There's an old joke that energy from nuclear fusion is 30 years away and always will be. For many decades, the industry did little to spoil the fun, running mammoth and unwieldy research projects, hoovering up a few billion dollars every science funding cycle and announcing breakthroughs that were scientifically interesting, but did little to move us towards actual production of electricity.
Then three things happened. First, perhaps emboldened by Elon Musk's successful disruption of the space industry with SpaceX, a bunch of wealthy investors started to invest in much nimbler startups. Second, in December 2022, the National Ignition Facility at the US's Lawrence Livermore National Lab announced that it had finally got more energy out of a fusion experiment than it had put in. And third, AI exploded on the scene, offering dramatic acceleration in everything. These developments have turned the slow-moving world of fusion on its head.
Over $14 billion has now flowed into fusion startups, and that is on top of the tens of billions already committed to national and international programs, like the International Thermonuclear Experimental Reactor (ITER) in France, the UK's Spherical Tokamak for Energy Production (STEP) and a variety of US national labs. This week on Cleaning Up, Michael Liebreich sits down with the director of one of those labs.
Professor Sir Steven Cowley has been entrusted with leading one of the jewels in the US's scientific crown. He has dedicated his academic career to trying to make fusion energy work, and he's been known to joke that he hopes it does so before he dies.
Michael and Steven discuss the breakthrough experiments that have shown fusion is possible, why getting from a successful experiment to a commercial power station is such a huge leap, and whether fusion can ever deliver affordable, dispatchable electricity. Steven puts forward the idea that fusion might have a Theranos moment, where over-promises in the industry come back to bite.
Topics discussed in this episode:
Leadership Circle
Cleaning Up is proud to be supported by its Leadership Circle. The members are Actis, Alcazar Energy, Arup, Climate Imperative Foundation, Copenhagen Infrastructure Partners, Cygnum Capital, Davidson Kempner, EcoPragma Capital, EDP, Eurelectric, KKR, Mitsubishi Heavy Industries, National Grid, Octopus Energy, Quadrature Climate Foundation, Schneider Electric, SDCL and Wärtsilä. For more information on the Leadership Circle, please visit cleaningup.live.
Links:
Acronyms:
Chapters:
Steven Cowley
What I worry about is we'll have sort of Fusion's Theranos moment, because there are some really good companies.
ML
Sorry… Fusion's?
SC
Theranos moment. Where investors say "you didn't exaggerate, you actually duped us", right? And if that happens, it'll cool the whole market. And there are a lot of really good companies. But there are 60 Fusion companies, all with different ideas, right? There can't be 60 ideas that are all, you know, and certainly some of them strike me as… I don't know anybody who's fraudulent, right? But I do know people who I think are over-egging what they have.
ML
Hello, I'm Michael Liebreich, and this is Cleaning Up. We're filming this episode in front of a live audience on the opening morning of Climate Week in New York. There's an old joke that fusion energy is 30 years away and always will be. For many decades, the industry did little to spoil the fun, running mammoth and unwieldy research projects, hoovering up a few billion dollars every science funding cycle in the major economies, and announcing breakthroughs that were scientifically interesting, but did little to move us towards actual production of electricity. Then three things happened. First, perhaps emboldened by Elon Musk's successful disruption of the space industry with SpaceX, a bunch of wealthy investors started to invest in much nimbler startups. Then, in December 2022, the National Ignition Facility at the US's Lawrence Livermore National Lab announced that it had finally got more energy out of a fusion experiment than it had put in. And third, AI exploded on the scene, offering dramatic acceleration in everything. These developments have turned the slow-moving world of fusion, with its foundation stories dating back to Ernest Rutherford, Enrico Fermi, Andrei Sakharov, on its head.
Over $14 billion has now flowed into fusion startups, and that is on top of the tens of billions already committed to national and international programs, like the International Thermonuclear Experimental Reactor, ITER, in France, the UK's Spherical Tokamak for Energy Production, STEP, and a variety of US national labs. My guest today runs one of those labs. Professor Sir Steven Cowley is a Brit, but he has nevertheless been entrusted with leading one of the jewels in the US's scientific crown. He has dedicated his academic career to trying to make fusion energy work, and he's been known to joke that he hopes it does so before he dies. Please welcome Professor Sir Steven Cowley to Cleaning Up.
ML
Sir Steven.
SC
Steve.
ML
Steve, if I can call you that, fine. Thank you very much for joining us here at our Cleaning Up breakfast in New York. Let's get started, there's a little intro which describes what you do at the Princeton Plasma Physics Lab, how did you end up there? You're a Brit. How come they let you do that?
SC
Well, I got fascinated by fusion. My father introduced it to me when I was 15 years old. I think he couldn't think of anything to talk to me about, so he talked to me about fusion. And when I went to college, I decided absolutely what I wanted to do was to make fusion work, right? It was this incredible promise of unlimited energy for the rest of the lifetime of the planet, who wouldn't want to work in it? And so when I finished Oxford as an undergraduate, I decided to go to the US for graduate school. And I went to Princeton, the Princeton Plasma Physics Lab, I was a graduate student there and I got the bug.
ML
And you went there as a Harkness Fellow.
SC
Yeah.
ML
And in fact, we met because the Harkness, which is kind of like the Rhodes, except that it's for people going to the US instead of from the US to the UK, and they were holding an event because I'm a Harkness Fellow as well but I went to Harvard Business School and took a very, very different path after Cambridge. But your dad used to talk to me about fluid dynamics and the Navier-Stokes equations because he was my professor of fluid dynamics.
