David Kirtley - Lex Fridman 知识图谱

Lex Fridman (00:13:31)

So in fusion, you take these lightweight isotopes like hydrogen and deuterium, and as you combine them and get these molecules closer and closer together, some really interesting fundamental physics happens. First, these atomic nuclei are charged. They have an electric charge, and like charges repel. I think everybody is familiar with that, where you take two positive charges and you try to push them together, and the electromagnetic force between them repels them. So you have a force that’s actually pushing against them. In fusion, you work to get your fuel very hot, very, very high temperatures—100 million degree temperatures. And temperature really is kinetic energy; it’s motion, it’s velocity.

Lex Fridman (00:14:14)

So these particles are moving so fast that even though they’re coming together and there’s this repulsive electromagnetic force, they can still come close enough that another force comes into play, which is the strong force. Once you get within a very close distance, on the order of the scale of those nuclei themselves—of those atomic nuclei, the tiniest thing you could imagine, and probably way smaller than that—these particles then are attracted to each other, and they combine and fuse together. At that point, you create heavier atomic nuclei that have a slightly less mass, slightly less total mass in the system, and that mass equals MC² as energy.

Lex Fridman (00:14:55)

So extremely high temperature, extremely high speed. Maybe that’s one of the other differences also with fusion and fission, is just the amount of temperature required for the reactions. Is that accurate to say?

David Kirtley (00:15:07)

Yeah, and I think fundamentally, in a lot of ways, fusion is hard and fission is easy. Nuclear fission happens at room temperature. This uranium and plutonium is so likely to break apart already that simply the adding of one of these neutrons, one extra particle, will then break it apart and release energy. If you have a lot of them together, it will create a chain reaction. Fusion, that doesn’t happen at all. Fusion is actually really hard to do. You have to overcome those electromagnetic forces to have a single fusion reaction happen. So it takes things like, in our sun, we have what is called gravitational confinement, where the gravity, literally the mass of the fuel itself, is pulling to the center of the sun and it’s pulling.

David Kirtley (00:15:55)

There’s a large force that’s pulling all that fuel together and holding it and confining it together such that it gets close enough and hot enough for long enough that fusion happens.

Lex Fridman (00:16:07)

And then we have to figure out, if we’re building fusion generators, how to do that confinement without the huge size gravity of the sun.

David Kirtley (00:16:18)

That’s right. Obviously, the sun is vastly larger than Earth, so we can’t do that same process here on Earth.

Lex Fridman (00:16:25)

Yet. No, I’m just kidding. All right.

Lex Fridman (00:16:27)

But we have other forces we get to use. We can use the electromagnetic force, which the sun doesn’t get to do, to apply those forces. I actually want to take a pause right there and point out a word. Historically, we’ve used the word reactor around fusion, but I don’t think that’s right. For me, we’re really careful about this terminology.

David Kirtley (00:16:47)

When we look to how that word is defined, and we can look to how the experts define it, it doesn’t really apply to fusion. The Nuclear Regulatory Commission, the NRC, defines reactor as—I have it right here—”A nuclear reactor is an apparatus, other than an atomic weapon, designed or used to sustain nuclear fission in a self-supporting chain reaction.” There are two big parts to that: one, fission reaction. Obviously, fusion is not that, and we’ve talked about why, but also the self-sustaining part. A reactor is self-sustaining; you take your hands off it, and it keeps going. In fusion, that doesn’t happen, and we know because we have to do it every day, and it’s really hard to do.

Lex Fridman (00:17:33)

So we actually use the word generator because we don’t talk about, for instance, a natural gas reactor. If you stop putting in fuel, it turns off. The same thing happens in fusion. We’re pretty careful about making sure we talk about that as a generator where you’re putting in fuel, you’re getting electricity out. When you stop putting in fuel, it just shuts off. You can go even one step further and say, “What am I going to do with this fusion that powers the universe? And what does humanity want out of this?” What we want is electricity. We don’t simply want a set of reactions or even heat and energy. That’s great, but what I really want is electricity.

Lex Fridman (00:18:12)

And yeah, we’ll talk about the technical details of one of the big benefits of the linear design of the approach that you do, is you get electricity directly as quickly as possible. Some of the other alternatives have an intermediate step, and those, again, are technical details, but let me still linger on the difference between… …Fusion and fission. What are some advantages at a high level of nuclear fusion as a source of energy? Is nuclear fusion safe?

David Kirtley (00:18:38)

…fundamentally as a source of energy. In fusion, you’re taking these lightweight isotopes, you’re bringing them together, you’re releasing energy, and that energy is in the form of charged particles. It’s already in the form of electricity. Fusion itself has electricity built into it without a lot of the steam or thermal system requirements. That’s a really nice fundamental benefit of fusion itself. Also, this reaction that’s really hard to do turns itself off, so you end up with fusion being fundamentally safe, and that’s really a key requirement of any industrial system, is that it turns itself off and is safe. You turn the key off on your car, you know it’s going to turn off.

Lex Fridman (00:19:20)

I guess the flip side of that, just stating the obvious, but it’s nice to lay it out. For nuclear fission, it’s a chain reaction, so it’s hard to shut off, and it works by boiling water into steam, which spins turbines and produces electricity. Can you talk through this process in a nuclear fission reactor?

David Kirtley (00:19:43)

In a nuclear fission reactor, you put enough of this fissile material, uranium or plutonium, together such that as these unstable molecules, these unstable atoms, crack open and break apart, they release heat. The component parts of those are actually quite hot. So not only are the component parts that the uranium breaks into—and it’s a whole spectrum of different atoms and atomic nuclei—hot, but it also releases neutrons. It also releases more of these uncharged particles. If you do it right, this fissile material will be next to other fissile material, and so that neutron will then go and bombard another uranium nucleus, again opening that up and releasing more heat and more of these neutrons.

David Kirtley (00:20:31)

That’s how you have those reactions of a self-supporting chain reaction, and that chain reaction then continues. People design fission reactors such that you have just the right balance of enough neutrons are made such that the reaction is continuing, but not so many neutrons are made that it speeds up… …Because you don’t want it to speed up.

Lex Fridman (00:20:50)

And there are some kind of cooling mechanisms also? That’s part of the art and the engineering of it?

David Kirtley (00:20:56)

The key is at the same time, you want to make sure that the whole thing is in water, which is typically the cooling fluid. There are some more advanced fission reactors that have different cooling fluids, but water typically, then absorbs both the heat and those extra neutrons. So you use the water and the fluid to then run a steam turbine to do traditional electricity generation and output electricity through your steam turbine. You end up with complicated systems of flowing liquids and flowing water, balancing the heat.

David Kirtley (00:21:28)

A lot of fission reactor design comes from that thermal balance of keeping this reaction going, making sure it doesn’t speed up, because that’s an uncontrolled chain reaction, which you would not want, and balancing the cooling and the output of getting the water out of it.

Lex Fridman (00:21:48)

So we should say that for reasons you already laid out—maybe you can speak to it a bit more—nuclear fusion is much safer. There’s no chain reaction going on; you can just shut it off. But it should also be said that, as far as I understand, the current fission nuclear reactors are also very safe. I think there’s a perception that nuclear fission reactors are unsafe, they’re dangerous, and if you just look empirically at the statistics, that fear is not justified by the actual safety data. Can you just speak to that a little bit?

David Kirtley (00:22:22)

Yeah, we’ve been talking about the reaction processes themselves, but I think fundamentally, let’s take a step back and look a little broader and say, let’s look at what we care about, which is the power plant, making electricity. I look at this from a nuclear engineer’s point of view. I spent a lot of years studying these systems. Modern fission reactors, I believe, are engineered to be safe. They’re engineered in ways where as those reactions maybe speed up and those systems get hotter, they actually are built to expand and cool down passively and natively. There are protection systems in place that modern systems are quite safe from an engineering perspective.

Lex Fridman (00:23:03)

So I believe that we have figured out how to build nuclear fission reactors in a way where the engineering of the power plant is safe. I would say that I look back at the history of what we’ve built over time, and the challenge hasn’t come to the engineering, actually. I believe the engineers have solved these problems. The problem comes from humans, and the problem comes from other things around nuclear power. You have to enrich that uranium to put it in a plant, and the plant’s safe, but you had to enrich that uranium, and that is some of the problem. Or a plant is designed to run for a certain number of decades safely, but do we run it longer than that?

Lex Fridman (00:23:42)

And so those are where I think the real challenges happen, is more with the humans around these systems than the engineering of the power plants themselves.

Lex Fridman (00:23:50)

Well, I have to ask then, what do you think happened in Chernobyl? What lessons do we learn from the Chernobyl nuclear disaster and maybe also Three Mile Island and Fukushima accidents? I think you’re suggesting that it has to do with the humans a bit.

Lex Fridman (00:24:02)

So with Chernobyl and Fukushima, I actually put Three Mile Island in a different category. In fact, some of the recent news in the last year is that we’re going to be restarting Three Mile Island, because there’s such a need for clean base load power. So that’s actually a very interesting other topic we should talk about is, is why and how we’re doing that. But more than that, going back to the accidents that did happen in both of those systems, you can point to the human failure rather than the engineering failures of those systems. That in Fukushima specifically, there were multiple nuclear fission reactors on the same site that successfully kept running through the tsunami, totally successfully, and were only later shut down for more political reasons.

Lex Fridman (00:24:48)

But the old one, the oldest of them that had been on site for long periods, and maybe too long, I think some experts have looked at this in the past was where some of the problems actually happened. And so, I look to that less as a failure of the engineering of the power plants, and more of the humans around those systems. That we should be operating these plants as designed, and then I believe they’re safe. And that gets to some of the atomic weapons questions that I think are the other part around nuclear reactors and fission reactors that are concerning for me.

Lex Fridman (00:25:30)

Can you speak to those? So, maybe this is a good place to also lay out the difference between nuclear fission power plants and nuclear fission weapons, and maybe also nuclear fusion power plants and nuclear fusion weapons. What are the differences here?

David Kirtley (00:25:53)

Fusion power plants can’t be used to make nuclear weapons. Fundamentally, the processes in fusion aren’t the same processes that happen in nuclear bombs and nuclear weapons. And so it’s actually one reason I started in fusion, and most of our team thinks about the mission of fusion, of delivering clean, safe electricity, is it also can’t be used to make weapons. And I think that’s a little bit of a distinction from traditional nuclear fission reactors, is that while I totally believe as a nuclear engineer we can build power plants now that are safe, that aren’t going to have reactions. They use a fuel, uranium and plutonium, that can be used to make nuclear weapons.

David Kirtley (00:26:40)

We know that if you take enough fissile material together, enough uranium and plutonium, put it in a small volume, that it will not just create a reaction, but it will create a supercritical reaction that will then continue and grow and release a tremendous amount of energy all at once. And that is a bomb. That is a bad situation, and that is what we want to avoid. A lot of the key is recognizing that even though there are things called fusion bombs, the H-bomb, the hydrogen bomb, the hydrogen bomb has uranium in it. It’s still a fission bomb. And so how this fundamentally works is that you have a fission reaction, a primary, and that creates radiation that induces a fusion reaction with a small amount of fusion fuel that then boosts that uranium reaction again.

Lex Fridman (00:27:29)

And so most of the energy, in fact 90% of the energy in an H-bomb, is all still from the uranium reactions themselves.

Lex Fridman (00:27:37)

Yeah, I think people call it sort of the nuclear fusion bomb, hydrogen bomb, but really it’s still a nuclear fission bomb. It’s just that fusion is a part of the process to make it more powerful, but you still need, like you said, the uranium fuel. So it’s not accurate to sort of think of it as a fusion bomb really.

Lex Fridman (00:27:53)

And if you take away that fissile material, that nuclear fission reaction, the fusion reaction doesn’t happen at all. In fact, there have been researchers that have over the decades tried to make an all-fusion bomb and been very unsuccessful at it. The physics and the engineering don’t support it can ever happen with our understanding today. The topic we’re talking about is more broadly called proliferation, and this is the creation of nuclear weapons in the world and the distribution of those weapons. And something we know as physicists and engineers is that fusion can’t be used to make nuclear weapons. We know that. But that is not widely known.

Lex Fridman (00:28:38)

And part of what we went out to do is work with the proliferation experts in the world, the people who work to prevent nuclear weapons from being made, being created, being shared throughout the world, because we know the challenges, the geopolitical challenges that happen. And we went to those proliferation experts, and we were worried they would have sort of the same historical question of, like, “Well, the word nuclear is in fusion, so therefore it must be related.” And in fact, the total opposite happened. What they told us is, “Please, please go develop fusion power plants absolutely as fast as possible.

David Kirtley (00:29:15)

The world needs this.” And the proliferation experts were telling us that otherwise people would start enriching uranium throughout the world, and we’d be building enriched uranium power plants because we need the electricity that’s clean and base load. But in those processes, they’ll be making fuel that could be one day used for atomic weapons, for nuclear weapons, and they were worried that the growth of this enriched uranium, think about the centrifuges, that having a lot more centrifuges happening all over the world would lead to more weapons, at least the possibility of it. And so they are pushing us as fast as possible, go build fusion generators and get them deployed everywhere.

David Kirtley (00:29:55)

Not that’s just in the United States, but all over the world so that we’re building fusion power and that’s meeting humanity’s needs, not this other thing. And so I was really pleasantly surprised. We’ve written a number of papers and worked with those communities on this of what does it mean, how is fusion power safe and can’t be used for nuclear weapons.

Lex Fridman (00:30:18)

So, this might be interesting to ask on the geopolitics side of things. I have the chance to interview a few world leaders coming up. By way of advice, what questions should I ask world leaders to figure out the geopolitics of nuclear proliferation? nuclear weapons, nuclear fission power plants, and nuclear fusion power plants? What’s the interesting intricate complexity there that you could maybe speak to?

Lex Fridman (00:30:47)

The question I would want to ask is, “What would you do?” If we could deliver for you low-cost, clean, industrial scale, tens or hundreds of megawatts of fusion power that’s low-cost, clean, base load and doesn’t have the geopolitical consequences of uranium and plutonium, of fissile material, what would you do there? How would that change your view of the next 30 years?

Lex Fridman (00:31:17)

But also, there’s a lot of geopolitics connected to oil, natural gas… …And other sources of energy which I think are important in Saudi Arabia, in the Middle East, in Russia, I mean, all across the world. And that’s interesting too. So do you think actually if everybody has nuclear fusion power plants, that alleviates some of the geopolitical tension that have to do with energy, other energy sources?

