggest investor and former chairman. Before fusion, David worked on rocket and space propulsion. In this episode of Summation, David and Auren discuss: Skipping the steam turbine and pulling electricity straight out of a starHow a Best Buy network card helped make fusion possibleRegulating a 150 million degree machine like ...

Speaker names marked ? are inferred from speaking time and the show’s known hosts — they are not voice-verified.

• 0:00Speaker 00

  You know, when you're building hardware technologies and you're innovating, so often I hear, well, that's the way we've always done it. I know we can do better. It's the one thing I know is that we're not doing it as good as we could.

• 0:11Auren Hoffman?

  My guest today is David Kirtley. David is the co-founder and CEO of Helion Energy, a fusion company.

• 0:17Speaker 00

  Let's optimize, let's develop, let's innovate each piece, each subsystem, so that you end up with fusion systems that generate more electricity at lower cost.

• 0:28Auren Hoffman?

  Hello, fellow data nerds. My guest today is David Kirtley. David is the co-founder and CEO of Helion Energy, a fusion company. Helion has raised more than $1.5 billion at a over $15 billion valuation, which is amazing. Before fusion, David worked on rocket and space propulsion. So he only tackles like the really easy things in life. David, welcome to Summation.

• 0:52Speaker 00

  Thank you very much for having me.

• 0:53Auren Hoffman?

  I'm really excited now. Okay, you're like essentially you're like recreating the reaction that powers the sun and building it in Washington state. Like, that kind of seems insane. Like, does it ever just seem insane to you what you're doing?

• 1:07Speaker 00

  I mean, it seems insane pretty much every day. It's a pretty wild job to be able to recreate what happens in stars every day. And so, yeah, so at Helion, that's what we do. We do fusion. We've built now seven generations of machine that have demonstrated our core technology and do fusion. And so I think that's pretty fabulous. It's exciting and it motivates the team too.

• 1:25Auren Hoffman?

  Your technology players, I think, hit like 150 million degrees. I don't even comprehend like what that is, but it is hotter than the center of the sun, right?

• 1:36Speaker 00

  Depending on how you count it, that's 10 times the temperature of the sun. And the key to that is magnets. The key to that is that no material on earth can survive that kind of temperature. And so what we do is we use high-power electromagnets, literally millions of amps of electrical current flowing through a large copper and aluminum magnets that generate a magnetic field that keeps that fusion fuel confined and off the walls.

• 2:00Auren Hoffman?

  For a layman, like, how does that work? Like, how does it like, how's it not like break out and just like melt everything or something?

• 2:08Speaker 00

  When you have temperatures that are that high, pressures that are that high, this is literally can be over 1,000 atmospheres, more pressure than the Marianas Trench in these systems. The universe doesn't like to stay contained. It wants to expand. And so a lot of our work is actually fighting that. It's like fighting that and harnessing the fundamental forces of the universe. That's one of the things that I tell our team all the time. So how we do that is we have large-scale electromagnets. Electrical current is flowing. When it's flowing, it generates a magnetic field. When you superheat a gas, it becomes a plasma. Literally, the nucleus of the proton and the electron, which is orbiting the proton, gets so hot, they vibrate, they rip off. And when that happens, you have charged particles, ions and electrons. And now those ions and electrons, when they're trapped in electromagnetic field,

• 2:53Speaker 00

  you can act on them. You can squeeze them until they get hotter and denser and fusion happens. You can expand them and extract electricity from them.

• 3:02Auren Hoffman?

  Was it like the magnet technology that didn't exist before? Why is this something that can happen today? But like, people have been talking about this for a really long time.

• 3:10Speaker 00

  Some of the earliest versions of our technology were theorized in the 1950s. And it's pretty wild to imagine what we do now, where we have tens of thousands of high-power semiconductors driving gigawatts of peak power, like just insane amounts of energy flowing around. And it was theorized before the transistor left the lab for the very first time. And what folks knew is that if you could generate electromagnetic fields of this intensity of the 15 Tesla and higher intensity, and you could do it quickly where you can compress and heat that fusion fuel, then like the math says fusion will work. You can get to the temperatures and densities, you can get fusion. You can actually even extract the electricity directly. A lot of that was theorized, but it took Moore's law, actually.

• 3:53Speaker 00

  It took modern semiconductors to be able to harness-

• 3:55Auren Hoffman?

  Just to get better and better and better, essentially.

• 3:58Speaker 00

  To get better and better by many orders of magnitude from where they were. And so now we literally have little transistors that are this big that can harness megawatts of peak power into the fusion reaction and then recover it at megawatts of scale. And that allows us now to build whole systems. There's some of the other key tricks is that fusion, when you superheat a gas, it moves fast. Like fundamentally, the gas particles, when you ionize them, they vibrate really quickly. But that means they also move around really quickly. They try to escape the confinement. And so you have to have systems that it can adjust really quickly. Literally, we measure our timelines in microseconds, millions of a second, or even sometimes nanoseconds. And so you now need electronics that are fast, that operate at gigahertz speeds.