SC
It's a very small world.
ML
It is an astonishingly small world. And he was, I will say, he was an incredibly thoughtful and kind, obviously, enormously eminent professor. And I was a scrappy little undergraduate and he was a tremendous professor.
SC
Well, he was a sweet dad.
ML
So you ended up, you went to Princeton, but you've done other things since. So you didn't stay at Princeton, you're not a career Jersey, can I call you a Jersey boy? You've done some extraordinary other things, running the UK's Atomic Energy Authority and the Joint European Torus, the JET program in Culham in Oxford. So what came first of those?
SC
Well, it was while running UK Atomic Energy Authority, we ran JET for Europe, right? And there are really only three places on the planet that have done any controlled fusion. That's very recently the spectacular results from the National Ignition Facility in California at Lawrence Livermore Lab, Princeton in 1994, we did 10 megawatts of fusion from the TFTR it was called, and JET, which did fusion several times. And just recently in December of 2023, did 69 megajoules of fusion energy in a 5 second shot.
ML
So that's the famous 12 megawatts 5 seconds shot, quite extraordinary. We do have a rule about acronyms on this show. Now we've already said Joint European Torus is JET, but you mentioned TF...
SC
Yes, the Tokamak Fusion Test Reactor.
ML
Which is, and where's that?
SC
That was at Princeton.
ML
That was at Princeton, but has that now been dismantled?
SC
That was dismantled, yeah. What happened after that device is, I think Princeton decided that if we were going to get commercial fusion, I mean the goal is not just to do fusion, the goal is to do commercial fusion, right? And we had to bring down the cost and scale of these devices. What we have now is we can probably get to fusion, but the real trick is can we get to fusion you can afford the electricity from?
ML
You're doing my job for me because I've got here in my notes that there's two big questions that I want to see if we can answer them or if we can jointly, if answers will be an emergent property of this conversation. And one is, when might we see fusion electricity? And the other is, how much might it cost? Because the world that I live in and a lot of the people listening to this or watching this live in, is the energy system as it is, the actual energy system, the one that you want to dock all of your innovations into. So that's the challenge today.
SC
I'll unpick some of those questions. Bringing a new energy system into a mature market like we have is extraordinarily difficult because we've honed the engineering of our energy sources and reduced the cost and got there. So I think we will see fusion coming in two stages. We will see it come in as we'll make our first electricity and then the first things that will actually make a profit will be some time after that. Now, the National Academy of Sciences had a report in 2021, we changed kind of tactics, we realised that we've got to really focus on getting the first electricity. And the National Academy report in 2021 said that the US should have an aim of producing first electricity between 2035 and 2040. And we'd had no aim to do that prior to that time, no governmental aim to do that prior to that time. The idea was we'll figure out the science, and once we figured out the science, then we'll start to think about how we are going to actually make some electricity? But I think first electricity and the first time somebody makes money from fusion, they're different things.
ML
There's a visionary designer of Formula One cars whose name I can't remember, but he talks about when he joins a new team, when he gets a fresh sheet of paper, he's not looking to design the best car. He's looking to design the best platform, it's the one that will ultimately become a dominant car in the race. So in a way, that's what you're looking at is not just the single design of a reactor, but actually something that could eventually become a player in that energy system, correct?
SC
Right. Well, let's wind this back a little bit, because there were a bunch of books 20 years ago that say, how could you put a sun in a bottle? Because to make fusion work, you've got to heat the fuel to about 200 million degrees, and that looks like it's impossible. That part we've done, we've actually made some fusion. We've actually made copious amounts of fusion out of our devices. Of course, we've been consuming energy to do that more than we've actually produced, but the conditions for fusion are here, that makes me believe it's going to happen. The question is, how fast?
ML
OK, and also what sort of what will the equipment look like? And can it be, you know, can it be done commercially? It was you brought the word commercial into this so I'm going to hold you to it. But let's do the following, let's take a step back and talk about that ignition experiment. We'll start with laser containment because there are different ways of putting sun in a bottle, different sorts of bottles, different ways of doing it and I'd love to run through some of the main ones, but laser containment. So that was the National Ignition Facility, which is part of Lawrence Livermore National Lab, chiefly famous in my world for Sankey diagrams, but they had this historic Q greater than 1. In other words, more energy out than in, it was December 2022. And they did it by shooting very powerful lasers at little pellets of deuterium tritium. And that then sort of kicked off, it didn't kick as there was already investment going into fusion, but it really turbocharged the kind of arms race. From your perspective, were you surprised, did that rock your world? Or were you saying, well first, we knew it was going to happen and anyway, it's not such a big deal. Where were you running in a sense of, I don't know, is it a competing platform, the magnetic containment platform, where were you in response to that breakthrough? Or was it a breakthrough?
SC
I think it was spectacular. I mean, it's the first time, well this quantity Q is usually defined as the fusion energy out divided by the energy into the target. And that's not net energy for your whole system, right? Because you have to put electrical energy into the laser and etcetera. But the fact that they actually got that little target to deliver more energy than you put in with the laser was a spectacular result. And it has some history because the National Ignition Facility came online I think in late 2009. And it was going to have the National Ignition Campaign and the then leader of it said, you know, we'll have fusion within two weeks in 2009, within two.