David Kirtley (00:31:42)

I certainly do, that the fuel is in seawater all over Earth. Everybody has deuterium, and everybody has it. And so you can’t have a monopoly on the fuel. And no one can control the fuel and no one can turn off the fuel, no one can cut a pipeline. That just cannot happen with fusion. And so if we can deploy those plants and we can deploy them quickly, then it decouples the ability of any one or any few countries to control energy. Extreme scenarios

Lex Fridman (00:32:12)

Okay, so let’s return to the basic question. We already mentioned it a little bit, but is nuclear fusion safe? So, are the fusion power plants we’re talking about safe?

David Kirtley (00:32:26)

Yes, fusion power is fundamentally safe. The physics and the reactions of the fusion system itself mean you don’t have runaways. We’ve talked about some of the human factors around power plants, power systems, and industrial scale systems, and that’s something that we build into the design of these from today. We look at how these systems might fail. In fact, some of the analysis we do, we did this analysis for the Nuclear Regulatory Commission over the last few years, looking at how to regulate fusion power. As we’re building the first fusion power plant, we need to make sure we’re regulated safely. So we spent a lot of time doing the technical and political case in the United States of how to regulate fusion.

David Kirtley (00:33:18)

The analysis we did assumes you have a fusion power plant that’s operating and then, at any one time, a meteor strikes it. The whole thing is vaporized. What is the impact of that? This is worse than you could ever imagine an actual physical scenario, but let’s start there. The answer is, you don’t need to evacuate the populace nearby the fusion power plant. One of the keys, I think, that I come to when I think about this is the fuel. In a fusion generator, you are continuously feeding in this hydrogen, these deuterium fuels. At any one time, in a Helion fusion system and most fusion systems, you have one second of fuel in that system. So what that means is if you stop turning on…

David Kirtley (00:34:08)

If you stop putting fuel into that system, fusion just stops. But what it also means is that if something really catastrophic happened and for whatever reason you have all that fuel that’s not in the system, and fusion is so hard to make happen, you hit it with a meteor, you do anything of that nature, and fusion doesn’t happen. That hydrogen, that heavy water, that deuterium, just goes back into the environment safely and cleanly without issue. That’s the fundamental safety mechanism of fusion. You can compare that with other types of power plants like an oil or a coal power plant. You might have a large pile of coal that then catches fire and burns.

David Kirtley (00:34:47)

It’s not catastrophic, but you have a large coal fire for a long time releasing toxic fumes that you may have to deal with. In nuclear power, in a fission power plant, you may have several years of fuel sitting in the core. In that case, if something bad happened, you have all that potential energy for things to happen. But in fusion, you have literally one second of fuel at any time in the system. Having a tank of deuterium, which we have around all the time, can’t do fusion by itself. It needs that complex system.

Lex Fridman (00:35:17)

I love that there’s like a PowerPoint going on in a secret meeting about what happens if a meteor hits a fusion power plant. Okay, so that’s really interesting. What about the waste? What kind of waste is there for fusion power plants?

David Kirtley (00:35:31)

The fusion reaction itself is still fundamentally an atomic reaction. So during this reaction, you do create ionizing radiation. You create X-rays, you create neutrons, and you create all these charged particles. The charged particles themselves for a fusion reaction are all contained in the fusion system. The X-rays, similar to a dentist office, although a lot more than that, but that same type of X-ray and X-ray energy is absorbed by the fusion system. But the thing we do care about is those neutrons. So we do have, in a fusion system, activation. During its operation, neutrons are made and leave, so we have to shield these fusion systems during their operation.

David Kirtley (00:36:13)

This is very similar, in fact, this is a lot of the work we did with the Nuclear Regulatory Commission over the last number of years. There was a landmark agreement that happened for the NRC that was then codified into law last year called the ADVANCE Act, which is really powerful because it says, for the very first time—how the U.S. government, leading the way on this, which I’m really proud of, will regulate fusion. This gets into a little bit of the details, but the way the Nuclear Regulatory Commission regulates nuclear things in the United States is in these different sets of statutes. Nuclear reactors are regulated under something called Part 50.

Lex Fridman (00:36:54)

And there’s a lot of variety in the regulatory language around that, but most of it is to handle special nuclear materials, uranium and plutonium. But fusion is not. Fusion is regulated under something called Part 30. Part 30 is how hospitals are regulated, particle accelerators, other types of irradiators where, as they’re operating, you have very high energy particles, ionizing radiation, and you have to protect operators from it. You have to shield them, so we build concrete shields. If you came and visited Helion, you would see plastic, borated polyethylene, and concrete shielding to protect operators and equipment from the fusion reactions while they’re happening. But again, you turn them off, and those fusion reactions stop. And that’s really the key.

David Kirtley (00:37:38)

There’s a funny story related to that. We’ve been building fusion systems that do fusion for a long time, and at some level, they got powerful enough, doing enough fusion, we started building these shields and shielding them like a particle accelerator. I went to the regulatory bodies that regulate Part 30. This is in Washington state, it’s the Department of Health. So I went to the Department of Health and said, “Here’s an application for a fusion generator shielding permit as a particle accelerator.” The very first question I got asked was, “Great, where do the patients go?” Because the standard form had a patient as a hospital, the patient dose for the particle accelerator, and then the shielding.

David Kirtley (00:38:26)

We talked all about the shielding and the operators, which is very similar for a Helion system. We said, “No, no patients at all. No one’s inside this thing. Our goal is to generate electricity one day.” This was a lot of years ago. We were able to go through and work with the state agencies to license these fusion particle accelerators. We were, as far as we know, the first licensed fusion system ever as a particle accelerator for those first systems. The first license we had was in 2020. We then have gone on and now licensed several of our fusion systems that we’ve built that do fusion, both the shielding as well as some of the fuel processes. How nuclear fusion works

Lex Fridman (00:39:08)

So high level, what are the different ways to build a nuclear fusion power plant? Can you explain what a tokamak is, what a stellarator is, and what’s the linear approach that Helion is using?

David Kirtley (00:39:25)

There are a number of ways to do fusion. Fundamentally, in all fusion approaches, you’re trying to do the same physical process, which is take these lightweight isotopes, heat them up so that they can move at high velocity—over 100 million degrees—bring enough of them together. We call it density. Enough of them together in a certain volume so that you have reactions happening at a higher rate, and keep them together long enough that they are able to collide into each other and do fusion and release energy. That’s the fundamental core. Now, how you do that, how you bring those particles together, how you hold them together long enough, there’s a wide range of technologies that, as humans, we’ve been exploring since the 1950s.

Lex Fridman (00:40:12)

And I think about several main categories. If you look at the fusion funding out there, government funding in the world, private funding actually has quite a different profile, which is an interesting thing to talk about. But in public funding, in federal funding in the United States, there are two mainline programs called inertial fusion and magnetic fusion. In inertial fusion, what you’re trying to do is bring together and push together by a variety of physical means, those particles. You push them together. The most common is called laser inertial fusion. Our colleagues at the National Ignition Facility did this really well and made world records in the last few years for being able to demonstrate you can do this and do it at scale.

David Kirtley (00:40:58)

Where you take very high power lasers and pulse them together to combine them to do fusion for a pulse, for a very short period of time: nanoseconds, billionths of a second. The other extreme, and you mentioned tokamaks and stellarators. Stellarators are actually my favorite, so we’ll talk about those. As a graduate student in fusion, the stellarator is the first thing you learn about. Because there’s a mathematical solution for a stellarator that solves perfectly. And you can write it out and you can solve it, and analytically, it’s very simple. Building one is very hard. It’s taken humanity a number of decades to be able to build stellarators, and we can do it now with the Wendelstein 7-X that came online in the last few years, being the premier stellarator in the world.

Lex Fridman (00:41:49)

I should say, all the different ways to do fusion all just look so badass in terms of engineering. Creating this containment, extremely high temperature, high density. Everything’s moving super fast. Everything is happening super fast. It’s just fascinating that humans are able to do it. Like, there are certain things, accelerators are that a little bit, but this is even cooler because you’re generating energy that can power humanity with this machine. Anyway, can you just speak a little bit more to the inertial and the magnetic fusion systems?

David Kirtley (00:42:23)

In a magnetic system, your goal is not to push together those particles as fast as possible. Your goal is to hold onto them for as long as possible. To do that, we use magnetic fields. So let’s take a step back. What is a magnetic field? In an electromagnet, there are a variety of ways to make a magnetic field. One of the most famous, I think everyone is familiar with, is Earth itself. Earth has what we call the magnetosphere, which is the magnetic protection that’s generated by the core of the Earth. We have a magnetic field around the Earth, and that magnetic field protects us from particles coming from the galaxy, galactic cosmic rays, and solar particles that would come to Earth.

David Kirtley (00:43:11)

That magnetic field, when you run a compass, you see the magnetic field from the Earth. So we know it’s happening. It’s all over. But how we generate it with electric currents is a little bit different. What we do is that we have a loop of wire, and the simplest way to think about it is literally a round loop. In that loop, you have electrons; you have electrical current that’s running. And when electrical current, this is some of Maxwell’s equations that we discovered in the 1800s, when you have an electrical current in a wire, it generates a magnetic field inside that wire. And so when you look at fusion systems, you always have these big magnetic coils with large amounts of current. We don’t run a little bit of current.

David Kirtley (00:43:52)

In our systems, we have hundreds of mega amps of current. If you think about your house, you have your breaker box with 200 amps or maybe a 400-amp breaker box, and we run 100 million amps of electrical current. So massive amounts of electrical current to be able to do this. That magnetic field that’s generated inside that magnetic coil has some really special properties, and we take advantage of those properties to do fusion. Some of those properties are not intuitive. So here’s one of my favorites: when you have an electromagnetic field, you have this coil with electricity going around it and you have a magnetic field inside of it, and then you have a test particle, a charged particle, an electron or an ion.

David Kirtley (00:44:39)

If you imagine to generate this, I have a coil with electrons moving around it. But if I put one in the middle of it, in this magnetic field, some really interesting things happen. That electron or that ion, that charged particle, is what’s called magnetized. And what magnetized means is that it’s trapped on that field line. In fact, even more interesting is that it oscillates around that field line. So the way I think about this is if you think about the Earth’s magnetosphere again, and you think about the charged particles, the aurora, the Northern Lights, is a charged particle trapped in the Earth’s magnetic field going around the Earth’s magnetic field.

David Kirtley (00:45:18)

In the same way, in fusion, we do the same thing here on Earth, but in a smaller direction where we trap these particles on magnetic fields, and they can go around and stay trapped to that magnetic field line.

Lex Fridman (00:45:29)

How much of the physics at this scale is understood here? Like, how these systems behave when you attract a magnetic field in this way. Is this fundamentally now an engineering problem, or is there new physics to be discovered about how the system is behaving?

David Kirtley (00:45:48)

In fusion, the physics we’re using is actually quite old. The fundamental electromagnetic physics is 1800s physics. The fundamental atomic physics is early 1900s. So the fundamental physics of how these work is very well understood. Putting them all together into a power plant, that’s hard. So you can do the math, every introductory grad student does the math on a stellarator and says, “This is all I need to do. I just need to make a magnetic coil in this very complicated shape, and then fusion will happen.” However, doing that in practice is actually quite challenging.

Lex Fridman (00:46:27)

So maybe you could speak a little bit more. So the stellarator and the tokamak, what’s the difference between those two? They’re both magnetic fusion systems. And then what does Helion do?

David Kirtley (00:46:38)

The tokamak and the stellarator are both magnetic systems. Their goal is to generate this magnetic field and hold onto the fusion fuel long enough. Like I mentioned, these charged particles are trapped on the magnetic field. In fact, they’re oscillating. We call that a gyro orbit; it’s the radius that they oscillate around this magnetic field. We’ve been talking about atomic physics, where everything is at this nanoscale. But gyro orbits are not. Gyro orbits for these fusion particles are measured in inches. So they’re on a scale that we can see and measure and understand really intuitively.

David Kirtley (00:47:16)

In a magnetic system, your goal is to simply trap as many of these particles as you can for long enough and heat them so they’re hot enough so that they bang into each other. They collide enough that you’re doing fusion, and you’re doing enough fusion to overcome as fast as you’re losing those particles. And so that’s what happens when you put particles in a magnetic field and you try to hold onto it. The challenge is that’s really hard to hold onto them long enough. These particles are moving around. They’re moving at very high velocity, millions of miles per hour. They’re colliding with each other, and they’re getting knocked off and getting knocked away.

Lex Fridman (00:47:51)

So we’ve talked about inertial fusion, where you try to confine a fusion plasma by crushing it as fast as possible. And magnetic fusion, where you just simply have a magnetic field and your goal is to hold onto it for as long as possible. But there’s another way to do fusion, and in some ways, it’s one of the earliest approaches for fusion that was successful. As scientists and engineers, maybe we’re not too creative with the terminology. We call the technique that Helion uses magneto-inertial fusion because it does a little bit of both.

David Kirtley (00:48:25)

To understand that, we can actually go back in history a little bit and think about the evolution of some of these approaches to fusion. From our perspective, we look at the technology that we use as built on physics experiments that were very successful in the 1950s. In those systems, the earliest pioneers of fusion said, “I know, we understand the physics. We have to take these gases, heat them to 100 million degrees, and then confine them, push them together so that fusion happens.” So, what is the best way to do that? Some of the earliest programs, we call them theta pinch. And what those programs were, were a linear topology, because we knew how to build these magnets.

David Kirtley (00:49:08)

It’s called a solenoid, where you take a series of electric coils, you run electrical current through them, that generates a magnetic field. Great, so you have a magnetic field. Now you add your fusion particles. Okay? So you’ve added fusion particles to this solenoid. Here’s the challenge. Those particles, as they’re sitting in that magnetic field in this nice magnet, escape. They leave out the ends because there’s nothing holding them in. Great. So that makes sense. And so that doesn’t work, okay? So then the next approach is to say, well, one branch of fusion said, “Okay, well, to solve that, why don’t we take the solenoid and bend it around? Let’s just make it a big donut. So as they’re escaping, they go around and around in a circle.” Great. That’s a great approach.