• 4:40Speaker 00

  And that's something when, even when I first started my career, we didn't have commercially.

• 4:44Auren Hoffman?

  Because they're basically are like, have to turn on or turn off or pulse differently or whatever they're doing and have to react in a very, very quick way?

• 4:53Speaker 00

  Yeah, so we are, our system's sequence where they, we inject the fusion fuel. We have a series of electromagnetic coils that accelerate the fuel over a million miles an hour. It merges into the middle, stops, stagnates, convert all that kinetic energy to heat. And then we squeeze that fusion plasma to increase its pressure and temperature. And that whole process happens in under a thousandth of a second. And it has to be sequenced in under a millionth of a second per sequence and under, in sub-microsecond time scales.

• 5:21Auren Hoffman?

  You get this thing, it's super hot, everything is moving. And then somehow you're like, somehow you extract electricity from it? And then it like all dissipates or like, how does that last step work?

• 5:32Speaker 00

  It's actually the fundamental thesis to what we do at Helion is directly recovering the electricity.

• 5:37Auren Hoffman?

  Instead of it like, in some things like they're going to heat some steam that then moves the turbine or something, like you're directly getting the electricity instead?

• 5:45Speaker 00

  It's what most folks in fusion do is they do a fusion reaction. It increases heat. And in fact, we measure some of the scientific units of how much heat did you create? Not how much electricity did you create, which is an interesting challenge in fusion. But then you take that heat, you boil water with it, you run a steam turbine into a cooling tower, and you make electricity. You can do that. It's also pretty inefficient.

• 6:07Auren Hoffman?

  Why create heat when you could just somehow get the electricity directly, essentially? It's like, why do these like extra steps? And then that heat is moving the turbine, the turbine then gets electricity? Like, if you can somehow capture electricity, like clearly that's better, but it does seem like really hard to do, right?

• 6:21Speaker 00

  Yeah, you got it. You should come, come, come to Helion. I mean, that's exactly right, that because you already have, we talked about these charged particles, you have these electromagnetic fields, it's already all electricity. So why are you then taking all this electricity and then boiling water with it? Can you directly extract it? The good analogy is actually electric cars. Is an electric car, you charge your battery overnight, you store all that electricity, you hit the gas or you hit the accelerator. It takes that stored electricity, turns it into motion. And then using that motor itself, you can regeneratively recapture that or directly recapture that kinetic energy and recharge. It's actually one of the things that enables electric cars. It's because you can recharge all that unused energy, you can use it for the next thing. You can use that electricity to then accelerate and go further. Same thing in fusion.

• 7:06Speaker 00

  If you can recover all that input electricity and recover to high efficiency, now you don't have to do as much fusion. Fusion's the hard part. That like 10 times the temperature of the sun and 1,000 atmospheres is awesome, but it's also really hard. And so the more you can do less of that, minimize the fusion reaction, minimize the amount you have to do, the better. And then you also always have the option on top of that is if you have leftover heat, you can use that too. But your steam turbine systems now will be much smaller, can be more efficient, and you can go build those systems quicker. So you get all that. In fact, it's an, it's an active debate in Helion is how, what do we do with that heat? How much do we harness it? How much do we just cool it? And so there's some interesting engineering things you get to do too. But the key is that is you just recover as much.

• 7:49Speaker 00

  And in our systems, we've demonstrated 95% of the electricity you put in, you recover and recharge all that.

• 7:54Auren Hoffman?

  Is it like you're doing like lots of little reactions? You're doing a reaction, it's heating, you get the electricity and then you stop it. You could like, and then you could just keep doing them over and over. Or you could literally just stop it for an hour if you want and then do like, do another one? Like, is that what's happening? Whereas like, when I think of like a classic nuclear reactor, it's just something that's like always going?

• 8:15Speaker 00

  You got it exactly. The good, the good analogy there is in a piston engine, is you do what? You have one piston that compresses, does the combustion, expands, and then takes that energy and puts it to the next piston, which compresses. And then you alternate it back and forth. And in our systems, we take electricity, we store it not in batteries, but in something called a capacitor. It's much larger in volume, but allows as much higher instantaneous power.

• 8:38Auren Hoffman?

  Essentially, it's a battery. It's a store, a way of storing energy.

• 8:41Speaker 00

  Exactly. Store that energy and then very quickly discharge it into the high-power electromagnet, compresses the fusion fuel, expands it, recovers electricity of the capacitor. If you did your job, it doesn't just take it to 95%, it takes it a little bit higher. And that extra electricity is what you actually sell. And then you exhaust your gas, just like in a piston. And you put in fresh fuel and you pulse it again. Right now in Pilaris, our seventh-generation system, that operates at about every once every five minutes for a pulse rate. And in a power plant, a lot of the work we're doing is turning that up from every five minutes to once a second and then really re-reating this system. From the business case, it allows you to do some cool things where you can change the power output by changing the repetition rate, you can actually change the RPM, just

• 9:25Speaker 00

  like in an engine, and change the output power and match the loads.

• 9:29Auren Hoffman?