ML
Two weeks? This is almost like we're going to stop a war in Ukraine in 24 hours or occupy Ukraine in three days.
SC
No comment.
ML
Oh, I forgot who your boss is!
SC
And they said this is going to happen, the calculations said it would ignite, it would actually yield much more actually than Q greater than 1, and that this was going to be great. Now, about mid to early 2011, I got a call to join a National Academy study, right. If this is successful, how will we then take this to commercial energy?
ML
And this being laser containment?
SC
Laser containment. right. But the problem was they hadn't got ignition and there was no sign of any yield of fusion from these targets.
ML
What had gone wrong?
SC
Well, what you do with laser fusion is what you actually do is the lasers shine on the inside of a little box. It's a little gold thing that's about that big. And inside the gold thing, it makes X-rays. Those X-rays ablate the outside of a pellet and that rocket effect squeezes the pellet down and when it gets small enough, as you compress it, it gets hotter and denser it has enough fusion and then it explodes. And it is exploded with like six megajoules of energy. One megajoule is a hand grenade. So there's this thing the size of a peppercorn, six hand grenades of energy right, it's fantastic. But in 2011, they were failing to be able to squeeze it down. If you take a balloon and try and squeeze it down with your fingers, you can never do it, it all squeezes out at the side. And they were getting asymmetries in the squeezing that were meaning that they could never get it dense enough and hot enough to actually do the explosion.
And what happened, a very interesting political thing happened. This National Academy review was full of very, very eminent people, we started out saying, calm down. Even if you do ignition, this is not commercial fusion so let's figure out what you would have to do if you did. But by the end of this National Academy review, because they take a year and a half, Congress is saying it's been a waste of time, let's close it. And we were saying, no, no wait, there's some good science here, they've got a wonderful laser, keep going. And gradually, year by year, they figured out what was going wrong and a whole series of steps. And then in 2022, they got ignition. Fantastic. It's the way science is supposed to work.
ML
OK, so they worked out how to squeeze the little gold, is it called a hohlraum?
SC
Well no, the hohlraum is outside. The pellet is in the middle of the hole.
ML
I was going to show off that I knew the name of something in plasma physics, but it turns out I was wrong.
SC
Sorry.
ML
This is a tough gig, cut me some slack here. So they got the Q greater than 1. And since then, they've done a number of shots, I think they've done 11 more, if I'm not wrong, or maybe it's more than that. And they've got Q of 4, as I understand it. But does that mean that that is the route to commercial fusion? What would have to happen for that to be the route to commercial fusion?
SC
So that work is funded by the nuclear weapons part of DOE. And it's funded because this helps you understand the stewardship of the stockpile, it is not actually funded to to generate an energy source. But now that they've done it, lots of people are considering how would we engineer that into a source? Because it's not obvious, right? You're making these very large explosions and to actually produce net energy, electrical energy and electrical energy out instead of a gain of about 4, you're talking of gains of 150. So each explosion would be 150 hand grenades or thereabouts, right? How do you engineer that into a commercial plant? There's a lot of work to do there.
ML
And so they're using lenses for these incredibly powerful lasers. Lasers are interesting to the military for all sorts of reasons, not least because they're a lot cheaper way of knocking out drones than shooting Patriot missiles at them. So there's lots of interest in it. But they're using they're using lenses and the lenses at the moment can't, how many shots can they do with a lens before the lens needs replacing?
SC
It's less than a handful. It depends, what happens if you have a very intense light going through a lens, if there's any flaw in the glass and there's always a little flaw somewhere, gradually passing very intense light through it that heats and causes fractures and then the lens starts to degrade.
ML
What I'm doing is I'm going to run through some of the problems they would have to solve. Because at the moment, they're doing one shot every couple of months and they need to do 600 a second and the lenses last a handful of shots. So that's how that feels like quite a big problem to solve. How does the heat get collected? How does the energy get collected? Because you've got this hand grenade that goes off, these are high energy neutrons, 80% of the energy comes out as high energy neutrons, which bury themselves in the wall of something. How does that get into the power station?
SC
So the fuels for the easiest kind of fusion, which is between one isotope of hydrogen, heavy hydrogen, it's called deuterium and another isotope of hydrogen is called tritium. Tritium doesn't exist in nature and you breed it with those neutrons that you make in your fusion reaction by bombarding lithium in the wall. That's called the blanket of the system. And so what you've got to do with those neutrons that come out of the explosion is that they've got to embed in the wall and they've got to hit a lithium in the wall and that lithium will then yield some tritium that you then put back into your reaction. That part we haven't done very much on at all. Because we haven't done enough fusion to do that research very much.
ML
OK, so because the deuterium we can get, that's heavy water, isn't it? So deuterium we can get, but the tritium we need to make because it decays so you can't, it's not in nature, we have to make it using lithium. The deuterium, I'm going to guess that quite a few people watching or listening to this might have seen the Heroes of Telemark film and that's the famous heavy water, which at the time interestingly enough, byproduct of electrolysis of hydrogen. I didn't know that electrolysing hydrogen is a good thing if you're after heavy water. But the deuterium is not a problem, tritium has to be made from these collisions with lithium. So you've got lithium in the wall, but lithium's melting point is, what is it? It's sort of 360°C..?
SC
It's 270°C, I think.
ML
270°C, 300°C and something, whatever it is. So it's less than 400°C. It's not hot enough to make the steam that you want for an efficient turbine. So you've got presumably liquid lithium in the wall, how does it hang there? How does it stay there?