David Kirtley (00:49:49)

One branch of fusion went down that direction, and that became, that evolved into the stellarator and the tokamak. Different ways of taking those solenoids and wrapping them around so that the plasmas go around and around in that magnetic field, and those charged particles are held long enough that fusion happens. But there’s a different way to do it. And so the theta pinch was what was born in the 1950s of, “Take this magnetic field, and oh, they’re trying to escape. Great. Let’s not let them escape. Let’s close the bottle, let’s close the ends.” And so we make the magnetic field much stronger at the ends. This one was called the mirror. And so the idea was that the particles would bounce in between.

Lex Fridman (00:50:30)

And that worked, and they got hotter and hotter and hotter. But guess what? As you would imagine, as this mirror topology, this linear topology, the pressure increased inside, the particle pressure, the particles tried to push back on the magnetic field. They were trying to escape now. They’re getting hotter and hotter. And just as you imagine, hot gas in a balloon tries to get out the ends, you could not hold it tight enough at the ends to keep those particles in. In fact, the problem is the hottest ones were the ones that would escape. So you do a good job of heating it, and they’d all leave out the ends. So then the next iteration said, “Okay, well, why don’t we just not try to hold on to it very long?”

Lex Fridman (00:51:07)

Why don’t we squeeze it?” So rather than just holding it constantly, let’s now crush it. So we built this solenoid, we pinched the ends, and then we crushed it. And what I mean by crushing it is not actually crushing any magnets or changing the topology or moving any parts, but just rapidly increasing the magnetic field. So going from a magnetic field that’s just holding it to now taking all those particles, if you imagine they were streaming around together, and then rapidly increasing the magnetic field so that those particles get closer and closer and closer together. So you increase the density. And now fusion starts to really happen. But they ended up hitting a technological limit.

Lex Fridman (00:51:53)

So this is the part that I look back and I look at the pioneers. In 1958, there was some pioneering work done. And this was in California, what later became Livermore Labs. There was also some work done at other national labs too. These were all federally funded programs to explore this theta pinch topology. Can you just squeeze the plasma down fast enough, hard enough? This was 1958. The transistor was sitting in the laboratory, and they were turning on millions of amps of electrical current. And they were doing it, we haven’t talked about the time scales, but they were doing it in millionths of a second. Microseconds, megahertz speeds. And this was in 1958.

David Kirtley (00:52:39)

No transistor, no CPUs, and no electrical switches, none of the things that I take for granted every day. And so they were able to show at that time the highest performing fusion systems. They got to temperatures… They didn’t get to 100 million degrees, not quite then, but they got to 50 million degrees. They were outperforming everything else in fusion, but they reached the technical limit where they just could not build it anymore. And so those pioneers went in a different direction, and they started down the laser inertial path, saying, “Okay, well, we can’t do these electromagnetic pinches, but we now have this new thing invented: the laser,” which turns on in nanoseconds. It’s fast. It’s interesting. Let’s go down that path.

Lex Fridman (00:53:25)

And it’s not… You have to fast-forward a couple of decades to researchers who found that with some of these theta pinches when they’re operated in a very specific way, something else happened, something new happened, and that these plasmas, where before they squeezed them very hard, and just like squeezing a tube of toothpaste, they squirted out the ends. Now it didn’t squirt out the ends. It actually pushed back. It stayed confined. It stayed trapped inside that linear topology. Even though the ends were open, the plasma didn’t leave. And so there was a large amount of programs asking, “What is happening here?” This is an accidental discovery in plasma physics that something new is happening. And what we discovered is what we now call the field-reversed configuration.

David Kirtley (00:54:09)

There are numerous FRC (field-reversed configuration) programs, both at national labs. There are actually a number of private companies now building field-reversed configurations. And they have some really unique properties, but fundamentally, talking about the main difference, I describe the solenoid with magnetic fields throughout the center of that volume, and plasma trapped, going back and forth. But some other things can happen, which is really interesting.

Lex Fridman (00:54:37)

And what they discovered early is if they have field going in one direction, so the plasma, the electrical current is going around the loop and the plasma is going back and forth along this magnetic field line inside that solenoid, inside that theta pinch, but then they change the direction of the magnetic field. And this is what we call field reversal, and this is really the key: that you start with the plasma going in one direction, and then very rapidly, you change the direction. You change the direction and reverse the direction of that field. And something really interesting happens, which is the plasma, this fusion fuel, these charged particles which are trapped on the magnetic field lines that are moving back and forth, you change the direction.

Lex Fridman (00:55:20)

What that means is that you’re trying to take that electrical current and that magnetic field and reverse its direction, flip it, but it can’t flip fast enough. The plasma is sitting there and you can’t move the particles. So what’s really interesting is what happens is that because the particles can’t move, but you’ve now flipped the direction of the magnetic field, you’ve inverted it, something really, really unique happens, which is that the plasma itself reconnects internally. So now what you’re left with is an outside magnetic field, an electrical coil, and inside, the plasma, where before it was moving along, it’s now moving internally.

Lex Fridman (00:56:06)

Rapidly reversing the magnetic field, plasma self-organizes into a closed field. What? So how…

Lex Fridman (00:56:16)

It sounds wild.

Lex Fridman (00:56:17)

It’s… Yeah. So first of all, there are a million questions I have. So one of them, what’s “rapidly”? What time scale are we talking about here?

David Kirtley (00:56:28)

You have to reverse the electrical current faster than a millionth of a degree, which is a very hot gas particle, can move.

Lex Fridman (00:56:37)

Okay.

Lex Fridman (00:56:37)

And so that means we have to do it on the order of a millionth of a second.

Lex Fridman (00:56:40)

Wow.

Lex Fridman (00:56:41)

We have to do it in a millionth of a second.

Lex Fridman (00:56:42)

Wow.

Lex Fridman (00:56:43)

And in practice, this is hard. And we can only do it now because of semiconductor switching. Because we can move things, we can switch things. Like the transistor in every CPU in a computer switches at a gigahertz; that means in a nanosecond, it’s switching in a billionth of a second. And so now, which we didn’t in the 1950s when these theta pinches were invented, but now we have the semiconductors to be able to do that.

Lex Fridman (00:57:10)

The self-organizing plasma. Can you just speak to that? What the heck is it doing? How do we discover? How do we understand the self-organizing mechanism, the dynamics of the plasma that’s able to contain itself?

Lex Fridman (00:57:23)

So what I like to do is use an analogy here. Once you’ve made it, it’s actually somewhat straightforward to understand. Getting to it is tricky, and how they discovered it the first time is absolutely amazing. But once you’ve made it, it’s a lot more straightforward to understand. So in a magnetic coil, when you have a round electrical coil, you have electrical current flowing in that coil. And if you have a conductor, if you have another metal inside that coil—and this is called Lenz’s Law in one of the Maxwell equations—is that as you have electrons and you have current flowing in that coil, an equal and opposite electrical current is induced in a piece of metal nearby.

David Kirtley (00:58:11)

This is the same thing that happens in a transformer where you have a primary on a transformer and you have electricity flowing in it, and you have a secondary where electricity flows exactly the opposite direction. We use this every day in our lives. And so in this condition, you have a conductor, an electrical conductor where current can flow, and you have an electrical current flowing on the outside, electrical current flows on the inside. And in that case, now you… I’ve described two pieces of metal. Now let’s go one step further, and that inner conductor is not a piece of metal anymore. It’s one of these high-temperature gases, this plasma, these charged particles. So now you have current, electrical current flowing in the plasma.

David Kirtley (00:58:54)

This is really, really interesting. We talked about these charges moving back and forth. Well, moving electrical charges is current. So in every plasma condition we’ve talked about—the tokamak, the theta pinch, the stellarator—there’s electrical current flowing in the plasma. But in the field-reversed configuration, you have a lot of electrical current flowing in the plasma, massive amounts of it. And that’s the key. So you have the center core where electrical current is flowing in this transformer, if you want to think about it, primary and secondary. And here’s the craziest part of it: this electrical current… Well, how did I describe a magnet? An electromagnet is a loop that has electrical current flowing in it that generates a magnetic field.

Lex Fridman (00:59:37)

And for a theta pinch, and for a mirror, and for a tokamak, in that magnetic field, the plasma gets trapped. But in an FRC, this electrical current is the plasma. And that plasma then generates its own magnetic field, and it’s then trapped in its own magnetic field.

Lex Fridman (00:59:59)

That’s fascinating.

Lex Fridman (01:00:00)

And that’s the key. And so in your tokamak, in your donut, in your funky donut, your stellarator— you make the magnets and you trap your plasma in it. In an FRC, you make the plasma which makes the magnets, and it traps itself. And the craziest part of this, in my mind, is that we actually see this in nature all the time. If you look at the sun, we see solar flares. And in a solar flare, we’ve all seen the pictures of the photosphere of the sun and this large arc of plasma coming out. That plasma has electrical current flowing in it, and then we see this solar flare rip off of the sun.

Lex Fridman (01:00:41)

And that solar flare then can flow throughout and continue into the solar system, and for a little while anyway, it makes something called a plasmoid. That plasmoid is, in fact, electrical current flowing in the plasma, generating a magnetic field and holding it for longer than it would otherwise. And so we’ve observed these for 100 years, and we’ve known about these plasmoids for a long time, and there are researchers that have tried intentionally to make them. But fundamentally, that’s what we do every day: we make one of these self-organized closed-field plasmas.

Lex Fridman (01:01:15)

In a more controlled way at this rapid rate of one-millionth of a second, and being able to make sure it’s reliable, stable, and all that kind of stuff. So, by the way, how do you keep the thing stable?

Lex Fridman (01:01:26)

And there’s the hard part, because I just described a solar flare. But, yes, we’ve seen the pictures of them, but we’ve also watched them, and they appear. They fly away from the sun and then they go away, and that’s not what we want in fusion, right? We want to be able to control this. And so that’s the hard part of the job. And so that’s what we’ve spent the last number of years learning how to do, ourselves and others, on these pulsed, closed-field FRC systems.

David Kirtley (01:01:53)

Let’s first talk about how to make them, and then we’ll talk about how to make them stable, because they’re two different things and we spend a lot of time on both. So, we talked about timescales. You have to reverse the field. You have to change the electrical current in a millionth of a second. And so how do you do that? So I’ve described this system as you have a series of magnets. You have a magnetic field on the outside, and then on the inside of this, you have this donut, this FRC that has its own electrical current. And we didn’t talk about this yet, but it’s generated a magnetic field and that magnetic field has pressure, and this is the other thing that’s really interesting. So we talked about how this theta pinch compresses a magnetic field.

David Kirtley (01:02:34)

It applies a pressure on the outside. But the plasma itself has a pressure on the inside, and it has both a particle pressure, literally the particles bouncing. Think about hot gas in a balloon, the particles expanding, the ideal gas law expanding and contracting inside a balloon. But they also have a magnetic pressure. They have the electromagnetism pushing back, and so I like to think about this as the motor in a Tesla. In your electric car, you have an electric motor, and what that motor has is a series of windings. Those windings, you flow electrical current, in this case, from a battery. Hit the gas, electricity flows from the battery into the motor, into those windings, and it generates an electromagnetic force, a Lorentz force is what it’s technically called.

David Kirtley (01:03:20)

This electromagnetic force induces an electrical current on the armature, on the shaft. And this is getting into the details, but in the armature of an electrical motor, that actually is what spins. And so the outside of a motor doesn’t spin. You flow electrical current through it, and the inside does spin. That electromagnetic force is what is spinning that armature. In our case, we’re inducing an electrical force in that electromagnet, and that’s putting an electrical current, just like in the armature, into that plasma. And we can use that force to do interesting things. So that electromagnetic force can compress the fusion plasma. It can expand the fusion plasma. But here’s the problem: it’s unstable.

Lex Fridman (01:04:02)

And so this is something you learn very early in your graduate work as a student in fusion, is you learn about plasmas that are called high beta plasmas.

Lex Fridman (01:04:13)

So I keep seeing this plasma beta thing everywhere. What is this ratio of plasma field energy to confining magnetic field energy? Please explain.

David Kirtley (01:04:23)

Plasma beta is the ratio of the magnetic pressure to the particle pressure. And so what that fundamentally means is I talked about how you have a magnetic field, and in that magnetic field, plasma is trapped in, on that magnetic field. But it’s not very well trapped. It can escape. It can leave either down the ends, it can freely travel, or it can also travel across the magnetic field. And so we have a term called plasma beta which gives us an understanding of how well trapped that plasma is. So, as you apply a magnetic pressure, a magnetic field to this plasma, it pushes back, and does it push back a little or does it push back a lot? And for a field-reversed configuration in one of our plasmas, beta is very close to one.

David Kirtley (01:05:09)

In fact, usually by definition, one at any point in the system, which means that every time I apply a magnetic force on this donut to compress it, the plasma particles on the inside push back.

Lex Fridman (01:05:23)

And what’s really interesting is you have an equation for magnetic pressure, which is B squared over two mu naught. The magnetic field squared is the external magnetic pressure. Any magnetic field anywhere generates this pressure. But the plasma particles themselves also have a pressure. This is the ideal gas law, and we use the definition NKT, density, Boltzmann constant, and temperature for pressure. And in high beta, they’re the same. B squared over two mu naught is NKT. So for a known magnetic field, I know what the density and the temperature of the plasma is. And just to circle back to it, when we talked about fusion, we talked about it had to be hot enough and it had to be dense enough. And that’s N and that’s T.

Lex Fridman (01:06:09)

And so now I have a very clear equation between magnetic field and density and temperature of the fusion fuel, and that’s really critical. All plasmas have some… All fusion plasmas have some beta, some number. The FRC has one of the highest betas, beta equal one. However, what you also learn in school when you learn about beta the first time, is you learn that high beta plasmas are typically unstable. And so the good way to think about this is a tokamak is a accelerator are stable, because those plasmas that are going around in the donut, there’s a force on that donut. But that plasma donut is very well held by all those magnetic fields, by all those magnetic coils. If it tried to move, it would be confined by that magnetic coil. But in an FRC, it’s unconfined.

Lex Fridman (01:06:59)

So the plasma is confined, but the whole topology can do something what is called tilt, is that this whole plasma donut, because it’s under pressure, can just turn over. The way I think about this is think about the a motor’s a good example. An armature in the center of your motor, you have a spinning armature. You have this spinning magnet on the inside, and it is held by the main axis of the magnet. It can’t go anywhere. We don’t have that axis. We don’t have any mechanical things inside these fusion systems. They’re 100 million degrees. You can’t put any mechanical things inside them, and so we have nothing to hold onto it, and so it’s unstable.