  Unlike a lot of these things, all these alternative energy producers that could potentially wreak havoc on the grid, like you could actually match the load on the grid potentially.

• 9:38Speaker 00

  We have a whole team that thinks about this a lot because as we are going to go build a power plant, and we're working on what we call Orion, our eighth generation system, to build a power plant for Microsoft, a lot of that is how do you actually interface with the grid? What does that look like? How do you turn on to the grid so you don't destabilize it? How do you turn off? And how can you support them?

• 9:58Auren Hoffman?

  A lot of these alternative energy producers, like, they have to produce energy all times, essentially. Like, in some ways, like, if something happens, I guess you could just stop, right? If you had to.

• 10:10Speaker 00

  Absolutely. And it's one of the powers of the fusion reaction itself, is that a fusion reaction is, we've talked about how hard it is, it's easy to turn off. And so we can do a compression, we can do an expansion, and then we can just not do the next one. And then that recharges our capacitors, and we can take that electricity, and at our leisure, take that extra electricity and go sell that.

• 10:31Auren Hoffman?

  I can imagine that there could be a point, assuming this works, where you're, there's points in time where you're producing way too much electricity than your customers need, or the grid needs, et cetera. Is, like, the next step, if it's successful, then, like, figuring out a better capacitor, like, a battery to, like, keep it all somewhere? Or you just, like, roll boulders up a hill and just keep them there for later?

• 10:56Speaker 00

  As you can operate these systems and dial the power up and down, you then, you can match the grid. You can, if you're producing too much more than the grid needs, you could go turn that down. Some of the wear and tear on the machines, the lifetime of them, is also a function of how you operate them and how often you operate them. So dialing down the power and going to a standby mode, that would be an option. Our goal, though, is to generate so much low-cost electricity that you would want to run it. And you'd maybe turn off something else on the grid if you could.

• 11:24Auren Hoffman?

  Maybe we'll start having all these batteries that are, like, taking the extra load when we need them. And that could be, like, a distributed system, it's like, my Tesla, just grabbing it when the cost is low, or cost is negative. Or it could be more industrial side batteries, or something like that.

• 11:41Speaker 00

  One of the unique things you get when you're not focused only on the steam turbine cycle and the cooling water cycle, is you're a lot less tied to where the systems are. And so now you can actually put these generators, fusion generators, in other sites where maybe that's where the battery storage is, maybe it's where the electric car charging is. And so now you can have more distributed power systems.

• 12:02Auren Hoffman?

  How big is this footprint of this thing?

• 12:04Speaker 00

  So we're building for Microsoft a 50-megawatt scale facility. We have a lot of work to do on it, but what we're projecting right now is, this can be in a several hundred thousand square foot building, so about 10 acres for the full facility. So this is still big industrial scale power, it's not going to be in your garage or in your basement. Fusion just doesn't scale that way, we don't believe. But it is now sightable, where you can actually be in a town, in a city, next to a data center, next to a factory.

• 12:33Auren Hoffman?

  There's a lot of hype and a lot of people who are negative about these, like, small modular reactors, these SMRs. Like, what are your thoughts about these things?

• 12:43Speaker 00

  The commercial aspect of small modular systems is absolutely a winner. That because you can make them, it's what we actually focus on, at a factory, literally build a factory, a giga factory of fusion generators, you can automate systems, it can come off the assembly line, which drives down cost, drives up reliability, up efficiency. Like, you can build and deploy generators that really win that game at scale. I think the challenge comes to the regulatory piece, is that when you have a small modular system, maybe a nuclear reactor, where you still have uranium or plutonium, you may have to still regulate it, just like the big one.

• 13:19Auren Hoffman?

  Yeah, also, like, I mean, if bigger, it's easier to put all your eggs in one basket and guard it. Like, these things need, like, a lot of physical security, they need a lot of other kind of things from a regulatory standpoint, that, like, okay, you might as well just put them all in one place, and have, if you're going to put all the barbed wire and all the guards around it, and stuff like that, I presume. And, you know, you might need, like, things so the drones don't attack it, and other types of stuff, too, right?

• 13:47Speaker 00

  They have a lot of work to do there. I think that's one of the beauties of fusion, is it doesn't have that. We don't have the uranium, plutonium, the nuclear materials. We can't, the systems literally cannot melt down. And so that's why we've been able to go through, working with the nuclear regulatory commissions, the regulatory processes at the federal level, that we got a law passed a few years ago, where now the states regulate fusion. And in fact, fusion is regulated like a particle accelerator in a hospital. So if you think about a proton beam therapy, or something along those lines at a hospital, that's the same regulatory framework that we use for our systems.

• 14:21Auren Hoffman?

  Is it like the NERC that does the normal, like, who does the old nuclear stuff and who does the new one?

• 14:26Speaker 00

  Nuclear power is regulated by the Nuclear Regulatory Commission, the NRC, in Washington DC. Each state is a little different. In Washington State, the Department of Health regulates Helion. And so we've been working with them on our systems for a number of years.

• 14:39Auren Hoffman?

  And it is a state body, not a federal body. And I should, obviously, some state bodies are going to be, like, quicker to move than other states, and so you can, you've got, like, 50 different ones to experiment with, or something like that.