SC
Not necessarily. People are talking about using lithium salt, fluoride lithium beryllium salt. This is getting very technical right now, sorry about this, I am a professor and afterwards there'll be a quiz.
ML
But that's okay because I think we'll keep going because the costs lie between the physics and the energy system.
SC
Using molten salts is something that we're talking about doing for fission, conventional nuclear power, and for fusion. So this kind of salt is called FLiBe, fluoride, lithium, beryllium. And that's one. The other one is a eutectic of lithium and lead and that's because it's not flammable like liquid lithium. So there are safety reasons for that. But one of the ones that's really interesting is to do lithium ceramics, lithium aluminate, lithium titanate. And you can make little pellets of these things and then you can cool with helium and then you can go to a high temperature wall.
ML
So we have done one other episode on fusion related stuff, which was with a material scientist called Aneeqa Khan from Manchester. And I think she was talking about some of those because I was under the impression that you were still using porous capillary, so having a sponge of tungsten or a sponge of steel and putting the lithium in that. But that's no longer what the cool kids are doing.
SC
Oh gosh, again, more very technical stuff. We're talking about coating the inside walls near the reacting substance with lithium, and the reason for that is that it protects the walls, the bombardment
ML
How far does the neutron go into the wall? What distance are we talking about?
SC
Many centimeters.
ML
Many centimeters.
SC
The width of the blanket is probably a meter to a meter and a half.
ML
OK, so you've got to do that. And then, of course, if you manage that, then because you can keep the wall cooled and then you take the heat off and you can generate your electricity. I wanted to go through it because I want to give people a little bit of an idea of what the actual machine might look like. It's very easy to focus on a little tiny frozen micro pellet of deuterium tritium, but it's got all this other engineering around it. There are a bunch of startups that are working on laser containment, so on exactly this, I've got to find a way of asking you, there's a bunch of them that have raised around the sort of well into the hundreds of millions. One most recently, actually in Germany, that wants to use the old Biblis nuclear power station location. That one's called Focused Energy, there's Inertia Enterprises, First Light Fusion, Marvel Fusion, Xcimer Energy. So there's a bunch of them. How closely do you follow their progress and with what amount of hope?
SC
That's such a carefully worded question, Michael. Focused Energy has some very smart people in it, they're doing some very interesting things. This is early stage deep tech. There's a lot of risk being taken, but there's a lot of reward at the end of the line. It is quite possible that at some point, fusion will be 30% of the energy market, that's a heck of a lot of money. And so a certain amount of risk taken to accelerate development of that. I think all the fusion companies are doing that. So I pay attention to pretty much all the fusion companies, there are a few that are just plain bonkers, and that's a technical term, there's bound to be in this space. But there's been $15 billion in investment in fusion companies in the last six or so years. That's a heck of a lot of investment compared to the public sector program. We've got to get something out of this process to accelerate fusion. So I'm paying a lot of attention.
ML
I could also rephrase my carefully phrased question, which of course was trying to trap you into saying things you might later regret. But if I rephrase it and say, is laser containment a worthy competitor to magnetic containment? Is it obvious that one's ahead, one's behind, that one's challenges are, we're going to come on to talk about magnetic, but do you think they have a chance of beating the magnetic containment route, which is the one that mainly you and your colleagues work on?
SC
I have preferences, but we're at a very early stage of the technology and things overtake. And the NIF results certainly said that, yes, inertial fusion is coming along very fast. But most of the laser fusion companies are aiming to do it differently than NIF. And in particular, most of them do what's called direct drive. So indirect drive is what they do at NIF. You have this little gold container, the hohlraum, and the laser shine in end holes in the hohlraum and they make x-rays from the gold and those x-rays then collapse the peppercorn sized target in the middle. That's easier because the x-rays make it more symmetrical and the push is more symmetrical, which you've got to have, right? But it's terribly inefficient because you convert laser energy into x-ray energy and then the x-ray energy does the implosion, etc. And that gold hohlraum, if you have to make targets with gold hohlraum, they have to cost roughly 20 cents each. At the moment, they cost $10 million each. So we're a little ways off, right? So it'd be much easier if you just made the peppercorn thing in the middle, right? And so people at the University of Rochester and I think Focus Fusion, they're aiming to shine the lasers directly on the pellet. Now, we've never made that work, but in principle it does work.
ML
OK, so thank you. I'm trying to get the right level between kind of, because it is the physics that tells you, it's the x-rays and the physics that tell you whether the thing is to be gold or whether you can get rid of the gold. So we are in the right zone, I think. I hope people are able to follow this, because I think the short version is that, yes, it is a worthy pathway and big challenges, but a worthy pathway and some very good science being done. Some of it is funded by private companies and good luck to them, but it sounds like there's a lot of science still needing to be done.
Cleaning Up is proud to be supported by its Leadership Circle. The members are Actis, Alcazar Energy, Arup, Climate Imperative Foundation, Copenhagen Infrastructure Partners, Cygnum Capital, Davidson Kempner, EcoPragma Capital, EDP, Eurelectric, KKR, Mitsubishi Heavy Industries, National Grid, Octopus Energy, Quadrature Climate Foundation, Schneider Electric, SDCL and Wärtsilä. For more information on the Leadership Circle, please visit cleaningup.live. To keep up with all that's going on in the Cleaning Up universe, make sure you subscribe to our newsletter. Written and edited by my longtime New Energy Finance and BloombergNEF colleague, Angus McCrone, it comes out every second Monday. Angus provides the latest on the episodes we're recording, the events we're hosting, stories we're watching and what Bryony Worthington and I are up to. To sign up for the Cleaning Up newsletter, visit cleaningup.live.