Lex Fridman (01:07:41)

So when you learn about the FRC, that’s the first thing you learn, and it took us a number of years to learn about a parameter of how to make them stable. And that’s pretty fundamental, but most people who’ve heard of an FRC haven’t understood this really key fact. And so we have a parameter we call S star over E. And we’re getting really into the physics weeds here, but…

Lex Fridman (01:08:04)

Let’s go.

Lex Fridman (01:08:05)

But it’s really important, and the good analogy here is a top. Literally a top, a spinning top, and so you have a top spinning on your desk. You know that it’ll spin for a little while and then it will fall over. It is unstable. However, if you spin it fast enough, if you take a top and you spin it fast enough, put enough angular momentum, enough angular inertia into that system, it’ll stay upright even though it wants to just fall over, even though it’s unstable. And we do the same thing in an FRC, is if you can drive it fast enough, if you can add enough kinetic energy and inertia to the particles, it will stay stable. However, you can do another really key thing. We are not limited now to having a very skinny top.

David Kirtley (01:08:51)

We can actually make it much bigger, so the good analogy here is if you have a coin and you know you’re spinning that coin, if you spin it faster and faster, it’ll stay spinning longer. However, eventually it’ll slow down and fall over, but if you had a roll of duct tape, if you had something thicker and heavier and longer, and it’s spinning around that same axis, it’ll stay spinning even longer both because of the inertia and because of the geometry. And so we have this parameter called S star over E. S star is the hybrid kinetic parameter which tells you how stable it is from that top point of view, and the E, which is the elongation of how long it is.

Lex Fridman (01:09:31)

And so, maybe fortuitously, thank you nature gave us a win here, which is that how we make these in these long solenoids is naturally very, very long. And so we can build these with very long lengths, and if we can drive them fast enough and hard enough and drive the ions to move at very high velocities, we can stabilize against those instabilities and hold them stable. And so we now know we can design with a given S star over E parameter, we can design these for very long lives. The theory of the systems we make say that they should last for a few microseconds at most. Us and others in the field have been able to make them last for thousands of microseconds, thousands of times what the basic criteria would tell you.

Lex Fridman (01:10:21)

And so we know now how to do this, and we just design them with this built into them.

Lex Fridman (01:10:26)

Can you explain a little bit more of the S star over E? Are you given that, or is that an emergent thing? So like at which stage… Is that the result or the- the requirement?

David Kirtley (01:10:39)

It’s a great question. So it is a requirement of the system, is that you must design it with this parameter in mind.

Lex Fridman (01:10:47)

Got it.

David Kirtley (01:10:48)

The hard part is you have to design it with S star over E being satisfied the whole time.

Lex Fridman (01:10:54)

Right.

Lex Fridman (01:10:54)

And here’s the extra trick here. S star over E is also a measure of temperature.

Lex Fridman (01:11:02)

Oh boy.

David Kirtley (01:11:02)

And, and, yup, this, it all comes back to temperature. The hotter you make them is the same thing, temperature as kinetic energy, is the faster you’re spinning. So if you take your top and you spin it faster, it’s more stable. But you’ve got to make it hot. And so here’s the trick: How do you make something hot that’s starting cold? And it has to be hot by definition. And so that’s part of the challenge of what we do day-to-day, is getting to these hot plasmas. And where other people have tried to make FRCs and not been very successful is because they couldn’t get it hot enough fast enough. It fell over, it tilted, before it got hot. And so we spend a lot of our electrical engineering…

David Kirtley (01:11:45)

In some ways, Helion is more of an electrical engineering company than a fusion company some days, focusing on how to make the electronics fast enough to be able to get it hot enough soon enough that you can keep it stable the whole time. Extreme temperatures

Lex Fridman (01:12:00)

So you’re trying to reach 100 million degrees. How do you get to that temperature fast? And by the way, what can you say to help somebody like me understand what 100 million degrees is like? It seems insane. What does that world look like? I guess just everything is moving really fast. Like you said, you can’t put anything mechanical in there.

David Kirtley (01:12:21)

Yeah, so a couple of key things happened. When gas is that hot, there’s… We talk about the states of matter. You have solids, where ice, it’s cold. The atoms are now bound in a lattice structure together. They’re held together. And then liquid, you’ve broken a lot of that lattice structure. They can move around. They have some kinetic energy, but they’re still pretty contained, they stay in the bowl. Keep heating it, now you’re in gas. And now these particles are free to move around. They’re moving around, they’re bouncing off of each other all the time, and you can keep heating it from there, and that’s where we talk about some more phases of matter. We can add a little bit more physics here. We talk about rarefied gasses.

Lex Fridman (01:13:02)

So when we think about most gasses that humans interact with, they act like a fluid. And what I mean by that is that they’re colliding with each other so often that the particles at any one place, here the air is roughly the same temperature as the air here. That these particles are bouncing off of each other as if you’ve put a really hot one right here, it would then cool enough that all the air is roughly on the same temperature. But you can be what is called rarefied, and this is like space. This is where now you have particles moving around, but they don’t collide with each other very often.

Lex Fridman (01:13:34)

And so you can have one very, very high energy particle and very cold energy particle, and they may not even touch each other, but maybe occasionally they bang into each other, they collide, and then they transfer energy. And that’s what we call rarefied. And then you can go even hotter than that, and that’s where now the actual atomic states, which has the nucleus, which is a proton and a neutron, and an electron gets so hot, that electron gets energized and then escapes, leaves the system. And now they’re charged. You have a positive nucleus and a negative electron floating out, and that happens on the order of 10,000 degrees. So way hotter than what we’re used to. But now, we’re going to go hotter. We’re going to take this plasma and go even hotter. What does that mean?

David Kirtley (01:14:14)

At that point, a lot of the way we think about temperature doesn’t really apply. The idea that you have these random motion of particles, because now they’re all individual particles moving at very high velocities. So there really is a measurement of its velocity. It’s really a measurement of how fast is that particle moving. And that’s how I really think about temperature when you get to that 100,000,000 degrees. And so it does more complex things. If you have this high energy particle, this is why we like fusion. It’s moving at a high velocity and there’s another one moving at high velocity. They will come together, they will collide, and they will fuse.

Lex Fridman (01:14:54)

But other things will happen. You don’t want to touch that high velocity particle with any kind of material, because it will collide with that material, damage that material, and usually blow off some chunks of that material. So we don’t do that. We keep those charged particles in a magnetic field. So they just bounce around and they don’t ever touch anything. And that’s really important. And so it’s less thinking about it from the way we normally think about hot and cold, and more thinking about it from a velocity point of view.

Lex Fridman (01:15:20)

So what we should be imagining is extremely fast moving, what is it? 1,000,000 miles per hour? Is that accurate?

David Kirtley (01:15:29)

That’s the right kind of order for these systems.

Lex Fridman (01:15:32)

Crazy. And so you’re looking for them to collide. Well, first of all, to get back, is there some interesting insights, tricks, anything you could say to the complexity of the problem of getting it to that high temperature quickly?

David Kirtley (01:15:46)

So, if temperature is velocity, that means they’re moving quickly over a given amount of space. Speed is distance divided by time. And so if you have a machine of a certain size and it’s moving very fast, that tells you the time that that particle is moving from place to place in that machine. And, in fact, if it’s a million miles per hour, these are on the order of 100 kilometers per second, which you can flip that around and you can say you’re moving at meters per microsecond.

Lex Fridman (01:16:19)

So feet per millionth of a second. That fundamentally tells you, and we’ve known this, as soon as you say, “I want to do fusion,” you know you need to react to the universe in microseconds and be able to understand the system at that speed. If you get it hotter, it goes even faster, and you have to go faster. So we look at those, and that’s how we think about the systems. We measure everything in microseconds, not in seconds. So when you do fusion, it’s pretty wild. It’s literally a flash. Fusion happens. And it’s over. You start it, you do a lot of fusion, you recover energy from it, and then you turn it off before the human eye can really respond. Fusion control and simulation

Lex Fridman (01:17:00)

And there’s a computer managing all this. How do you even program these kinds of systems to do the switching? Is there some innovation required there?

David Kirtley (01:17:09)

I’m continuously amazed by what the pioneers in fusion were able to do before the computer existed, because they had to control things at this scale. But maybe it was pretty hard, and why we’ve been able to take what they did and build on it, is because now we use modern gigahertz-scale computing to be able to do this. Even when I started my career, we talked about megahertz processors. Megahertz is microseconds. That’s great. You’re kind of at the border of fast enough, but you can’t do computation at that speed if all it can do is respond in one microsecond. But now gigahertz means I can do a thousand operations in that one microsecond, so I can do more useful things. So we use mostly…

David Kirtley (01:17:55)

This is way too fast for any human to respond to, so we use what’s called programmable logic. We program in sequences to the fusion system to be able to do this reversal. We pre-program it and then we run a sequence, and then fusion happens. In this sequence programming language, we use a variety of them. Some of the fusion codes are actually written in Fortran still.

Lex Fridman (01:18:17)

Nice.

David Kirtley (01:18:18)

A lot is now, more and more, run in Python. So we do a lot of Python. We do some Java, and because of the speed of this, it’s a lot of assembly language programming. So we go right to the assembly level of the programmable logic FPGAs and we program those. To be able to run one of these systems, we typically have a series of electrical switches that turn on this electrical current. Those are controlled via fiber optic because the wires are just too slow. Fiber optic can respond, I can send photons at the speed of light. So those fiber optics can respond in nanoseconds. Then I trigger those fiber optics with programmable logic that we programmed in the hardware assembly language.

Lex Fridman (01:19:00)

As a small tangent, let me do a call to action out there. I’m still looking for the best Fortran programmer in the world if people want to talk to them, because so many of the essential systems the world runs on are still programmed in Fortran. I think it’s a fascinating programming language. COBOL too, but Fortran even more so. It’s one of the great computational numerical programming languages. Anyway, in terms of the sensors that are giving you some kind of information about the system, in terms of the diagnostics, what kind of, at this time scale, …what can you collect about the system such that you can respond at a similar time scale?

Lex Fridman (01:19:49)

So I’m also calling out for Fortran programmers, for different reasons.

Lex Fridman (01:19:53)

Yes, great.

David Kirtley (01:19:55)

The diagnostic systems are really one of the keys to how we do this effectively, because you need to be able to tell the system, “We’re going to trigger electrical current, and we’re going to do it in a microsecond, and we need to know if it’s working right.” In one of these FRC or these pulsed magnetic systems, you won’t have just one electrical switch. I’ve mentioned 100 mega amps, 100 million amps of electrical current. Even the big transistors we use can only run at 30,000 amps, so you’ll end up with tens of thousands. In fact, the systems we build now, tens of thousands of parallel electrical switches all operating in harmony together. So you need to be able to build a system, and this is what we spend a lot of time with.

Lex Fridman (01:20:38)

And I made the joke that in a lot of ways, Helion is an electrical engineering company. To be able to both program, control, and then detect how they’re operating, and do it all very fast. In a typical sequence, we will pre-program. The operators will pre-program a sequence usually fed from a numerical simulation of expecting how the fusion system will perform. We start with a set of calculations. We then pre-program all of these electrical switches to a certain sequence to be able to inject the fuel, reverse it, and then compress it up to fusion conditions. Then we trigger that, and then let it go, and measure fusion happening.

Lex Fridman (01:21:22)

But during that process, we have to be real-time recording and measuring all of the semiconductors and all of the switching in the system. I’m not going to talk about measuring fusion diagnostics. That’s a whole other thing, which we can talk about. This is just on the electrical control side. Some of the pioneering things we’ve been able to do is that real-time you’re monitoring all of these switches. You’re watching who is triggering correctly, who is not triggering correctly. And if systems aren’t working, you’re shutting this down because you want to make sure that all the sequences are operating correctly.

David Kirtley (01:21:56)

Some of the key diagnostics, it’s actually pretty amazing that even early in my career, we didn’t have a lot of fiber optics built into the system. Now it’s absolutely essential. So every one of these electrical switches has fiber optic signals going into it and fiber optic signals coming out, understanding how it’s actually operating. And real-time, all of these systems are being monitored by more fiber optics. We call these Rogowski coils, but they’re electromagnetic coils that are powered by the electrical current themselves. So as the switches are conducting, they broadcast an optical signal that says, “Yes, I’m electrically conducting,” fiber optics that come back to a central repository where we detect those signals.

Lex Fridman (01:22:40)

So real-time, we’re monitoring all of this so that we know that these systems are behaving and operating at their optimal performance.

Lex Fridman (01:22:48)

What’s the role of numerical simulation in all of this? I guess, ahead of time, how much numerical simulation are you doing to understand how the system is going to behave, how the different parameters all come together? The electrical system and how that all maps to the fusion that’s actually generated?

David Kirtley (01:23:12)

Yeah. The operation of a fusion system is pretty fascinating because all of this happens on a time scale where human operators cannot be involved.

David Kirtley (01:23:19)

…cannot really be involved. So you have to have pre-programmed the majority, we call them shots. You’re going to do a shot, and when you’re operating them repetitively and you’re running long periods of time, you still have all computers doing both the triggering and the measuring of how they’re performing, real-time the whole time. So how this typically works, at least in our systems, is that we will design a system with a combination of numerical simulation tools that we’ve developed based off of decades and decades of amazing government programs. National programs developed these numerical codes. We use a code called an MHD, magnetohydrodynamic code.

Lex Fridman (01:24:07)

And that’s for people, for the engineers out there who are used to CFD, computational fluid dynamics. This is very similar where you take the same sets of equations actually and add electromagnetic equations on top of those. And so you get magnetohydrodynamic.

Lex Fridman (01:24:23)

Are you simulating at the level of a particle? Is there some quantum mechanical aspects to this also? How low does it go?

David Kirtley (01:24:30)

We have multiple codes at different levels, because one of the main computational challenges is… amazingly, even given all that we have been building for fusion systems, computers are still not fast enough to simulate everything. And so, we have a number of codes that we use. One we call fluid codes, where you treat the ions, the electrons, all these fusion particles. You treat them as fluids, as gases, ideal gas law, with electromagnetic forces. In those, we can simulate not just the fusion fuel, which is important, but all of the electrical circuitry. We talked about capacitors, and magnetic coils and the electrical current and the switches. Well, we actually simulate the full thing, starting literally with a SPICE model.

