The Race to Nuclear Fusion (Patreon)

[This is a transcript with references.]

Nuclear fusion is hot, both literally and financially. Nuclear fusion startups have attracted billions of dollars in the past years. And each of them thinks they’ll be the first to make it work.

The promise is no doubt huge and has attracted supporters such as Bill Gates, Jeff Bezos, and George Soros. But there are so many of those start-ups now, it’s become really confusing. So, we made this video to give you an overview on nuclear fusion startups: Who is doing what and how far along are they? That’s what we’ll talk about today.

The nuclear fusion business is booming. 4 point 8 billion US dollars have been invested into nuclear fusion start-ups so far, according to a 2022 report by the Fusion Industry Association. 2 point 8 of that money arrived in the past year alone. Almost all of it is private funding.

These numbers pale in comparison to dozens of billions that have gone into huge governmentally funded projects, such as ITER or the NIF. But those fulfil a very different purpose. While the big projects lay the scientific groundwork, the startups now want to get the power into the grid.

A majority of those in the business think fusion is close to being commercially relevant. In the FIA report, more than three quarters of the respondents were confident that fusion power would feed an electric grid in the 2030s; two thirds of them said it’ll be the first half of the decade. And most of the rest said it’ll be even earlier.

So far, none of those start-ups have actually produced net power. But the promise is huge because we know it works, in principle. It’s “just” the engineering part that isn’t quite working.

In principle it works like this. If you push two small atomic nuclei into each other, you create a heavier nucleus with slightly less mass. The mass that’s lost is released as energy. It’s a staggeringly huge amount per fusion reaction. This is why nuclear fusion, like fission, produces a lot of energy from very little fuel. The amount of energy you get out of a certain volume of fuel is more than a million times higher for nuclear fuel than for fossil fuel.

What makes nuclear fusion so difficult is that atomic nuclei are all positively charged and therefore repel each other. To get them to merge, you must overcome this electric repulsion in some way. And then you must do sufficiently many of those reactions in a short amount of time.

To quantify how well this works, scientists use the “gain” normally denoted with the letter “Q”. It’s the ratio of energy out over energy in. So, a gain larger than one means more out than in, and that’s what we want to reach.

However, as I explained in an earlier video, the scientists and engineers who work on nuclear fusion are often interested in what goes on in the reaction, so they will instead just quote the energy that goes into and comes out of the reaction. But the gain factor from the reaction alone does not take into account the energy that is required for powering the entire equipment. The total gain is often considerably smaller than the gain from the nuclear reaction alone. So you have to be careful with interpreting the numbers that they quote.

Nuclear fusion isn’t all sunny, it does have some downsides. For example, it creates radioactivity. Mostly that’s the material in the vicinity of the reaction which gets irradiated with neutrons. This is why nuclear fusion reactors must be heavily shielded. But this radioactivity is short-lived and will decay to background level within a few decades. So nuclear fusion *does create some radioactive waste, but it isn’t as difficult to handle as that created by *splitting atomic nuclei.

The biggest downside of nuclear fusion is that the most-studied and easiest fuel is a mixture of deuterium and tritium. Deuterium and tritium are both isotopes of hydrogen which is the smallest atom. This means this combination has the smallest possible electric repulsion and it’s the easiest case to make nuclear fusion work.

The vast majority of fusion startups use this combination as fuel. But while deuterium is abundant on earth, tritium isn’t. This means one either has to use a different fuel, or one has to find a way for the fusion reactor to also produce the tritium. 

Another issue with fusion reactors is that while we might get them to work eventually, they might turn out to be prohibitively expensive, but I guess we’ll see about that. The currently biggest problem is that a total gain of larger than one has so-far not been achieved. Reaching it is the major challenge that all nuclear fusion startups must solve. They’ve come up with three different methods for that.

First, there’s field confinement, that works by stripping the electrons off the atomic nuclei to create a plasma. Then one uses magnetic and electric fields to hold and heat the plasma. If all goes well, this ignites a fusion reaction in the plasma which creates a lot of heat from which energy can be extracted.

This is the method that has been used in the most-explored governmental projects, such as the Joint European Torus and that will also be used by the mega project ITER. The big benefit of field confinement is that it’s been well-studied and is well-understood, and if you can get it to work you have a lot of plasma burning at once. The big downside is that the plasma is very difficult to keep in place.