Let's move on to magnetic containment. Because then we get to talk about plasma, your favourite subject. I'm assuming, given that you run the plasma physics lab, that it's a favourite subject. These are the big circular structures, the tokamaks, all these things called stellarators and a stellarator means a star maker. But broadly, it's stuff that's going round and round and it's contained by big magnets. What are you trying to do with that, and why is it hard? Let's start with the real basics.
SC
You know, fusion doesn't happen until you get above about 100 million°C. I mean, it happens, but it's so low level that you wouldn't worry. But you've got to get to 100, 200 million°C. And then you've got to hold your fusion, your deuterium and your tritium, which are your fuels, while they come together, they bombard each other. Most of the time, about 999 times out of 1000, when deuterium hits tritium, it just bounces off, right? That one part in 1000, when it fuses, you've got to get it hot enough that it can get close enough that it can fuse, right? So you've got to hold something at 200 million°C. And the magnetic fusion is to hold it steadily in a cage of magnetic field at 200 million°C and give it time to fuse and do it continuously. Not an explosion, but just continuous output of fusion power from that.
And that, as I say, on JET, we actually did that, right? We held it there five seconds. It doesn't sound like a long time, but it's well past any of the other times that you worry about. We held it steady, it produced 12 megawatts of power for five seconds. Excellent, right? So the cage kind of works. But at this point, the problem is you also want the fusion to heat the fuel because you want to put more fuel on and you want it to heat it, like a burning fire. If you have a burning fire and you add another log, you don't want it to go out, you want to ignite the log and then put another log on and then put another log on. So we want it to heat itself as it's doing fusion, and it will do that. But to do that, you've got to get it self-heating to beat the leakage of heat. Otherwise, it will just get colder and colder and colder and it'll stop doing fusion, right? And to do that, we have to make very large magnetic bottles because in the middle, it's bubbling away. This plasma is actually turbulent and it's bubbling away and the heat is leaking out of your device. Now, there are two ways to make the leakage, actually, there's really three ways to make the leakage less.
ML
Let me just come in there because there was something very critical that I'd like to just highlight, which is that you use this word hold it there and 5 seconds. And of course, if it's 5 seconds, then you say, well, why can't it do 10, why can't it do 20, why can't it do 500 seconds, why can't it do continuous? Because what you're working on is continuous production of power. It's not lots of explosions. It's not an internal combustion engine with fusion. It's almost like it's a jet engine with fusion. It's just continuous. If you can make that stable, then that feels like a very useful thing and possibly avoid some of those engineering problems of hand grenades blowing up and lenses going opaque and all sorts of things.
SC
Yeah, and it becomes self-sustaining in which point you're not putting energy in to get energy out.
ML
But to be self-sustaining, Q of 1, just getting a bit more energy out than goes in, Q of 1 won't do it, correct?
SC
No, you have to get to the point that you're putting no energy in. And that's possible in magnetic fusion, it's not possible in inertial fusion.
ML
Okay, and to get there, I'm going to see have I studied well, Professor? Because 80% of the energy goes out as the neutron, which embeds itself in the same sort of blanket that we talked about with the lithium that's produced, because you still need the tritium, and sticking with the deuterium tritium fuels for the moment, you still have the same sort of blanket, it produces tritium. That's 80% of the energy that's kind of lost to the plasma. It's going off to the blanket and off to make steam and power all those data centres and all that good stuff. But 20% is available for the plasma.
SC
Self-heating.
ML
Self-heating, okay.
SC
In magnetic fusion, we've done fusion, but we've not done a self-heated fusion plasma. And scientifically, I would say that's the next big step by far. Right? We would like to get to a point where you're not putting in any energy at all and the fire is burning.
ML
So at the moment, we're putting in kind of 1 unit and getting 0.3 out.
SC
Yes.
ML
But you need to be putting in, from speeches that you've given, you're kind of looking for like 10, a factor of 10, because then 80% can go away, 20% stays, a bit of it gets lost. And there's enough though to heat, to maintain the true ignition. Self-heating ignition would require a Q of 10.
SC
Right. Well, Q of 10 is sort of a threshold for that. I would actually say commercial reactor probably wants to work at Q equals 30.
ML
Okay, but these are only one, one and a half orders of magnitude beyond where we are. It's not like you're billions of, you're not nine orders of magnitude. I mean, if you're doing one laser shot every two months and you need to get to 600 a second, I calculated that you need a 4 billion times improvement, but you're much closer, you're much closer.
SC
Thank you, yeah, we are very close. But to do this, we have to make very large devices, right? Because to minimise the heat that's lost, we make it very big. It takes a long time for the heat to get from the middle to the outside. Or, and there's a very interesting company that's doing some very interesting and good work called Commonwealth Fusion, it's probably the most funded fusion company. It's got $4 billion in investment, Breakthrough Energy invested in them, a bunch of investors. Commonwealth Fusion is trying to develop stronger magnetic fields because the cage is better if the field is stronger. And so they're using high temperature superconducting magnets to try and go to more than double the field that we had in JET, but close to three times the field.