• 14:52Speaker 00

  Yeah, it's a good question. Do you want 50 small regulators, or one big one? One thing we've been able to do at Washington is work, you get to work with them a lot, and actually bring them along, license, we've now licensed several machines through the Department of Health, educating both ways. They teach us how to be licensed and safely operate machines. And we teach them about fusion and how these systems work.

• 15:13Auren Hoffman?

  Are you going to Washington because it's the best place, where you found, like, the best regulator fit? Because I know Washington, like, there's a lot of hydro power, there's a lot of cheap power in Washington, of all the states, like, it seems like one of the best, you know, it seems like the cheapest power. I don't know that much about power, but I assume power's, like, a lot cheaper there than it is in California, or something. But so is that the reason why you started Washington? Is because, like, you had a good regulator you could work with?

• 15:41Speaker 00

  We started in Washington because it was the right mix of some of the high technology, the fusion physics people were here, as well as all the big industrial manufacturing.

• 15:50Auren Hoffman?

  Oh, you want the people near the site, essentially. Okay, that makes sense. Okay, obviously, there's a lot of smart people in the state of Washington.

• 15:57Speaker 00

  And then big industrial base. We're right next to Boeing's big manufacturing headquarters here. We're in Everett, Washington, actually right behind me.

• 16:04Auren Hoffman?

  I've toured the Boeing factory, it's freaking awesome.

• 16:07Speaker 00

  We're not the big building, we're a little building down the street, but we have now five buildings in our complex as we continue to build here. And so the workforce is here, folks that want to build and want to build fast, and so, and have the right mindset of building, you know, mass production at scale. And so that's enabled us to learn how to build quickly, and then, and then in parallel to that, you can build quickly and test new things, iterate, and deploy. So we've now built seven fusion machines that do fusion, Polaris being our latest, we announced earlier this year, exceeded 150 million degrees, and then operated on deuterium-tritium. It's the first fusion energy company to be able to do that.

• 16:42Auren Hoffman?

  Now, when I was in high school, I remember my physics teacher, and AP physics, like, getting me excited about fusion. And then when I took physics in college, I remember my professors and my TAs getting me excited about fusion. And it seems like something, like, it gets, like, the students super excited about, but just, like, was never a reality, or was always, like, 20 years away for the last 70, 80 years? Like, is it really just Moore's Law of why this is something that can be done today? Or is there some other factors that, like, make this? Or is it just, like, we can, we need a bunch, a lot of capital, and you just couldn't raise the capital? Or what are the big reasons why, like, this is something we can do today?

• 17:22Speaker 00

  I make the joke with my team that, you know, what they say in fusion is that fusion is 20 years away, and always will be. And I flipped that around, and I'm like, and what that means is that we're 20 years late, that the world needs clean and safe baseload power. Demand for that has always been there, and now we have even more demand than ever before. So there's a huge demand function for new, clean, safe technologies that we can't meet otherwise. We know that, you know, the lead time on natural gas generator turbines is, some people say, a decade out is the lead time now.

• 17:55Auren Hoffman?

  Yeah, and people are buying old, like, turbines from planes and putting them into the things right now, just because they need, they need something.

• 18:02Speaker 00

  So you have the demand side is there, and then the technology is finally there, too, is that we needed power electronics that could commute at high efficiencies, these large-scale pulse power systems. That didn't exist in the beginning of my career. We needed fiber optics that can turn on and off in a nanosecond, literally one billionth of a second. That's gigahertz fiber optics. Even when we started Helion, that wasn't a thing you could buy yet. But now you can go to Best Buy and get a 10-gigabit network card that can do this all day long, every day. And so the technology behind Moore's Law has really driven us to be in this place. There's a computational piece to it, too. We can now simulate a lot of these things, reducing the amount of hardware we need to

• 18:42Speaker 00

  build to go learn. And so that's been a big piece, I think, too. For us, one of the keys is that we actually have, in every one of these pulse power systems that we build, little teeny brains, little teeny high-speed modules that are always calculating and measuring how it's performing. So we get real-time feedback back into the operators, so we know how the system's operating, its health, how efficiently it's operating, all of those things. And, you know, I couldn't have dreamed that when we first started down this road.

• 19:10Auren Hoffman?

  I was reading, and you can correct me here, but, okay, when you worked with Microsoft, you put this, like, kind of aggressive timeline in the contract, and is that just a way of just aligning, like, how do you align, you've got this big team of people, you have a lot of capital needs, you have all these other things, is that just, like, aligning everyone, like, on this clear, hard-to-hit deadline, and, okay, you gotta hit it or else type of thing? Or what's the, was there, like, a strategic reason behind that?

• 19:40Speaker 00

  It is always, as you're running a business, the continual focus on the mission. And our, our mission is not to do fusion. Our mission is to deliver global scale, clean electricity.

• 19:52Auren Hoffman?

  Energy at a, at a certain price, essentially.