David Kirtley (01:25:21)

More of that electrical engineering. We start with the SPICE model and use that to drive the plasma physics model, and that’s one level of simulation. We use that to do design work, and then also to try to understand how we think the machine will run. But then we go one level deeper and we start thinking about particles, and we think about the ions, and we treat the ions as particles, and we look at the ion behavior. For that one, the computational resources are several orders of magnitude larger. Luckily, a lot of the work in GPUs, the AI data center work is directly applicable to those simulations. It’s been able to speed up our work, which is pretty fascinating. That’s a whole another tangent we can go down.

David Kirtley (01:26:00)

Those hybrid codes we call them, particle and cell codes now treat the ions as particles, and that lets us measure and simulate the behavior. I mentioned the stability criteria, S star over E, the top behavior. That behavior, we now need these more advanced codes to be able to simulate, and those are more modern. We’ve only been able to apply those in practice for the last few years, actually, which is pretty fascinating that the old stability rules were built off of testing, empirical tests, where now we can simulate that, and we know why they work and how they work and we can do some predictions on them. And so that’s really fascinating that we’ve been able to push those boundaries.

Lex Fridman (01:26:40)

And what are the different variables you’re playing with? Are you still playing with topology? What are the different variables in play here?

David Kirtley (01:26:47)

Each of the different simulations we analyze and use to design different parts of the machine. So, at the MHD level where we have the SPICE, where we actually have the circuit model, now our design team uses this to design the circuitry, where we’re designing which capacitor to use, which switch to use, how many cables to use, literally to that level, how big of a cable to use. So as we’re doing power plant designs right now, those are the tools we’re using today, every day, the team is using. Then you can go one level deeper and say, “Okay, let’s use these more advanced computational tools about stability,” to say, “Okay, great, but I now know the circuitry, but let’s look at the magnetic field topology.”

Lex Fridman (01:27:30)

How do I design the magnet, the shape of the magnet exactly, the timing of the magnet exactly? I have to trigger one magnet and the next magnet next to it and the next magnet next to it. How do I have that shape and that design?” And so that’s where you’re using those more advanced tools. Now, unfortunately, those are still too slow. And so those simulations may take a day or two to run. And so an operator right now does a lot of simulations ahead of time, then collects data through their operations of the machines, making these field-reversed configurations, going through parameter sweeps. And then the simulation team then goes back and looks at that data and compares it with simulations.

David Kirtley (01:28:12)

I’m really excited about some of the things we’re seeing in artificial intelligence and reinforced learning to be able to speed up that process. So we’re watching and starting to work on that now, of can we now, rather than using it where we use it today, where we do a simulation to design a machine or a test, run the test, and then over the next couple of days compare the testing with the simulation and use that to inform what we’re going to run for the next set of tests. But in fact do it more real time, where an operator can pull up what the AI or what the machine learning would have predicted it should have done. And then use that to understand what’s happening in the actual programs, in the actual generators themselves. Electricity from fusion

Lex Fridman (01:28:54)

All right. So there’s a million questions there. So first of all, how much understanding do we have about how many collisions happen? Can we go to the fusion? How many collisions are there and how does that map to the electricity? And maybe can you just even speak to the directly mapping to the electricity, which is one of the differences between this approach and the tokamak approach?

Lex Fridman (01:29:20)

So how much fusion do you get out from these systems? And that’s really the key question. So we already talked about beta, that B squared, the magnetic pressure is equal to NKT, N being the density, T being temperature. And then we talked about fusion, where your goal for fusion is to get particles hot, high temperature, get enough of them together, density. And then you want to get them together long enough. We call that tau. So N, T, and tau, long enough that fusion happens and a lot of fusion happens, more than any of the loss rates that are happening, NTT. And in beta with B squared, you know already two of those parameters, NNT, are equal. And so that tells you right away the goal is to maximize magnetic field, absolutely maximize magnetic field.

Lex Fridman (01:30:10)

And most folks in magnetic fusion, whether it’s a tokamak or it’s Theta Pinch or it’s an FRC, are attempting to do that, maximize the magnetic field. So we’re all pushing to that. What’s really nice in pulsed systems is that we know how to do that. In fact, in a pulsed system, researchers in pulsed magnetic fields have demonstrated over 100 tesla magnetic fields in pulsed magnets. That’s much higher than you can get in a steady magnet, or what’s been demonstrated so far.

Lex Fridman (01:30:40)

Just a clarification question. So maximizing magnetic field is about the N and the T, the beta? So we’re not talking about tau yet.

Lex Fridman (01:30:49)

Not yet, but we need to, because that’s really important.

Lex Fridman (01:30:52)

Okay.

David Kirtley (01:30:53)

We can even talk a little bit further about how fusion scales. In fusion, the hotter you get the fuel, the more fusion you get. And we know that by increasing the magnetic field, B squared as NT, you increase density and temperature together. More density, more temperature is more fusion, plus more temperature is even more fusion. And so what we see is that in these types of systems, a scaling very clearly of magnetic field to the 3.75 power, or even in a lot of demonstrations, 3.77. That specific scaling.

Lex Fridman (01:31:29)

Wow.

David Kirtley (01:31:29)

That’s a very strong scaling of fusion power output and fusion reactions. And so that tells you you want to go to a maximum magnetic field as you can. Pulsed systems are really powerful. Pulsed systems have showed when you do pulsed magnetic fields compared to a steady magnetic field, researchers have shown over 100 Tesla magnetic fields, where in a steady system, people have showed in the 20, maybe high 20 Tesla systems. And if it’s B to the 3.77 power, already you can see massive fusion power outputs by doing a pulsed system.

Lex Fridman (01:32:02)

Okay, got it. So we’re maximizing the magnetic field. So that’s a number going up, super up. How do you get the duration, the tau?

Lex Fridman (01:32:11)

But then I said pulsed.

Lex Fridman (01:32:12)

Oh, I see.

Lex Fridman (01:32:12)

And pulsed already implies shorter tau.

Lex Fridman (01:32:15)

Yes.

Lex Fridman (01:32:15)

And so that is in the fusion field, the name of the game. Folks will have inertial fusion, with a nanosecond tau. Very short, but then very high pressure. They don’t have magnetic fields, but very high pressure. And then in stellarators and tokamaks, your goal is very long tau, but you’ll have much lower density, and you can’t really go too much in temperature, but they’ll have much lower density. And so where we live in the pulsed magnetic or the magneto-inertial fusion is in the middle, in extremely high magnetic fields, increasing pressure as much as you can, and then keeping them around long enough. And so that gets to the tau. That gets to that energy confinement lifetime, and also, it gets to stability.

Lex Fridman (01:33:03)

And so this is the thing that this field-reversed configuration has shown that we can build. These plasmas can last for hundreds or thousands of times the basic theory has shown that now you can have long enough lifetimes. So what that means is in a practical fusion system, there are lifetimes of these high beta pulse systems between 100 microseconds and a few milliseconds, thousandths of a second. And you hold onto it for a few thousandths of a second. You do fusion, and then you exhaust it. And so the whole process in this is we start with a magnetic field that fills the full chamber. You then inject fusion fuel. You ionize it. Superheating it now to a nice, cold one million degrees. But hot enough that you have charged particles.

David Kirtley (01:33:57)

You have plasmas. You can then start increasing the magnetic field. You form a field-reversed configuration, and then rapidly increase the magnetic field further. Increasing from one to five to 10, 20, to even higher magnetic fields. And as you do that, the plasma heats. You compress it, increasing the field and pressure. Fusion is now happening. New charged particles are being born inside this system with a tremendous amount of heat and energy, but in charged particles. And this is where the beta really works to your advantage, is that just like magnetic pressure on the outside, magnetic pressure, NKT, compresses the fuel, increasing pressure and temperature. When the pressure and temperature of the plasma increase, NKT increases.

David Kirtley (01:34:55)

It pushes back on the magnetic field, increasing the magnetic field on the outside of the plasma, and what that does is magnetic field is electromagnetic current, and current running in a wire. And what that does is push current back in the wire. And so the plasma itself now pushes back on the magnetic field, pushing electrical current out of the system and recharging the capacitors where we started this whole process.

Lex Fridman (01:35:21)

All in a self-organizing way. So I think it’s good to clarify how fusion usually generates energy, where this intermediate step of heating up water, then the steam is the thing that leads to electricity. And then, of course, the FRC method that you use leads directly to electricity. I was wondering if you could describe the difference between those two.

David Kirtley (01:35:44)

Yeah, I like the analogy of the match and the campfire, and I hear that a lot in fusion. Where a lot of what steady fusion, think a stellarator or a Tokamak, is attempting to do is take a little bit of fuel, that match, and then add heat to ignite that match, and then put it with enough fuel and in the right conditions and hold onto it for a long time that it grows into a campfire. Even if they do a good job, a bonfire.

David Kirtley (01:36:15)

It’s creating a tremendous amount of energy in that steady system. Burning fuel in the same place, generating some ash, generating a lot of heat in that reaction. And in a traditional Tokamak or a stellarator, that’s a lot of what you’re doing, is you’re holding onto the heat as much as possible to keep that reaction going. And the optimal fuel is called deuterium and tritium, where you have deuterium, which is a heavy isotope of hydrogen where you have an extra neutron, and tritium is a very rare form of hydrogen that’s an unstable form. It’s so rare it’s hard to get.

David Kirtley (01:36:53)

Where it has two neutrons and a proton, and when you fuse those together at very high temperatures, at very high densities or high enough densities and very high temperatures, they make helium, which is a charged particle, which stays inside the campfire, inside the Tokamak, continuing to heat it and stoke the flames, and it makes a neutron which leaves the system because it’s uncharged. It has no charge, and in that system, it’s actually ideal. It’s really great because in a campfire, you have this reaction going and you want to get the energy out of it. You want to use it, and you don’t want to just burn up all the fuel and do nothing. That’s not really valuable. What’s really valuable is to stand next to the campfire and get the heat, get what comes off of it.

Lex Fridman (01:37:38)

And then use that in a traditional fusion system to boil water, to heat the water, and then at 30, 35% efficiency, then convert that through a steam turbine into a cooling tower, and cool off the fuel and extract electricity.

Lex Fridman (01:37:53)

And we know steam turbines. Coal plants do this. Nuclear fission reactors do this. And so we know how to do that, and that’s the traditional way of doing it. But I think there are other ways to do it with a pulsed magnetic system. There’s one more thing you get to do, because you have this high beta where there’s an electric field and an electromagnetic force that’s now compressing the fusion fuel. It’s increasing in temperature. It’s getting hotter. It’s increasing in temperature. Density fusion is happening. New fusion particles are being born, and those particles are not just stoking the flame. They’re not just holding on to the campfire like in the Tokamak, but they’re doing another thing which is really powerful, which is they’re pushing back on the magnetic field.

David Kirtley (01:38:40)

They’re applying a pressure. That pressure induces a current. We can extract that electrical current. But it takes you into another direction, so your analogy of the campfire now breaks down, because now the campfire is expanding. It’s pushing back on something, and so now it’s the analogy of the piston engine… …As you move from the match, the campfire, to now pistons. And so in a piston engine, you use the motion of the piston, the pressure on it, and the motion of it to do something useful. And in a piston engine, it’s to turn a crankshaft and run wheels, or maybe even a piston engine to turn a crankshaft and run a generator and make electricity. And in fact, you can do it pretty high efficiency.

David Kirtley (01:39:24)

…and a generator using that method, using the expansion of that piston. What we do is use the expansion of the magnetic field to extract that electricity, and we believe you can do it with much, much higher efficiencies. In fact, there’s been theoretical papers that show not 30% to 35% efficiency, like a steam turbine can do, but 80% efficiency, 85% efficiency, extracting much more of the energy of the fuel in that process.

Lex Fridman (01:39:52)

Can you actually just take a tiny tangent… …On the word “efficiency” here? So, you said 30%, so it’s inefficient. Is that efficiency measure how much of the energy is actually converted to electricity?

Lex Fridman (01:40:06)

That measure is how much of the thermal energy that gets outside of the system is then converted into electricity, which is the thing we care about. We’re not in this to make fusion. We’re in this to make electricity. And we’re using fusion to make electricity. So from my point of view, that should be the focus: how do we get to that? So that’s the efficiency of that thermal energy that makes it out to electricity. What it is not a measure of is how much energy you put into the system and what happens to that. In terms of, you started this campfire with a blowtorch, what about all that blowtorch energy? What are you getting for that?

Lex Fridman (01:40:42)

And so I think that’s something that high beta is one more side benefit that it turns out is actually maybe the tail that wags the dog. Not only do you, at high efficiency, get out any of the new fusion energy, which is great, because that’s what you want: make electricity from fusion, but you also get to recover all of that magnetic energy you put back into it.

Lex Fridman (01:41:02)

And that’s the really powerful one, and that’s something that folks have demonstrated: over 95% efficiency, that you can put electricity into fusion and then get that electricity back out at 95% efficiency, plus some very high efficiency, maybe 80%, maybe higher, of all the fusion product electricity too. So now you’re just making a tremendous amount of electricity in one of these systems, and that has all kinds of performance and engineering benefits that are really powerful, but it also pushes you to other fuels. So we talked about how deuterium and tritium fuels make this neutron…

David Kirtley (01:41:38)

…which leaves the system to boil water to run steam turbines, but it doesn’t push back on the magnetic field. So in one of these high beta systems, it’s actually not a great fuel at all. So the other fuels that are out there are even more interesting, and one of the candidate fuels that’s really interesting is called deuterium and helium-3. And we talked about deuterium, heavy, heavy hydrogen. Well, helium-3’s nucleus is also called a helion. That’s why we named the company that.

David Kirtley (01:42:05)

It is light helium, which is… In normal helium, which is what you find in a balloon, there are two protons, two neutrons. It’s very stable and found commonly. Helium-3 is also stable, but it’s not found commonly. Fortunately, it’s lightweight, so it leaves. It literally leaves the atmosphere and goes into space, so we don’t have a lot of it here on Earth, and so you have to make it, or you have to go into space. There’s a whole other thing about where do you get it? Do you get it from the moon? Jupiter, it turns out, has massive amounts of helium-3. So when you take deuterium and helium-3 and you fuse those together, you also get that helium particle, that alpha particle, as we call it in fusion.