ML
And you've got some handy equations which you use to communicate with folks like us, what might drive the costs and a better magnetic field, a bigger device, which you mentioned that we do big devices, we have better magnetic fields. And the other one that you highlight in your equation is the quality, the lack of turbulence or the ability to manage turbulence. But which of those, you've got different players that are betting on different versions of those three, size, magnetic field strength and turbulence.
SC
So I think reducing the turbulence, I've always said that, you know, you can make a brick fly if you put a great big jet engine on the back of it, right? Doesn't fly very well. But until you've got an aerofoil, you haven't got the right shape. And reducing the turbulence is about finding the right configuration of the magnetic bottle that reduces the turbulence. And that's the smart way to solve this. And there was something that happened about 15 years ago. We started to be able to calculate the turbulence inside this. It's a horrendously large calculation and we're using a lot of very large computers, but we're now able to, in the computer, start to look at optimum solutions. And part of the fusion commercial market, the private sector fusion companies, is fascinated by stellarators, right? That stellarator, unlike the tokamak which JET was, is a device that doesn't have any electrical current in it or electrical currents outside it. It's an invention that the founder of the Princeton Lab, Lyman Spitzer, invented in 1951. But in 1951, we couldn't calculate it very well and we certainly couldn't build it because it requires precision engineering, right? But stellarator in Germany has been getting spectacular results. And so there's a company in Germany, there's a company in New Jersey called Thea Energy that spun out of our lab, there's a company called Type One in Tennessee. These are companies that have been using these massive simulations to find optimum solutions.
ML
Let's take a step back because what we can do for those who will be watching this on video is we'll put in a picture or a few pictures of stellarators because these things look like the Death Star. They're this sort of twisted nuclear spaghetti, I can't describe them, we'll put an image. Anybody who's listening on podcast, you're allowed to put this on hold, put it on pause and go and have a look at a stellarator because the geometry is incredibly complex. So Lyman Spitzer worked out that if you just have a straight doughnut, then you've got some of the plasma taking the short route and some taking the long route and it won't work. They'll all drift all over the place and it won't work. But if you twist it, then you can get it all to link up and then nobody could make one, I think there were many efforts to try in the lab's history. And then was it 1968,we discovered that the Russians had got a tokamak apparently working. Is that correct?
SC
You know, some technologies do this. I mean, we were flying hot air balloons for more than 100 years before their airplane. And I think that the promise of the stellarator is probably the biggest promise in fusion. Spitzer was right at the beginning. You want a steady state machine that just does fusion and keeps burning away and the stellarator gives you that. And you want something in which there is no current flowing in your plasma because that current can be unstable. And so there are a number of physics reasons why his idea was great, but in 1951 and in the 1950s, you just couldn't build this thing. And in the 1990s, the Germans got the bug and they built a machine called Wendelstein in Munich. And then they've now built an improved version in Greifswald on the Baltic coast. And it's getting great results, and we think we have better designs than that. And at least in the computer, it performs better than anything else.
ML
And in the computer, that's now been turbocharged because we now got AI. So the issue is no computer can solve all the equations from quantum scale phenomena up to the five seconds or the five months or whatever you want to run these things. So you're having to break the problem down into chunks and we're grinding away at that. And then suddenly, or over a period of time, AI comes along. What is that doing to our ability to control, to eliminate or manage turbulence?
SC
We're already using AI algorithms to control instabilities. And that's proving very, very, because it's so rapid. You can be on the timescale of what's happening inside the plasma and that's very good. But the big payoff, and this is going to be part of the Department of Energy's Genesis Mission, the big initiative in the Department of Energy is the Genesis Mission, which is AI for science, and one of the top problems among AI for science is fusion. And the idea here is to take those projections from high performance computing and let AI learn from them. And that will make it easier for us to iterate faster and faster and faster and look for optimum solutions.
ML
Optimum solutions, it feels to me like there's two things those might look like. One is learning more and designing or doing more of the design work on a computer rather than a billion dollar science project. So one way is design, the other would be real time control. So changing the magnetic fields so that when turbulence starts to happen, you can actually intervene and do something about it. Which one are you on? Which one does Genesis target? Or is the second one impossible?
SC
Both really. One of the very interesting things happening actually in the accelerator company, Thea, which is in northern New Jersey, spun out from our lab, is they're making the magnetic field out of many, many small coils. This is a way of making the magnetic field that allows you to make subtle changes in order to get the best possible turbulence suppression. I think that's a very interesting idea because we've never had the kind of control of the plasma. We make plasmas and they do what they want. We want to be able to shape them and control them really in intimate detail and AI will allow that.
ML
So if you think big is going to win, you bet on ITER. If you think magnets are going to win, you bet on Commonwealth Fusion. And if you think that AI and smarts are going to win, you bet on Princeton or Thea, or who would you bet on?
SC
Well, I think the whole community is rallying around AI. I would say Princeton, of course, right? I mean, absolutely. If you're talking smart Princeton, in same breath, I'm your man. But no, I think that everybody in fusion realises that this is going to be a game changing technology for us. We always say, you know, AI for fusion and maybe fusion for AI eventually, because hopefully we'll power data centers in the future, right? But it's interesting. But there's one problem in science that AI solved and that's the protein folding problem. Alpha fold.
ML
Two now, Navier-Stokes.
SC
Oh, yeah, that's a math problem, right? And I agree.
ML
That's going back to your father's favorite subjects, right?