• 19:55Speaker 00

  Right. If, and what I say to the team is that if we are the first to build the world's first fusion power plant, and that's all we do, and that's where we stop, then we've totally failed. That it would be a great achievement, and we've still failed because our goal is to be building these systems that can deploy power at scale throughout the world and build them quickly. I have a background, and you mentioned in rocketry and space propulsion and plasma physics as well. And our focus can't be that. It can't be making heat. It can't be making rockets. It has to be on deploying electricity to the grid as soon as possible. So engaging customers, having those customers hold you accountable for delivering. I think that is a key, a key part of business and has been pretty catalyzing to be able to go through

• 20:38Speaker 00

  and, and get a PPA and then start moving. It also unlocks doors too. You know, as you go to regulate a fusion power plant and start citing things, having a customer, having a power agreement in place makes that, makes that go a lot smoother and a lot faster because it's not a theory exercise now. It's, I need to go build a, a system to, to answer my customers' needs. And having a customer like Microsoft doesn't hurt either because they can, they can help in that too.

• 21:02Auren Hoffman?

  If you think about like a SpaceX or Tesla, if you think of like the success of the company, there is definitely some innovations that they do, which are super amazing. But so much of their success is just like these continual 1% improvements. Either driving down costs or driving down accuracy or driving down, driving up reliability or, you know, whatever it might be. How much of your job is the innovations versus just these continual small improvements?

• 21:32Speaker 00

  That rapid iteration that SpaceX, Tesla and others now have shown that you, that you can do in hard technology is really the key. And it's one of the things that also like pivots us away from some of the traditional fusion business models, or even especially nuclear, where you go off and build one giant plant every 10 or 20 years. You just have to learn faster. And so be able to build a business that can build things and test things quickly. So you can build and iterate and learn. You know, I, I said, oh, one of the keys is that we even have these new semiconductors that are higher efficiency and higher power. But even over the course of the last five years, we've iterated on those semiconductors. Polaris, our seventh generation system, has actually several generations of semiconductors in it

• 22:16Speaker 00

  because we were iterating the technology as we were building the machine. And we talk about, oh, it's this, this thyrist or this SCR version. What version is, is it version three? Is it version four? Oh, version five is now getting produced. And we're going to fold that into the next design or the next pieces of equipment. And that's all across the board. That's on the, the computational side. That's on the hardware side. And of course, manufacturing. Can you, can you, the more you can automate manufacturing and get away from, you know, an army of hand soldering to automatic welding, like those are the kinds of things that we are pushing across the board. And you have to, you have to push them all in parallel.

• 22:51Auren Hoffman?

  And are these weird materials things that like, is 3D printing important? Do you have ceramics? Like what, what, what's happening in like the materials that are new and different?

• 23:01Speaker 00

  Many technology folks don't recognize, or, or if they do, they recognize it too late, is that the material science and material development drives so much forward progress. You see that at SpaceX too, even Tesla, new kinds of stainless steel and, and things. And how it drives it. So for us, some of the key, it's a good analogy you made because some of the key materials requirements and technologies are very similar to rockets. And a high power magnet, you have very high current, high electrical current. So high electrical resistivity, but also high thermal conductivity. You want to extract all the electric, electrical heating that happens in that at high speed. And it's high pressure. And so it really, we use some of the same copper technologies as the inside of the reaction vessel

• 23:44Speaker 00

  of a, of a, of a rocket engine for that exact same reason, for our copper magnets. And then high, high powered semiconductors, we're always innovating there and pushing on that, on that space. I mean, then infusion, the other really hard one is called the first wall. Is that we have this 150 million degrees. We do everything in our power to confine it with magnetic fields, so it never touches the wall. But even then, there's X-rays and some amount of neutrons and ionizing radiation that comes off and then heats that first wall. And so now you have this really tough material science challenge on that first wall. So we have teams working, we actually fund external research outside of Helion, universities and national labs to be able to, to push the forefront of material science even more than we can do internal to the company.

• 24:28Auren Hoffman?

  At a classic, old school nuclear reactor, the, the fuel is very highly regulated. It's almost like a national security risk if it gets out and stuff like that. Cause it, it has all these other dual uses and, and stuff like that. Walk us through like what kind of, how does the fuel work for you? Or do you think about it in the same way?

• 24:50Speaker 00

  So for the fuel itself, it's actually, it's actually very easy. In fact, I have this sitting on my desk. Probably won't even show on camera, but this is literally heavy water. This is deuterium oxide, not H2O, but D2O. And this is the deuterium. We buy it in gas format that we actually put in the machines. It's easy and it's safe and it's common. It's not the whole story on the fuel. In our systems, we use a deuterium and helium three fuel. That's our eventual fuel goals, where we use deuterium, that heavy water. And we also use an isotope of helium called helium three, which is lightweight helium. That's not very common. In fact, the nucleus is called a helium. It's what we named the company after.

• 25:28Auren Hoffman?

  And how do you get this H3? Like how does, is there like a, is there like a manufacturer in Pennsylvania that like makes all the H3 or something? Or where does that come from?