Lex Fridman (01:42:50)

But instead of the neutron, you get a proton, and that proton is a charged particle. It’s a hydrogen nucleus. That proton is now trapped in the magnetic field, pushes back, and you can extract that electricity. Now, there are some prices to be paid for this helium-3 fuel, but for a high-beta system like a pulsed magnetic fusion system, that’s really the ideal fuel.

Lex Fridman (01:43:14)

When you say “prices,” what is the… Is there like technical costs, or what are the prices? What shape do the prices take?

Lex Fridman (01:43:22)

All kinds of shapes: physics, engineering, technical, and business costs.

Lex Fridman (01:43:28)

Okay.

Lex Fridman (01:43:28)

And so, let’s dive in. So, yeah, we talked about how helium-3 is… From the fusion physics point of view, we talked about 100 million degrees. That’s the temperature that deuterium and tritium fusion works really well. And that’s the temperature that traditional fusion folks have really focused on getting to. That’s the threshold. When you get to 100 million degrees, you’re at the operating point of fusion, and you know it works, colloquially anyway. Helium-3 requires higher temperatures. That’s not enough. Yes, fusion happens for deuterium and helium-3 at 100 million degrees, but it’s not its optimal temperature. And in fact, in a high-beta system, the optimal temperature is higher, 200, even sometimes 300 million degrees. So you have to get to even higher temperatures.

David Kirtley (01:44:16)

Temperature’s hard, and so you have to push to even higher temperatures than you had before. And so that’s one of the downsides. The other downside can be as you get to those higher temperatures, we talked about B squared is NT. B squared is density times temperature. Well, for a given magnetic field, density and temperature are now inverse. So as I increase temperature, density decreases. And so now you have an issue of you may have fewer particles to do fusion, which means your fusion system has to get bigger than it was before.

Lex Fridman (01:44:52)

So for the same reaction rates, a helium-3 system compared to deuterium-tritium has to operate at a higher temperature and be bigger. However, the flip side is, if you can now recover energy at three times the energy efficiency, at 80-some percent versus 30-some percent, and recover all your input energy, then now it’s actually about the same size. Because it’s the same electricity output. Not energy; it’s electricity we’re worried about. Electricity output, now you can actually build systems of similar size and similar energy. Only they’re now at this much higher efficiency.

Lex Fridman (01:45:31)

Got it. Can you say more about size? What are we talking about here? Why is size an important constraint?

Lex Fridman (01:45:37)

And that gets to one of the other prices. That gets to money. So—

Lex Fridman (01:45:40)

Our goal is we want to build clean, low-cost electricity and get it out in the world, but that means it needs to be low cost. That’s fundamental. If it’s really expensive, no one’s going to buy it. And while it can be clean, it’s not going to be deployed. And so that always has to be a part of why the promise of fusion is that it can be low cost. So how do we know how much fusion systems cost? That’s a really great question. And a lot of it comes down to fundamental size, that you have to just build things. And so there’s some really first principles cost engineering you can do around power plants for fundamentally what do they cost? How much concrete went into it? Fundamentally, how big is it?

Lex Fridman (01:46:26)

And if you’re doing a good job of manufacturing, your goal is to manufacture a product for as low of a cost as you can so you can sell it for as low a price as you can. It asymptotes to the material cost.

Lex Fridman (01:46:43)

Ah.

Lex Fridman (01:46:43)

Because you never get cheaper than that.

Lex Fridman (01:46:45)

So this literally, in some sense, some sort of first-principle sense, is how much concrete—

Lex Fridman (01:46:52)

How- how-

Lex Fridman (01:46:52)

goes into building the power plant.

Lex Fridman (01:46:54)

How much concrete, how much steel, how much copper and aluminum. Different materials cost different amounts, but at the end of the day, the cheapest function is the least amount of materials.

Lex Fridman (01:47:06)

Wow. Okay.

Lex Fridman (01:47:08)

And so that’s, we think a lot about that and how we can make these systems smaller so they can be developed at lower cost. Now, there’s a flip side: you still need to produce electricity. So if you make them really small and they don’t produce electricity, there is some minimum size to fusion, and that’s really important. Fusion scientists and engineers don’t think you’d ever have a fusion generator on the back of your DeLorean, for instance. The physics doesn’t let that happen, at least physics as we’ve understood for the last, you know, 100 or 200 years.

Lex Fridman (01:47:37)

Well, there’s a lot of really interesting business questions here, because you’re basically at the cutting edge of science, of technology, of physics, of engineering, trying to basically innovate into the future rapidly. How do you do that? Because the R&D here, the research alone, is a lot of money. So what, well, I mean, what can you say about that? Like, how to be bold and fearless in pushing this technology into the future when so much is unknown and it costs so much to just do the research?

Lex Fridman (01:48:14)

So I think about this in a couple of ways. One, the need. We look to the world and we know the world needs clean, low-cost, safe electricity. And just to meet our needs today, and not to even talk about the needs of tomorrow or the needs of AI or any of the growth that’s probably coming. Just to meet today. But fundamental to that is it has to be a product that people will buy. It has to be a generator that is making that electricity at low cost. And it’s got to be soon. So a lot of what I think about is how do we do those two things together? And a lot of that is scale, and a lot of that is thinking about… And not big scale. In fact, it’s the opposite of that. It’s small scale. It’s how do you build a product that’s mass-producible, that you can build quickly and learn quickly?

Lex Fridman (01:49:18)

And what I’ve found in my career is that they’re actually the same thing. And that the faster you can build a thing, the faster you can learn if that thing works, the faster you can now iterate on that and build the next thing. And so what I have spent my career building is teams of humans and a company that are builders, that can build high-technology things quickly. If you want to do R&D, you don’t want large-scale, multinational, complex, huge systems.

David Kirtley (01:49:58)

You want to actually take the smallest thing you can build that accomplishes the mission—and in fusion, there is a minimum size—but accomplishes the mission, and then build it quickly and build whole teams around building it quickly and incentivize folks to move quickly, iterate and learn. And the irony I think of one of the things that I’ve discovered is that by focusing on manufacturing, by focusing on low-cost, very rapid manufacturing, you actually get to do science faster. And at the beginning of my career, I would never have guessed that. I would have thought the way to do science is to make a giant demonstration particle accelerator somewhere.

David Kirtley (01:50:37)

Like to make a large complex science experiment is the best way to do science. And what I’ve found is actually small, iterative, just building as fast as possible gets you there faster, because you can learn, you can build, you can iterate. You can solve the problems, and then you can learn the fundamental physics, learn the scaling, learn the FRC, and the B to the 3.77 power and learn those things way sooner than if you would have just started on one mega project and then waited decades to get to the answer.

Lex Fridman (01:51:11)

There’s a profound truth in that, something about the constraints of pushing for the simple, for the low cost, for the manufacturable. That pushes everything, pushes the science, pushes the innovation. In fact, you should maybe explain that you’re, I believe, on the seventh prototype. The rate of innovation here is insane. Can you maybe speak to all the different prototypes you went through, what it took to just iterate rapidly? And maybe it would be really interesting for people, like what can you say about the teams that’s required to make that happen? Like what kind of people are required to make that happen at that fast rate? And we’re not talking about software here. We’re talking about everything, the full stack.

Lex Fridman (01:51:58)

All the way down to the physics at 100 million degrees, at speeds of 1 million miles per hour. I mean, it’s insane. Anyway, so how do you iterate the prototypes, and what kind of teams make it happen?

Lex Fridman (01:52:13)

So at Helion, we’ve built seven systems. The first six were a series of prototypes that we built end to end that were focused on scaling the process of making these field-reversed configurations, compressing them to thermonuclear fusion conditions, and demonstrating that you can do fusion and then increasing the scale, increasing the temperature and the energy. The very first ones were named after beer. Actually, the most successful was the Inductive Plasmoid Accelerator, the IPA. And it was the first system that showed that the team could make these FRCs and hold onto them and understand some of the stability criteria, the heating criteria. And then we started increasing the field. Now, okay, great, we can hold onto one of these FRCs.

David Kirtley (01:53:03)

We know how long and how to make them, but now can we squeeze on them and start doing fusion? Increasing in pressure and temperature. What we noticed is, you know, machine after machine, we always used Starbucks. We were in Redmond at the time, Redmond, Washington, and Starbucks cups sitting on top of the machine as the, this is the scale. They were too small to have a human really in the picture all the time, so the Starbucks cup was enough. And so then we switched to Tall, Grande, Venti. And then the biggest, Trenta, was the biggest system that came online in 2020.

David Kirtley (01:53:41)

That was a system that showed 100 million degrees and was the first system that did deuterium and helium-3 fusion. In fact, as far as we know, the only bulk deuterium-helium-3 fusion that has been done and also showed the 100-million-degree fusion temperatures from an FRC. And throughout that time, the earliest work was government funded, government grants, SBIRs and other types of government grants. And actually the team involved, myself and the rest of the founding team, were really good at winning government programs, doing fundamental science, but moving very quickly. And there’s a lot of ways to think about how to iterate and how to build quickly. I want to talk about the teams first, and then we can talk about some of the technology.

David Kirtley (01:54:25)

It uses to do that, but a lot of it is thinking about if your goal is to get the product, electricity, out to the world as soon as possible, then you should be looking at everything you do through that lens. And so that’s thinking about the materials you choose. You want to, at every turn, choose commonly available materials. If you have to wait for a supply chain for an ultra-rare material, it’s going to take you a lot more time. And so do everything you can to engineer a system that uses simple aluminum alloys, simple copper alloys. And if you have to use tungsten, and maybe you have to use tungsten in some of your systems, which is a hard-to-find alloy, make sure you’re using commonly available thicknesses of tungsten sheet.

David Kirtley (01:55:07)

You know, those kinds of engineering analyses and thought processes at every step. And that’s how we built these systems, from IPA to Venti up to Trenta, was always looking at how do we build systems that are easy to build and mass-produced? Because this is the other thing that I don’t know that early in my career I’d have predicted, is that by making a hundred of a thing, you can actually make it faster than if you go make one of a thing.

Lex Fridman (01:55:36)

And that’s because when you look at our fusion systems, we talked about these big magnets, and you could build one giant, big, complex, hard-to-make magnet that’s heavy and you have to move it around with a crane and requires very complex machining by ultra-rare CNCs, or you could then make that out of a composite of 100 smaller magnets. Each of those magnets now can be made on a simple machine. Each of these magnets can be picked up by a human, they’re light enough. They can be made and manufactured and mass-produced, and that’s what we did. And that was our whole design philosophy on these machines is, at every turn, how do we go faster? A classic one that still to this day I push the team on is, again, thinking about how do you move fast, eBay.

Lex Fridman (01:56:31)

We buy, and I don’t know that I’ve ever said this publicly…

Lex Fridman (01:56:35)

Oh boy, here we go. This is great.

Lex Fridman (01:56:38)

We spend a lot of time on eBay.

Lex Fridman (01:56:40)

You got to find a way, yeah.

David Kirtley (01:56:41)

You got to move, and here’s an example: we use a vacuum pump because in these systems you got to pull out all the air. So we use a vacuum pump called a turbomolecular vacuum pump. This is a commodity. This is used in a variety of particle accelerators, scientific applications. There are many of them. They’re robust. They last a long time. They also have a very small supply chain. So if you want to buy a brand new turbomolecular pump, you can, and you might wait nine months from the manufacturer…

David Kirtley (01:57:09)

…to go make one for you and deliver it to you. But I can go today and get the same model that was made 10 years ago and get it on eBay today, right now. However, it might not work. You don’t know yet how well it works or how clean it is or any of those things. And so what we do is you don’t go to eBay to save money. It does, it’s cheaper, and that’s great. But you can also go and get three of those turbo pumps that are sitting on eBay right now, bring those in-house, test them. Maybe only one of them meets the specifications you need, but guess what? You just got a pump in two weeks instead of nine months. And you got it, and it’s in the door and it’s operational and it’s running, and you’re moving.

Lex Fridman (01:57:50)

See, I love this. I love that kind of stuff. One of the only people I’ve really seen do that is Elon. He put together that cluster in Memphis in a matter of weeks, which is nothing like that has ever been done before. And this eBay way is really the kind of thing that’s required to make that happen, as you shortcut the supply chain.

Lex Fridman (01:58:15)

And everywhere you can, you still have to deliver the working product, right?

Lex Fridman (01:58:19)

Right.

David Kirtley (01:58:19)

That is, cannot sacrifice the quality. But do you really need the shiny brand-new one when the used one is going to do the job? And we think about that across the board. Do we take the best plasma diagnostic, the most sophisticated plasma diagnostic in the world that has an accuracy of within 3%? And it’s going to take me three years and maybe a few million dollars to go build. Or do I take a technology from 10 years ago that’s 5% accurate, that’s good enough that I can go build in a month? And the answer for us, for Helion and for the team that we’ve put together, is that scrappy, “I want to just solve the problem. I don’t need necessarily the best solution, but let’s go make it happen.” And so that’s something that we routinely do.

Lex Fridman (01:59:09)

I think sometimes I have challenges with my academic colleagues on this, is that we have a difference of opinion. Because that 3%, well, that’s way better than 5%. So shouldn’t you do that? You’ll know your data better. But 5% is good enough. Now, 50% would not be good enough. And so that technology wouldn’t have been applicable. And so finding that middle ground is a hard thing to do, and never compromising on the quality and the safety. Like it’s got to work and it’s got to be safe, but can you still go fast?

Lex Fridman (01:59:40)

But in general, just having a culture of pushing the rate of iterations here.

Lex Fridman (01:59:45)

And building the team that wants to go build things. Everyone at Helion, or at least the vast majority of Helion, we hire engineers, scientists, technicians, and machinists, are hands-on builders. The company at Helion is very weird for a fusion company. Today, we are 50% technicians, not scientists.

Lex Fridman (02:00:06)

Nice.

Lex Fridman (02:00:06)

And we have a ton of scientists, because the science is critically important too, but they’re supported by a huge manufacturing company. Our goal is to build as fast as possible. Some of the other things we try to do there, vertically integrate. This is to your point on Elon Musk. This is one of the things he’s focused on at his companies, has been how do you bring inside the critical things that are going to drive timelines, the things you can’t just go buy as a commodity product and get it here soon, and make sure that you can go build those fast. We’ve done a number of key vertical integrated manufacturing lines at Helion. I think we may be the only fusion company with a conveyor belt.