SC
You got me there. Yes, but the alpha fold problem, the problem is not that we couldn't calculate the folding of proteins, it just took too long, right? And by having AI recognise, “I've seen that fold before, it folds like this”, and then it goes, “you move along and then it says it folds like this”. And you can calculate the folding of an enormous protein in seconds now, right? It's fantastic. Same thing is needed for fusion, right? We need to be able to calculate the future and not just keep guessing. Thomas Edison had this famous way of inventing, you make 50 of them, right? Each one is a bit different and you find out which one works the best. And then you take that one and you do a sort of genetic algorithm, right? You breed from it and you make 50 more that have those advantages and then you keep doing this. If we do this in fusion and it costs a billion dollars a time, you're soon talking real money.
ML
The one architecture that we've not talked about or there's a few others I want to touch on, spherical tokamaks because the UK has stepped out of ITER. ITER was initially going to in 2006, it was supposed to finish construction in 2016, first experiments, 2020. We're now talking about full operation with deuterium tritium in 2039 and it's cost 25 billion euros, whatever. It's enormously late and over budget. The UK stepped out, but they're putting all their, you know, the UK's government program is all about spherical tokamak. It's the STEP program, Spherical Tokamak Energy Production, is that a useful contender in the race?
SC
Absolutely. Yeah, so we're just about, in November we'll finish construction of the world's most powerful spherical tokamak at Princeton. The spherical tokamak is a way to bring down the size and the cost of the tokamak because I'm a big supporter of ITER, but I don't think that that architecture will make commercial fusion. I think you've got to shrink it, you've got to make it cheaper. And so the spherical tokamak may work to do that. So we are building a near fusion scale spherical tokamak at Princeton and the UK will come over and do experiments at Princeton on that device to see how it will scale to this reactor scale.
ML
So it's another good horse in the race.
SC
It's a good horse.
ML
And then while we're on different architectures, there's the reverse field, a completely different approach, reversed field configurations, which is creating some plasma and then blowing it together or doing something similar and there's a number of companies doing that. Helion, Tri Alpha, I'm not sure how many others are trying to do that, is that another good horse in the race?
SC
It has a big advantage in that it's small and it's not a torus. The problem from an engineering point of view of a torus, donut shape, is that everything has to link the torus. It's hard to disassemble if you want to do maintenance, right? So the field reverse configurations have a wonderful thing that they are just a ball of plasma, but it's a current ring inside the plasma. And the problem is that they're always unstable to tilting and you have to stabilise that tilt. And it's still a struggle for those companies, but they've done some really interesting stuff.
ML
They have to not only stabilise it, but they have to then ram it together. And there's a lot of different ways of doing that. I have to find a good way to ask this, are any of those ways of stabilising the plasma and then shoving it together so that you get that ignition or that you get the fusion reaction starting, are any of them close? And the background here is that one of them, Helion, has got a deal with Microsoft to sell electricity in 2028, that's just two years away. Are they going to achieve that?
SC
I don't know. They don't publish and so it's very difficult to test exactly what their plasma parameters are. I'd be surprised, but if I don't know, it would be wrong of me to comment on my worries.
ML
But I mean, it is plasma so you could be helpful, you could run a program. Are you looking into that architecture as well?
SC
I tell you one of the things that does frustrate me with the private fusion industry. We have developed computer models of plasmas in the last 15 years that are very close to what we measure. We're often measuring exactly what we calculated in the computer. And so guessing what's going to happen as you get towards fusion is not the right strategy, you need to calculate. And we can help a lot of the companies, we're helping 15 fusion companies by doing simulations, basically, of their things. We're not doing full simulations, although we're doing some work for those field reverse configurations.
ML
And does the moment we're in with private money coming in, does it concern you? Because there's nothing I've heard that's convinced me, you started by saying that the US government's target is 2035 to 2040, and I tried to push you to say whether you thought 2028 was realistic, and I'm going to read between the lines and say that you don't seem overly convinced. Are you worried that this sort of moment of excitement just ends up dissipating? You couldn't really call it a bubble and a bust because it hasn't involved the stock market, but of course, you've got Tri Alpha, TAE, which has now become part of a quoted entity, as has I think, General Fusion, which is a Canadian-based, who were going to make a ball and then collapse a ball using pistons and explosives and all sorts of things, also quoted now. Are you worried that there'll be a pushback, that people will end up in a few years saying, ah, that didn't work and it was all nonsense and that that would affect the programs that you work on?
SC
I very much worry. This is not my area of expertise, finance, but what I worry about is we'll have sort of Fusion's Theranos moment, because there are some really good companies.
ML
So Fusion's Theranos moment.
SC
Theranos moment right, where investors say you didn't exaggerate, you actually duped us, right? And if that happens, it'll cool the whole market. And there are a lot of really good companies. But there are 60 Fusion companies, all with different ideas, right? They can't be 60 ideas that are all, you know, and certainly some of them strike me as, I don't know anybody who's fraudulent, right? But I do know people who I think, you know, are over egging what they have.
ML
I want to finish by talking about the costs and the economics of this, because doing it is one thing, but doing it at a cost point where it can compete into the energy system, and there's this marvellous phrase which I'm afraid you've used a few times in other conversations, baseload. And baseload is regarded by many as the sort of the holy grail, you have baseload. But baseload, if you have an electricity source that produces 24/7 365 power, in a world which is dominated by intermittent power, it's actually not fulfilling a very useful function in the energy system. Because every time it's windy or sunny, or you've got very cheap batteries as well, every time there's electricity available at a very, very low marginal cost, you're much more expensive to run equipment is going to be shut down.