• 25:39Speaker 00

  There are manufacturers that take it, that make it. And we could talk about how they make it, but more interesting to me is that we make it. And so one of the, one of the, the sort of the hacks of fusion is that if you take two of these deuterons, these deuterium atoms, and then can fuse them. If you have a very high efficiency fusion system and you fuse them, half the time they fuse and they make a new material called helium three directly in the gas. And then in, the other thing they make is called a tritium, which is even heavier hydrogen. We were the first fusion company to show you, we could fuse tritium as well. But it over time decays into helium three. And so you end up with this surplus of helium three in the system. That part's the easy part.

• 26:21Speaker 00

  The hard part is actually now taking that isotope soup, that, that combination of hydrogens and helium and separating those out. And so we've now spent a number of years of building technology to demonstrate that. And we can actually now do that where we can separate out the different hydrogen species, the helium species, separate those out, purify them and put them back in the machine.

• 26:41Auren Hoffman?

  When you separate it out, is it like somewhat stable? Can you just like store it in a canister somewhere? Or is it like itself like very combustible and changeable?

• 26:52Speaker 00

  In my office right now, I don't have a helium three canister. Sometimes I do, but the, yeah, so, so helium three is, is stable and can be stored everywhere. The tritium though, that's a more traditional fusion fuel. We also use that in some applications. That one's not stable. It decays over 12 years and is radioactive. And so you have to be really careful and thoughtful about that. In fact, the tritium production, because when we do fusion, we make that tritium. That is actually the highest, the, the most serious regulated part of our business is the radioactive materials, that tritium itself. And so we've spent a lot of years learning how to handle it, how to license it, how to operate with it, how to fuse it as well, but also how to handle it safely. And so that one, we spent a lot, a lot of time and, and money on.

• 27:34Auren Hoffman?

  Are these things valuable on the market? Like are there people trying to steal this? Like they would be trying to steal plutonium or something? Like do you have like James Bond people like, you know, landing on your roof and going through the, the duct system to try to steal your stuff? Or you need like high physical security or?

• 27:53Speaker 00

  No visit from Tom Cruise yet in, in, at Helion, maybe one day. But I think there, there is a market for it. And in fact, we get asked all the time if we can sell the helium three we make. And there's a couple of startups trying to do this to make helium three in various ways. But it comes back to that mission question that you said, is that helium three is really interesting? And there's a market for it, but it's not like electricity. The world needs electricity and it needs massive amounts of it. And so we just dial, dial the, the team right down and say, after we've deployed many fusion generators, then I'll think about selling helium three, but not till then.

• 28:25Auren Hoffman?

  What are the second order effects of, let's say we're able to?

• 28:31Auren Hoffman?

  Increase energy production by 10x and decrease costs by 10x. Like, what happens? Like, how would society look differently? Or, you know, obviously 10x, or even if you take the next step, 100x in both directions. Like, what happens?

• 28:47Speaker 00

  Our ultimate goal for these systems is to drive down costs tremendously. Our goal is to get down to one cent per kilowatt hour.

• 28:55Auren Hoffman?

  What is it today?

• 28:56Speaker 00

  Depends on where you are. It's seven cents up to maybe 12 or 20 cents in some places.

• 29:01Auren Hoffman?

  Okay, so you're trying to basically drive it down 10x right now?

• 29:05Speaker 00

  That's the goal, yeah. And it's hard to do. We have a lot of work to do to get there, but that's the goal. And I think, you know, when we founded Helion, we used to talk about things like increasing the standard of living throughout the world, which is a great thing to go do. Desalination, so you can actually have clean water everywhere. But what we're seeing now is there's even more need, right? This is what enables that large-scale data centers for artificial intelligence. This is the power needs, like, we know, and the cost of AI. At least, some folks believe it will be the cost of the power. And so I think that's, for me, a lot of what's driving us right now to actually go meet that need.

• 29:39Auren Hoffman?

  Are there things like desalination and things like that? Like, we're just not doing because it's just too, it just takes too much energy? Or we're not producing enough steel, or, you know, whatever it is? Because it just takes too much energy to produce these things?

• 29:51Speaker 00

  Our other customer, besides Microsoft, is Nucor, the steel company.

• 29:55Auren Hoffman?

  Yeah, famous company.

• 29:57Speaker 00

  Yep, and we have a deal with Nucor to investigate how do we go build and deploy 500 megawatts of fusion power at one of their facilities? Because their technology is steel recycling, is that they use high-power electric arcs to melt steel at very high temperatures to allow you to recycle steel in a way you couldn't do otherwise and make steel that's very low carbon and low carbon emissions. And so you can make steel much cleaner and recycle steel in a way you can't otherwise without their technology, but they need power, a lot of power.

• 30:26Auren Hoffman?

  Yeah, that's one of their biggest costs is power.

• 30:29Speaker 00

  And so that our goal is now to be able to provide that power to recycle that steel, to allow that low-carbon steel, and do it in a way where the power itself is not just making more carbon.

• 30:38Auren Hoffman?

  You got a bunch of naysayers, like any new technology out there. Like, what do the naysayers just, like, not understand?