David Kirtley (02:00:49)

Actually, our second one just came online now, where we literally have our production line manufacturing power supplies at Helion so that we can move at maximum velocity, rather than finding an external consultant or an external supplier to go do those.

Lex Fridman (02:01:06)

Well, I love it. Builder-first company, and you’re also thinking about manufacturing… …Throughout all of this. I’m looking at the photo of Trenta. It’s beautiful.

Lex Fridman (02:01:16)

And you can actually, I can point out on this picture one perfect example of what I’m talking about. So on the end is a green structure, green fiberglass. This is called G10. Actually, ironically, one of the main structural elements we use is this G10 fiberglass material. It’s the same thing that’s in PCB boards. It’s the same substrate that’s in every circuit board. We know it’s strong, it’s good with electricity only we get big pieces of it and machine it. But even in the end, you can see the bolts halfway through. There’s nine bolts in the middle there.

David Kirtley (02:01:52)

The standard piece of G10 was not big enough to fit the end of the machine, so we could have had one custom manufacturer manufacture a brand-new piece of a custom size, build a new mold and a new machine. It would’ve taken, I don’t remember anymore now, but probably on the order of usually these are about six to 12 months, or I could go to a supplier off the shelf, have that delivered in a week, and now machine it with all the bolts in between.

Lex Fridman (02:02:18)

And then in-house, have the G10 machine shop that can now machine the bolt holes to actually bolt those pieces together. That took extra engineering and having really clever and brilliant mechanical and structural engineers to figure out how to do that and still meet the needs of the fusion system. That’s what we try. That’s the kinds of teams we try to build at Helion, is folks that want to really get their hands dirty, get hands on, build things, move quickly. And everywhere you can, without sacrificing quality or safety, take shortcuts. That’s the name of the game. We’ve got to get fusion online as soon as possible. First fusion power plant in 2028

Lex Fridman (02:02:57)

Yeah, this is really exciting and really inspiring. So I have to ask then, what timeline do you think, like first working, out there, nuclear fusion power plant? When do you think?

David Kirtley (02:03:10)

Yeah, so what we’ve been able to do is rapidly build, every few years, bring a new fusion system online. In 2023, we signed a deal with Microsoft to build a power plant for Microsoft for one of their data centers. This is a power plant that is plugged into the grid, generating electricity from fusion, with a very tough, ambitious timeline of 2028 for the first electrons from that power plant.

Lex Fridman (02:03:38)

And that power plant will be powering a data center.

David Kirtley (02:03:43)

That power plant will be powering the grid that the data center is plugged into. We can get into the details of how the power grid works and such, but yes, Microsoft will be buying the power from that power plant.

Lex Fridman (02:03:55)

Props to Microsoft for creating a hard deadline. I love it.

David Kirtley (02:03:58)

They are. They are. It is daily that we think about that deadline. We had been working with them on and off through all of those machines, through Grande, Venti, Trenta, so they had seen us build, hit milestones, show that we can do fusion, scale up by orders of magnitude, and then access these advanced fusion fuels. So they had seen all of those things and seen the manufacturing we built. We’re already, right now, building the manufacturing to support that power plant. We’re doing that today. We started two years ago on doing the work around siting, around the interconnects. How do you plug fusion in? What does it look like? How do you site it? What are the environmental consequences? Who’s going to regulate it? All of those things.

Lex Fridman (02:04:45)

So we’ve spent a lot of time already and we’re on our way, and it’s going to be hard. No joke about it. This is tough, and it’s something that I think about every day.

Lex Fridman (02:04:55)

I’m sure you’ve had a bunch of people probably still tell you that this is a pipe dream, that this is impossible. Are there days that you and the team think that this is indeed impossible, then you wake up the next day and you’re like, “All right, we’re going to do it anyway.”

David Kirtley (02:05:08)

I mean, that’s the thought process. That’s the mentality. We’re going to do it anyway, let’s go do it. The world needs it. There’s no physics reason this can’t be done. Now it’s a question of how fast can you build it? And can you engineer it to be as efficient as it needs to be? And those are… engineering and manufacturing are ridiculously hard challenges. So do not short sell that. But that’s the goal, and that’s what we get up every day thinking about. This is something I was actually just thinking about and talking with some of my team in the last few days. We certainly have people that say like, “No, this can never be done,” and we had that before.

David Kirtley (02:05:46)

We had that at the very beginning of, “I want to go merge these plasmas together,” and folks said, “Nope, that can never happen,” and we went off and did it. And, “You can’t compress an FRC because it’s unstable.” In fact, I actually still hear that. “FRCs are unstable.” And I say, “Yes, I know. Now let me introduce you to S* over E, and 20 years of studies on what we know about that and how we can combat that.” And so we’ve been able to show, through lots of skepticism, that we can still build and iterate. And there are things I don’t know. Let’s just be totally honest. As we’re going to go build these things, we’re going to discover new hard problems. If we’re not doing our job, if we’re not discovering new hard problems, we probably didn’t push hard enough.

David Kirtley (02:06:30)

We probably didn’t push fast enough. And I think that’s really critical: that we build the team and we do the hiring to make sure that everybody is doing their problem. Now that doesn’t mean it’s not a hard challenge, and to keep folks motivated. Helion now is over 500 people, but when we built Trenta, we were 50 people.

Lex Fridman (02:06:55)

Okay.

Lex Fridman (02:06:56)

So now there are over 300 humans working at Helion that didn’t see us build a system from a computer model, bring it online, and do fusion with it. But even already for Polaris, there are lots of humans that started for our seventh-generation system. When we were running Trenta, doing fusion, they were able to see that, see the measurements, know we were doing fusion. But yet, this next machine was just a simulation. And so, seeing that get built, seeing that, it’s just awe-inspiring for folks. And I’ll tell you, the first time that it comes online and flashes pink and you see that fusion glow, it’s awe-inspiring. It’s awe-inspiring.

Lex Fridman (02:07:41)

I love that.

Lex Fridman (02:07:42)

I’ve, uh-

Lex Fridman (02:07:42)

The fu- the fusion glow, yeah. Yeah.

David Kirtley (02:07:44)

Everybody changes their Windows desktop backgrounds to now the fusion background, the plasma glow.

Lex Fridman (02:07:50)

So how can you actually see it?

David Kirtley (02:07:53)

A couple of things. So one, to get access to it, we have windows. We have small windows all the way around, that we look into it with cameras, spectroscopy, lasers, other kinds of scientific diagnostics that we use to measure. And so you see the light emission through that. But also, it’s very bright.

Lex Fridman (02:08:11)

And so, the actual vacuum vessels themselves that we use are ceramic. There are some versions of silicon and oxygen, typically quartz, but there’s also some other sintered materials. And it’s so bright that they can shine through those materials, and so what you see is you see the light of not fusion. When fusion’s happening, thermonuclear fusion is so hot that the light is in the X-ray spectrum, and the human eye can’t see that. But as your ice-cold, one-million-degree plasma, when you’re just getting started, it’s emitting photons in a range and light in a range that humans can see. And so you see that bright, purple, fuchsia color.

Lex Fridman (02:08:51)

And this would be, if you’re doing actual cameras, this would be, like, extremely high-speed cameras, that kinda thing?

Lex Fridman (02:08:56)

We have high-speed ones and low-speed ones. The traditional SLR cameras, which are the ones that represent the right color, all they catch is the light, the integrated light, the flash. They don’t know, they can’t see the plasma forming, accelerating, compressing. They can’t see any of those things. They just see all of it integrated into one bright flash. But the high-speed cameras, they can see that. And so the high-speed cameras we can use to actually measure that. In fact, we put special filters on them to measure different wavelengths of light, so we can tell is it the hydrogen, is it the helium, is it the helium-3? Who’s emitting the light? When are they emitting? What particles are emitting the light and when?

Lex Fridman (02:09:38)

And so by using those advanced diagnostics we can now take movies of that, though it’s not as great as just seeing that flash. Energy needs of GPU clusters

Lex Fridman (02:09:46)

Yeah, I mean, it’s beautiful, right, that human beings are able to create something like that. It’s truly beautiful. Just out of curiosity, are there some interesting intricacies connecting a nuclear fusion power plant to the power grid? Like, are there some constraints to the old-school nature of the power grid in, let’s say, in the United States? Like, how do you get that Microsoft thing you mentioned, how do you get from the nuclear fusion power plant to a computer with some GPUs? How do we make that connection? Or is that a trivial thing?

David Kirtley (02:10:20)

None of this is trivial, but there are, I think, simple ways, and there are some really interesting engineering ways to do this. So just from the fundamental basics as we’re doing fusion, we push back on the magnetic field. We recharge these capacitors that start where the electricity started from. And that electricity then sits on a capacitor at high voltage, DC voltage, that’s steady. At that point, it’s reasonably easy to make 60 hertz power, make traditional AC power. It’s the same way as you can take electricity in a battery and use an inverter and just invert that to AC power. And large-scale grid inverters, we know how to do pretty well.

David Kirtley (02:11:04)

One of the unique things about a pulsed version of this, because it’s pulsed at a repetition rate between one and ten times a second, we can adjust the power output. And so as the grid needs more power, we can actually dial it up and down. And we’ve been able to demonstrate that with our fusion systems. The smaller ones, the smaller plasma systems, we’ve gone from zero, from off, to all the way to 100 times a second, and shown we can do 100 hertz operation. In fact, that system we ran for over a billion operations, and just ran it steady all day long.

Lex Fridman (02:11:36)

So each individual pulse is independent in some sense.

David Kirtley (02:11:39)

Each individual pulse is different. Where you put in your fuel, you do fusion, you exhaust it…

Lex Fridman (02:11:44)

…through those pumps from eBay, and then power output and electricity output.

Lex Fridman (02:11:50)

Oh, wow.

Lex Fridman (02:11:51)

But there’s probably some more clever ways to do this, and when we founded Helion, the goal was to build low-cost baseload electricity. And what we started to see working with Microsoft and others now, is that data centers are going to be one of the biggest power needs in the future. And we know that’s coming up. And what’s really unique is that power in this form is direct recovery, not the steam turbine part, but direct electricity. Electricity is already DC, which is steady, which is what computers really want anyway. And so are there really unique ways to take DC power sitting on this capacitor, and rather than going AC to the grid and having all these transmission losses, just going direct DC to the data center? Can you plug right in?

Lex Fridman (02:12:38)

And so that’s some of the things that my team is looking at now, is, can you do that direct DC conversion at super high efficiencies and run those GPUs directly? That would be really powerful if we could figure out how to do it. But those are some of the things that I think there might be some unique ways that fusion and data centers can really couple together. There’s a whole cooling part to it too. Most of my cooling is cooling semiconductors and cooling power switching, just like a data center. So there’s a lot of interesting engineering ways that we can bring those two together.

Lex Fridman (02:13:11)

So a deeper integration between the power plant and the thing that it’s powering. And it does seem like in the future, quite possibly, a lot of the energy that’s needed will be for compute, for AI-related applications. So if you just look out into the future 10, 20, 50 years from now, do you see nuclear fusion as a thing that powers these gigantic data centers of millions of GPUs? Just basically, the surface of the Earth covered in compute and nuclear fusion power plants. Maybe that’s 100 years out.

Lex Fridman (02:13:55)

So when I talk to AI experts, they talk pretty routinely about the power needs for AI. And in fact, in the same way in manufacturing that the cost of any one thing asymptotes to the raw material, for AI, the cost of computation asymptotes to the power… …To the cost of the electricity. And even more, that electricity’s concentrated. It’s in that AI data center, that brain where all the power is, and you really want a lot of high-energy density. You want power generation right there…

David Kirtley (02:14:32)

…on-site. So it seems like, just take those two facts, a really nice match between fusion, which is baseload, high energy density, can be sited most places, and a data center, which is going to be high energy requirements in a local location, and large amounts of it. There’s been predictions recently from energy institutes that suggest we will have growth that, rather than a 2% growth per year in electricity, may be a 4 or a 6% growth in electricity due to data center use. I think that is probably wildly underestimating where we’re moving. And, and so-

Lex Fridman (02:15:13)

Oh, man.

Lex Fridman (02:15:14)

And so the idea that AI can grow human cognition, and our ability to solve problems, we can’t let it be limited by power. And so I’m going to push as hard as I can so that that’s not the limit.

Lex Fridman (02:15:30)

Do you ever think about like 2050 or something like that? I know you’re focused on a few years out, just getting… A fusion power plant working. But do you ever think about a longer-term future? By what year do you think there’ll be over 1000 nuclear fusion power plants?

Lex Fridman (02:15:50)

So I tell the team that if we demonstrate fusion one time and that’s it, then we failed. But that’s not enough. The universe is powered by fusion. Humans need to be harnessing this, and can harness this for our society, for the good of society, for the good of technology. And so that’s something that we push towards. And in fact, it’s baked into how we design these machines.

David Kirtley (02:16:23)

Coils are mass produced. Capacitors are mass produced, and we make them. All across the board is thinking about not what the next system’s going to be, but making sure we’re building the manufacturing and the infrastructure to build all of those systems. So we had a call from the White House a number of years ago for the Bold Decadal Study in Fusion of how do we get fusion. And it was Helion and a variety of other companies from the fusion industry. And it’s pretty awesome to be able to say there’s a fusion industry now. That it’s not, it’s not-

David Kirtley (02:16:55)

just a one-off thing, or there’s a fusion experiment or somebody has a prototype. But, like, there’s an industry. That Helion has competitors. That’s, that’s great.

Lex Fridman (02:17:04)

I’ve never heard anyone so excited to have competitors. But yes, that’s like a serious thing. That’s a real possibility. Yeah.

Lex Fridman (02:17:12)

And the goal was, how do we not just demonstrate fusion in the next decade, but meaningfully deploy it and start to answer? We have 4000 gigawatts of installed fossil fuel capacity.

Lex Fridman (02:17:25)

How do we, how do we start replacing that with fusion in a meaningful way? And how do we get to not just making a generator every few years? But we want a factory, a Gigafactory of these fusion generators rolling off the line, one a month, one a week, one a day. That, that’s the kind of plans that I task my supply chain team with. Like, how do you do this? How do we actually go build this? How do we go build a Gigafactory so we can have 50 megawatt generators coming off the line, being deployed on a truck, and then driving off the factory every day? And it’s a tough challenge. I see what we’ve, what others have been able to do in rockets, in electric vehicles, turning around huge factories.