SC
Well, you're making the assumption it's much more expensive, but it probably at least initially will be. Fusion can be switched on and switched off, there's no question that you could do that. But there's a cycle. This is true also, one can make nuclear power stations that you switch on and switch off, you just don't want to do it because the marginal cost of keeping running is not very much. Whereas because your fuel costs are so low, and your running costs are so low, you've got to keep the reactor there. So it is true that when you have a technology that's dominated by its investment cost and not its operational costs, you want to keep it running all the time.
ML
Correct. But the problem here is that the running cost of a solar farm is almost exactly zero. I mean, in wind, you've got some maintenance. So what we've got, even in nuclear now, just running and maintaining an efficient power station is $30, $40, $50 per megawatt hour. And so they're being competed out by wind and solar and batteries, where the marginal cost is almost exactly zero. So it's a problem. And the reason you don't want to cycle, apart from the fact that you get temperature cycling and cracks and all sorts of bad things, the moment you power it down, it's not earning any money. And so the cost of the rest of the power goes up. So I guess I worry that you could do all of this and in the end, other than the first few that are heavily subsidised, that it may never find its role in life in the energy system.
SC
Well, if we can get the cost down and it can become a firm source as opposed to a baseload source. A firm source, I think the definition is that you can switch it on when you need it.
ML
Dispatchable, I would call it dispatchable for the avoidance of doubt, because firm sounds like it's going all the time. Dispatchable is what you need, that's very valuable, no question.
SC
Right. And I think, you know, there's a hope that we get fusion into that role because it's going to be very difficult because the cost of solar and wind is not going to go up. It's probably going to go down even further than it is now.
ML
Batteries, most certainly.
SC
Yeah. But I think fusion has to get into that market and we don't know what the cost of fusion will be.
ML
So there are estimates and I've watched you talking about them on other things and other programs on YouTube. And you always sort of, a bit like you're doing now, you always sort of smile and say, well, you know, oh, it could become very cheap.
SC
A supercilious smile.
ML
No, not supercilious, it's almost sort of naughty and mischievous because your role is to raise money for fusion for the plasma lab, for the Princeton plasma lab. So I don't know but you're confident that it can become cheap enough to play a role, clearly.
SC
I am, yes. Estimates of what it's going to cost, you know, if you estimated the cost of a fission power station and you didn't know anything about the regulatory environment or that these kind of things or the restrictions on the way you build and that kind of thing, you would say it's a very competitive technology. So the cost of a future technology is very, very hard to estimate. We have to get closer before we know.
ML
By the way, just for the avoidance of doubt, because some people listening or watching might think, oh Michael's trying to destroy the business case. I'm not, I'm challenging, but I'm enormously supportive because in the grand scheme of things, the amounts of money are very small. You know, when the US spends, I don't know, $45 billion every year on vitamin supplements and only a couple of billion on plasma and potentially on fusion. I'm absolutely supportive of spend the money and let's figure it out and if it takes to 2040 or 2050, we still want to do it. I hope, though, for your sake, that it doesn't take quite that long because you've said you'd like to see it during your career.
SC
Absolutely. I would like to see a self-sustained fusion burn and I think we're close, right? That's maybe not first electricity, it's before first electricity, but I'd like to see that and I'd like to see some electricity.
ML
So I would like to see self-sustained fusion burn, I'd like to see electricity and then I'd like to see people buying the things, and I'd like to see some of the project financiers that I know getting able to come in behind and for the technology to be bankable. And I intend to have a very, very long life.
SC
OK, I'm with you.
ML
Very good. Steven, thank you so much for spending time with us here today.
SC
You're very welcome, thank you for having me.
ML
So that was Professor Sir Steven Cowley, Director of the Princeton Plasma Physics Lab. As always, we'll put links in the show notes to resources we mentioned during our conversation or which might be useful for further research. That includes a link to Steven's TED Talk from 2009, his talk at the Institute for Pure and Applied Mathematics, IPAM, from April this year, his talk at SOSV Deep Tech Live, also from April, a link to the Princeton Plasma Physics Lab itself and to the Plasma Forge Incubator, which the lab has set up with venture capital provider SOSV.
With that, I'd like to thank our live audience for joining us today, Oscar Boyd, our producer, Jamie Oliver, our video editor, Kendall Smith and Jo Jagger for their sterling work organising the event at which this was filmed, our Leadership Circle, without whom none of this would be possible, and you, the audience out in interweb land for taking the time to listen or watch. Please make sure you sign up to our newsletter at cleaningup.live and join us at this time next week for another episode of Cleaning Up.
Cleaning Up is proud to be supported by its Leadership Circle. The members are Actis, Alcazar Energy, Arup, Climate Imperative Foundation, Copenhagen Infrastructure Partners, Cygnum Capital, Davidson Kempner, EcoPragma Capital, EDP, Eurelectric, KKR, Mitsubishi Heavy Industries, National Grid, Octopus Energy, Quadrature Climate Foundation, Schneider Electric, SDCL and Wärtsilä. For more information on the Leadership Circle, please visit cleaningup.live. If you're enjoying this episode, please hit like, leave a comment, and also recommend it to friends, family, colleagues, and absolutely everyone. To browse our archive of around 250 past episodes and to subscribe to our free newsletter, visit cleaningup.live.