• 30:45Speaker 00

  The big key thing is that folks will say that, sure, you can do fusion in this way because you've demonstrated in a number of machines. But that efficiency piece, that actually recovering electricity, that you can't do that piece. You're going to end up doing other things. And so that's what we've gone out to go demonstrate time and time again. In fact, when we were initially invested, or our first Series A round was Mithril Capital. And so when Mithril invested, the first demo we did was can we build pulse power electronics at high efficiency? Can we recover 95% of the electricity we put in? And 95% sounds like a good number, but it's pretty wild in that most fusion folks talk in single-digit efficiencies.

• 31:27Speaker 00

  Round-trip efficiency of 5% is generally considered pretty good. And so 95% just blows that out of the water by orders of magnitude, by over an order of magnitude, you know, close to two. And so now you're the case where there's just that much less fusion to do. The other one, and it actually gets to something you said earlier, is around that self-sustaining reaction. Is that even in fusion, a lot of the concepts that I learned in school and some of our competitors are working on building these large-scale systems that are self-sustaining in what's called ignition, where now the fusion is happening, it's creating enough heat that it's now self-heating itself and creates a reaction that continues. It doesn't, like ours, just shut off every pulse and then we keep going, but it actually contains, it continues.

• 32:10Speaker 00

  And using our approach to fusion doesn't lend itself to doing ignition very easily. And the critics there are totally right, that, man, for Helion's approach, it's really hard to get to ignition. And what I would, my response is always the same, it's, I agree with you, except we don't do ignition.

• 32:25Auren Hoffman?

  Yeah, in some ways, you could say that that's a feature, not a bug, or something.

• 32:30Speaker 00

  As long as you can then recover electricity that you put into that system, that's the key.

• 32:35Auren Hoffman?

  Sam Altman, he put in, like, hundreds of millions of dollars, his biggest personal investment into you guys. I think he was the chairman until maybe six months ago or so. How did that relationship develop?

• 32:47Speaker 00

  It was pretty early in Helion where Sam was interested in all of the things, nuclear and fusion. And so I know talked with a lot of fusion companies and then came and visited Helion and came and visited and had literally a stack of fusion textbooks. And it was like, tell me why these are wrong. Tell me why your technology is better than what these books say. And we were able to, like I said, let me show you in the warehouse and let me show you the machine we built and how it works and why it works and why the business case closes. And so that ends up being, you know, one thing I've appreciated about working with Sam over these years is his focus is always on that on-scale, is always on deploying, you know, the large-scale deployments that actually impact and help humanity.

• 33:30Speaker 00

  And so us to be able to talk about, be able to mass-produce these systems, be able to have smaller systems that are easily deployable, that you could put on the back of a truck or at least maybe multiple trucks, but you can move it around and you can make them in a factory and then go deploy them, that's the kind of thing that resonated with Sam. And then we were part of Y Combinator. And so with Y Combinator, that's also where, you know, to be frankly, I learned to run a business not like a government lab, but run it like a startup. And that's, we took those lessons, applied it to the science and the engineering, and that's where we are now.

• 34:03Auren Hoffman?

  I'm used to talking to people working more bits and atoms and maybe they're hiring different types of people. Like, how is your management of your company different when you're managing people different than I would expect in a large software company, let's say Stripe or something like that? Or is it, like, basically the same?

• 34:25Speaker 00

  You know, there's some famous discussions around founder mode, around folks that are actively hands-on, involved in the business and leadership at Helion. And it has some pluses and minuses, but our focus really is on leaders that are also very hands-on and very technical, that are involved in the business, that are involved in the product. And so our science teams are led by scientists and our engineering teams are led by engineers. And our executive team is led by some of the technical co-founders at Helion.

• 34:53Auren Hoffman?

  That would be similar at a company like Stripe or Databricks or, you know, whatever it might be, right? Even though they're dealing with bits and atoms.

• 35:01Speaker 00

  And a lot of that same philosophy, we want to take and learn as much as we can from those, you know, high-performing, super-successful businesses and then apply that to the hardware pieces. And some of the things are just harder and slower of, you have large-scale supply chain challenges and a large manufacturing, we have about a half a million square feet in Everett here of total manufacturing space. And so you're now running a big, much more infrastructure of the business. But the more lessons you can learn, the better of how you move faster and how you lead in that way.

• 35:29Auren Hoffman?

  Cool. Two personal questions we ask all of our guests. The first one is how are you personally using AI on your day-to-day, either at work or in your personal life?

• 35:38Speaker 00

  This is super interesting. In fact, even in the last few weeks at Helion, this has evolved. So just for me, I mean, the more I've seen that you can automate and you can build independent AI systems and agents that are just running in the background and doing the life stuff that you normally would spend a lot of time doing. You know, it frees you up to be able to do the thing I love doing. It frees me up to go build.

• 36:04Auren Hoffman?

  What's an example, like, the life stuff that it's doing now for you?