David Kirtley (02:18:12)

We know this can be done, and so for fusion, the call is there, and the market is there too. If you can get electricity generators cheap enough, then it’s worth doing.

Lex Fridman (02:18:23)

Yeah. I mean, all of this is really exciting and inspiring what you’re doing. And obviously the world needs it, and the more cheap energy we have of this kind, that we described, clean, and it’s not constrained to geographical locations and so on, first of all, that alleviates a lot of the tension in geopolitics. But second of all, it enables a lot of the technological breakthroughs on the AI side. On all the different things that we use compute for. It’s really, really exciting. So yeah, I hope there’s like millions of them in the coming decades.

Lex Fridman (02:19:00)

And so if we can get to that, if we can get to making a generator a day, you’re not- Now talking about hundreds a year, and you’re deploying them.

Lex Fridman (02:19:08)

And deploying them is also hard at this scale. How do you go and deploy power plants and deploy generators at this scale and do it quickly? Interestingly, data centers are a little bit of a nicer challenge in that way, because we wouldn’t build one 50-megawatt system and have to go build a site for it. We’d build a site and put 100 of them on that site and have large amounts of power for that large data center. So that in some ways is actually the chicken and egg problem of how do you go deploy hundreds or thousands of fusion generators. Data centers are an interesting application where very immediately you need a lot of power in a very small area. And you can go, you can go do that. Now, what does that mean? That means I’m going to need more than two conveyor belts, that’s for sure.

Lex Fridman (02:19:52)

Yeah, manufacturing is really hard. But like you said, the fascinating thing is it’s hard, but as you’re doing it, you figure out all the other things: the science, the physics, and everything. The innovation is accelerated when you have to manufacture at scale. It’s actually fascinating to watch. You see that in the space industry as well. When do we humans get to Kardashev Type One civilization status? And when do we get to a Kardashev Type Two? Kardashev scale

Lex Fridman (02:20:25)

So the Kardashev scale: a Kardashev Type One civilization is when humans are either catching or generating as much power as what’s incident on the Earth from the sun. Type Two is the next big one, where you’re catching as much energy from all the way around the sun, so massive amounts of energy. A lot of times, people talk about it as incident, as in you had solar panels the size of the entire planet blocking all of the sun. But I think really, you should be thinking about it as what can we generate? What can we make here on Earth? And what we know is that, you know, we’re only a fraction right now of Kardashev Type One, and we’ve got some work to do. And there are not a lot of technologies that can get there, just from the point of view of the fuel.

Lex Fridman (02:21:14)

Right.

Lex Fridman (02:21:14)

But if, as some research says, that there’s 100 million to a billion years of fusion fuel on the Earth, we have room to go, and that’s at today’s use. So 100 times today’s use, we still have tons of fuel. Let’s go do it. And what does that unlock? What does it unlock to have power 100 times the output that we actually do here on Earth right now? And I think that’s pretty transformational. Do we have those huge AI data centers? Do we have brains that can now think at rapid speeds and now innovate? I think that’s a pretty powerful future.

Lex Fridman (02:21:48)

Yeah, I can just imagine a giant AI brain and rockets just constantly shipping more and more humans out into space, into colonizing space, and we’re expanding out into the universe. I mean, it’s a beautiful… obviously there’s a lot to be concerned about. Technology in itself is always a double-edged sword. There’s always a concern that we humans, in the power we create, will also destroy ourselves in obvious ways and less than obvious ways. I’ve been spending a lot of time in nature.

Lex Fridman (02:22:25)

And you become distinctly aware that there’s something truly special about the simplicity, the balance that is achieved by nature. In some sense, we disturb that balance by creating sophisticated technologies. But in another sense, we’re building something in the spirit of nature that’s more and more beautiful and allows us humans to flourish in a richer and richer way. So, a double-edged sword.

David Kirtley (02:22:52)

I think a lot about what vast amounts of low-cost energy, low-cost electricity enable, and how that works with nature. If you have power—and this is why, one of the reasons we love fusion—it’s energy-dense. So a 50-megawatt facility, we believe, fits in a 27,000-square-foot building, on the order of an acre, for 50 megawatts. Compare that to solar, which would be 2,000 acres, at least in Seattle. And what you can do there is transformational. And a lot of folks talk about desalination and clean water so that we can be in places where there’s not a lot of water and those things. I actually think about food, ironically, is that how much of the Earth’s surface that used to be nature is now farmland.

Lex Fridman (02:23:45)

And we need it. We’re going to grow food because humans need to eat, and that’s really critical, but it’s about five feet tall all over the Earth. Why can’t you do it at 500 feet? Why can’t you build a building where you’re actually growing plants inside? I spend a lot of time thinking about growing plants, ironically, at high densities of food, so that we can eat and we can exist and we can coexist in a way that’s energy-dense and rich. You mentioned actually going to space. How do we go to space now? We take methane fuels or hydrogen fuels and we burn them and we launch a rocket.

David Kirtley (02:24:24)

There are all kinds of cool beamed rocket technologies that I looked at early in my career, where you can beam microwaves, and so you have a microwave craft that doesn’t have to burn any fuel. And so if you have really dense, really good power on Earth, you can beam it to that microwave craft. It can now use electricity as its rocket fuel. And so there are some really powerful, interesting things you can do. Even deep space, it gets also more enabling, but even just launching from Earth.

Lex Fridman (02:24:55)

And so I think it opens up things we don’t really even think about, but it’s just been theorized, “Wow, if I had massive amounts of power in a small place that is low cost, this is what it could do.” But I’m excited by what it can unlock that we can even think about now, but even what we can’t think about or we don’t know yet.

Lex Fridman (02:25:14)

Since you mentioned propulsion, is there some interesting use, possible use of nuclear fusion in propulsion, whether it’s getting off of Earth or in going into deep space?

David Kirtley (02:25:26)

I mean, that’s … Honestly, in a lot of ways, that’s how I got into fusion, is thinking about that intersection of energy and space travel.

David Kirtley (02:25:35)

When you are in the solar system around Earth’s orbit, collecting the sun’s energy makes a lot of sense. It’s there, it’s free. When you’re in space, you get a lot more of it because the atmosphere’s not blocking it. That’s why spacecraft run on solar panels. But if you want to go further out, the sun’s irradiance falls off as R-squared, radius squared, and it’s a long way out there. It doesn’t take very long before there is not a lot of energy anywhere from the sun. So you have to bring it with you, and in space, mass is expensive. Mass is hard. That’s the rocket equation. Being able to bring high energy density fuel is really exciting and that’s what fusion enables. But here’s one of the challenges.

David Kirtley (02:26:22)

If you make electricity from fusion using a steam cycle, you need something cool. You get hot water, you now have to be able to cool it. In space, there’s nothing to cool. There’s no working fluid to cool off of. So, a lot of the steam-based systems in fusion don’t make sense for space. That’s where some of this direct energy, this energy efficiency, matters. It actually comes to some of the origin story of the team that founded Helion. Before spinning off Helion to focus only on fusion, we worked on a mix of things: advanced materials, rocket propulsion, fusion rockets, fusion materials, all of those things.

Lex Fridman (02:27:07)

Nice.

David Kirtley (02:27:08)

One thing that people in the aerospace field know, especially if you’re in deep space, is you can’t waste anything. Every watt of electricity you make, you better use, because it was expensive to get it, or the solar panel. Every ounce of every joule of heat, every watt of heat you make, you have to reject with a radiator, and it’s super expensive and heavy. So you build in space as efficient as possible. You recirculate your water and your air, and all of those things, you’re efficient. It’s something we brought into thinking about fusion energy efficiency, is that you want to… If my goal is to make the product, what’s the product? The product is electricity. Don’t waste any of it.

David Kirtley (02:27:53)

Recover every watt you can by recovering electricity directly. Recover every watt of electricity from the fusion process as efficiently as you can and you end up with, just like in space, systems that are smaller, have higher performance, and can deliver more, whatever the mission is. In our case, the mission is electricity.

Lex Fridman (02:28:13)

When you look out there at the stars, I’m really confused by what’s going on, because I think there is for sure thousands, if not millions, of advanced alien civilizations out there. I’m really confused why we have not, in a definitive way, met any of them. So again, continuing the pothead questions, what energy source do you think they’re using? If what I’m saying is true, that there is alien civilizations out there, do you think it’s, like, pretty certain that they, in order to expand out into the cosmos, they would be using nuclear fusion?

David Kirtley (02:28:50)

It’s hard to imagine anything else. Right now, where does energy in the universe come from? It comes from fusion. It comes from stars, and we know that that’s the process. So whether they’re harnessing the star itself, Kardashev Type II, or are they bringing fusion along because they want to go somewhere and they’re bringing it with them to go visit, I think that’s pretty likely. You bring up the Fermi paradox.

Lex Fridman (02:29:20)

How come we don’t see alien civilizations? Even if there’s an infinitesimally small chance that there is life on any one planet, and infinitesimally small that life grows into intelligent life, there are, however, almost infinite planets around infinite stars in our galaxy that have been around for vastly longer than we’ve been around. But we don’t see it, and I think that’s a question that many scientists and everyone has wrestled with over the years.

Lex Fridman (02:29:50)

I mean, I’m very scared by the implications of that. The scary thing is that, to the point that we made earlier, as we become more and more technologically advanced, we end up destroying ourselves. Like, there could be things we unlock, like nuclear weapons, but plus plus. New things happen as you develop super advanced systems that have close to 100% probability to destroy ourselves, destroy any intelligent being. The kind of intelligent being that’s ambitious enough to keep innovating will eventually destroy itself, will be one explanation. And that’s scary. That should be a sobering… that’s at least an inspiring, sobering thought to be careful with the stuff we create. But I also just look at humans. We create dangerous stuff.

Lex Fridman (02:30:44)

And then figure out, sometimes almost last minute, how to not destroy ourselves. We’re good with deadlines.

Lex Fridman (02:30:53)

We’re good with deadlines. We…

Lex Fridman (02:30:55)

And we’re good at surviving. Life as we know it on Earth seems to find a way, and intelligent life as we know it, human life, seems to find a way. We do a lot of painful things along the way, but in the end, we somehow survive. It’s interesting. There’s something in the human spirit… …That allows us to survive. So I have a lot of optimism that the super powerful technologies that we create will eventually lead to us still surviving for thousands of years. But then, why are the aliens not here, though? So maybe it’s also possible that it’s really difficult to traverse space. Maybe it really is that difficult. The physics makes it not easy. There’s a lot of space, and it’s just hard to travel.

David Kirtley (02:31:45)

I think as I have gone further and further in building fusion systems that work, I’ve become more optimistic around the Fermi paradox specifically. And there are several of them. I think you’re referring to something called the great filter. Something happens that filters out life. The dark forest is another philosophy around, “Sure, it’s out there, but everybody’s hiding ’cause they don’t want to be noticed.” But I think about something else, actually. The philosophy that I’ve always loved, and I’m going to pronounce this wrong, so I apologize, Matrioshka brains…

David Kirtley (02:32:22)

…is that, and that’s Kardashev level two, that civilizations get so advanced, and they focus not on expanding physically and expanding in space and expanding their reach by planting flags in new places, but grow their cognition, grow their ability to think. They grow their brain. They grow their intellect. And I feel like in the last few years, we’ve seen a massive trend that maybe this is the thing that happens, and that we do grow our intellect, and we grow the intellect of the species by AI and advanced tools. And as a society, we can just get smart enough that we don’t need to go plant those flags everywhere.

Lex Fridman (02:33:10)

And so the Matryoshka brain is a Dyson sphere where a civilization has covered the entire sun in essentially solar panels or collects its light in some way and uses all of that power to power intelligence, to power computers, and to power brains. And I think we’re a ways away from that, but maybe AI and fusion together gets you actually along that path sooner. And I’m excited by that outcome of the Fermi paradox. And then at that point, those civilizations have a star that you can’t find anymore ’cause it’s all covered and are there thinking and growing their intellects rather than actually having to physically expand.

Lex Fridman (02:33:52)

Yeah. Exploring and expanding in the realm of cognition and consciousness versus in the realm of space and time as we 21st-century colonizer humans think like. Maybe 22nd-century humans will be thinking fundamentally differently. Yeah, that’s a beautiful, beautiful vision of the future. Speaking of beauty, you’ve been doing a lot of really interesting things in a lot of interesting disciplines. What to you is… ridiculous question, is the most beautiful idea in physics and nuclear engineering, in nuclear fusion and power plants? What ideas do you just step back and are in awe of?

David Kirtley (02:34:45)

I’m continuously in awe that it works. And I know that sounds a little silly to say. But the more that I learned in my career around the balance of exactly the right temperatures where life works, exactly the right balance between the electromagnetic force and the strong force. Those are things that it’s hard to imagine are accidental. And so we talk about how beautiful nature is, but then you look at what each of the leaves on the tree really is, and each of the cells and each of the atoms and each of the quantum substructure of that atom, and I’m just amazed that all the pieces come together.

Lex Fridman (02:35:42)

We humans are somehow able to find that perfect balance where it just works.

Lex Fridman (02:35:49)

Just works. Last minute sometimes, but it does work.

Lex Fridman (02:35:53)

The kind of deadlines you’re operating, the group of brilliant people that you’re working with or operating under, it just stresses me out. But it excites me, so I’m deeply grateful that you’re doing this work. You’re one of the people building an exciting future. So thank you for doing that. And thank you so much for talking today.

Lex Fridman (02:36:14)

Thank you very much. It’s been fun.

Lex Fridman (02:36:17)

Thanks for listening to this conversation with David Kirtley. To support this podcast, please check out our sponsors in the description where you can also find links to contact me, ask questions, give feedback and so on. And now, let me leave you with some words from the great John F. Kennedy. “We choose to do these things, not because they are easy, but because they are hard.” Thank you for listening, and hope to see you next time. This entry was posted in transcripts on November 17, 2025 by

Lex Fridman

.

Lex Fridman

Host of Lex Fridman Podcast. Research Scientist at MIT, working on human-AI interaction, robotics, and machine learning. View all posts by Lex Fridman →