• 36:08Speaker 00

  In our Malaga facility where we're building the power plant, I'm building a house out there also because we're going to go and have my family be able to spend time there. And so like building a house now, when you have AI, is like a completely different animal. You can take a picture of an empty room and furnish it and just literally like add to my cart and add the different items to my cart, like something simple like that, that used to just take days and days of discussion and thought, like, just gets done. Same thing around structural requirements and how you route electrical and some of those things just really just get solved on a personal way. And that's finding contractors. Oh, my goodness gracious, is that different? And that applies to the business, too.

• 36:53Speaker 00

  You can automate now your supply chain systems, how you buy things, how you find vendors, how do you interact with them? And those are pretty powerful. And we're starting to see the cusp of the science impact of actually taking the particle physics and the plasma physics and accelerating that, too.

• 37:09Auren Hoffman?

  Yeah, I bet you in material science, we're going to see a lot of innovations as well.

• 37:13Speaker 00

  And we're seeing now, we've seen even in the last few weeks at Helion, taking our old computational code bases that are five or 10 years old and taking and refactoring them to take advantage of modern GPUs and other things. Have talked, have seen, we've seen really high speed-ups of those codes already. And now we're starting to see code improvements on the physics side. That I'm really excited about, you know, my vision. What we do now is that our operators will run our fusion machines, hundreds or thousands of high-power pulses, measure all the data, and then our computational team will analyze that and simulate it and say, like, spend a week and say, this is how it performed over the last set of tests. We're getting to the place where now we can simulate it overnight.

• 37:56Speaker 00

  We run the machine during the day, we simulate it overnight, and then bam, we're informed the next day of running. And we see visions of how you can get to, like, I do a high-power pulse, I do fusion, I set a new world record for temperature or energy, and now I'm going to analyze, how can I do better the next shot? Tell me the next steps. And that's where we're, we're not real-time yet, but if we can get to minutes or hours instead of days and weeks, that's pretty transformative for our ability to build and operate and optimize these systems.

• 38:24Auren Hoffman?

  That's super cool. Last question we ask all of our guests, what is the conventional wisdom or advice you think is generally bad advice?

• 38:31Speaker 00

  Actually, one thing I have on my wall here is a picture of running of the bulls. Just because you've always done it that way doesn't mean it's not incredibly stupid.

• 38:42Speaker 00

  And, and obviously for your audience, right, their contrarianism and, and, and pushing it's the status quo is probably not something new, but you know, when you're building hardware technologies and you're innovating, so often I hear, well, that's the way we've already always done it, or that's the way, that's the way the sixth generation machine was built. Let's just do it that way on the seventh generation or the eighth generation. I'm like, I know we can do better. It's the one thing I know is that we're not doing it as good as we could. Let's optimize, let's develop, let's innovate each piece, each subsystem, so that you end up with fusion systems that generate more electricity at lower costs.

• 39:19Auren Hoffman?

  Is that like a more of a hiring challenge? You have to hire people that like are always fit that way? Or is it like a culture challenge or a management challenge? Like, how does one like have that as an ethos in a company? Or is it all the above?

• 39:30Speaker 00

  It's all the above. You have to have that as the ethos of the company. You have to reinforce it when you're hiring and making sure you're hiring for that. And then every day, it's easy to say, well, I've already done it that way. I'm just going to cut and paste. And the question should ask is what is the actual design requirement? What is the minimum viable product? And what are the opportunities we have to do better than we've ever done before?

• 39:50Auren Hoffman?

  When you're hiring, I assume like everyone's like, well, everyone's like, oh yeah, I'm open minded. I love being open minded. I'm always, you know, I'm always willing to change my mind if there's better evidence. Like, how does one actually hire for that?

• 40:01Speaker 00

  For me, it's, give me examples. Everyone has great philosophy, but give me the actual hard example of what you did. And then ask questions. Ask questions on that. One of the things that I focus on, on hiring is making sure that, that everyone should know something better than I do. And so let's go find that thing. Let's dig in, let's drill in. Tell me that thing that you are that world expert at, that you've done, that no one else has done. And then in that, I will see, did you innovate in that process? And did you do something truly new?

• 40:31Auren Hoffman?

  And hiring is, it is extremely disappointing when you talk to someone who like should clearly know more than you. Like they wrote their college thesis about, I don't know, medieval instruments or something. And then you ask them about that and you actually know more about medieval instruments than them? Like that is a huge like red flag, or, or they've forgotten it all, you know, because it's 15 years later or something like that. This has been amazing. This is, so by the way, what you're doing is so inspiring. I'm so excited about it and I so hope it is really successful. Thank you, David Curley, for joining us on Summation. By the way, I follow you on X, at D Curley. I hope the

• 41:12Auren Hoffman?

  rest of our audience follows you as well. This has been super amazing. Thank you again.

• 41:17Speaker 00

  I appreciate it. It's been fun. Thank you.

• 41:19Auren Hoffman?

  One more thing before we go. I read a blog called Summation. It's the same as this podcast. The blog is about non-obvious idea sharing on business, talent, data, longevity, and random contrarian takes. If you like the conversations on this show, you'll probably like the blog. It's free. New content comes out twice a month. You could subscribe at orin.substack.com. That's orin.substack.com.