The second method is inertial confinement. Inertia is the reason why we “hang up” phones that haven’t had anything to hang up for decades, and it’s the reason why, if you shoot at an object, it takes a while for it to start moving. This means if you just shoot nuclei at each other violently enough, they will merge, at least sometimes.

Particle colliders, such as the LHC, do this too, but powering a 27 kilometer ring of superconducting magnets to fuse a few protons isn’t exactly energy efficient. For inertially confined nuclear fusion, one therefore instead shoots at a fuel pellet, hopes that it blows up, and makes energy from the heat created in that process.

The big benefit of inertial confinement is that it requires about a thousand times less tritium than field confinement. This is because with field confinement a large amount of the fuel goes unused, whereas inertial confinement, if it’d work as desired, uses most of the fuel. The big downside of inertial fusion is that you have to reliably and very exactly reproduce those shots, and the science of that isn’t as well understood as for field confinement.

And the third method is to use a hybrid approach of both field and inertial confinement.

In this video we’ll not go through all startups because you’d get bored and I’d cry over my YouTube stats, and in any case most of them use quite similar methods. But we have picked a super non-objective sample of those who we found the most interesting. We’ll start with those using field confinement.

If you want to get hydrogen nuclei to fuse, you need to heat the plasma to about one hundred million Kelvin. With other types of fuel, the temperature might be even higher. There’s no material that could hold a plasma that’s so hot. So, you have to try and levitate it in a vacuum chamber by using magnetic fields, and you have to prevent the plasma from touching the containment vessel. Because, if it touches the vessel, that’ll damage it, and that isn’t good neither for efficiency nor for your budget.

Trouble is, the motion of this insanely hot plasma is partly chaotic which makes it difficult to control. The record for the duration of a controlled fusion reaction is currently held by the Joint European Torus. Guess what the record is? It’s about 5 seconds. But artificial intelligence has a lot to contribute to this area, I talked about this in my previous video on chaos control.

The most common devices that use strong magnetic fields to confine the plasma are called tokamaks, which have the shape of a doughnut, but if you want to sound educated you call it a torus.  

These devices have so-far relied on superconducting magnetic coils cooled by liquid helium because that’s the only way they can reach sufficiently strong magnetic fields. The existing tokamaks are huge, for example, the one at ITER will be contained in a 70 meters high building, that’s about 20 floors. There are several startups which want to improve on this with “mini-tokamaks” that will use stronger magnetic fields and require less cooling.

One company which works on a small tokamak is Tokamak Energy. Like the Joint European Torus, they are based near Oxford, in the UK. They have attracted about 250 million US dollars in funding and have about 190 employees.

The company was founded in 2019 by a team of scientists including Alan Sykes, who in the 1980’s did a calculation which revealed that one can improve a tokamak by using a torus whose entire cross-section approaches that of a sphere. In such a device, the plasma is confined closer to the center of the ring, where the magnetic field is higher. This increases both the efficiency of the reaction and the stability of the plasma.

It’s called a spherical tokamak, but the word is somewhat confusing because the thing isn’t a sphere, it’s still a torus, it just doesn’t look very doughnut-like any more, it’s more like one of those inflatable bumper balls that people play soccer in.

Tokamak Energy has called their machine “Peter”. Nah, just kidding, they’ve called it after one of Elon Musk’s kids: ST80-HTS.

They are using the common fuel of deuterium and tritium, and high temperature superconducting magnets to keep it in place. Those magnets can be cooled with liquid nitrogen rather than liquid helium. This has the advantage that liquid nitrogen is both considerably cheaper and easier to handle.

ST80-HTS will be about 3 point 5 meters in diameter, so that’s pretty compact, and I really think they should have called it Peter. They hope to complete it by 2026. Their goal is to construct the world’s first commercial fusion power plant by the early or mid 2030s.

Another startup which is working on a small tokamak is Commonwealth Fusion Systems based in Cambridge, Massachusetts. The company emerged in 2018 as a spin-off from the Plasma Science and Fusion Center at MIT. At the moment it’s the best funded of fusion startups, has attracted more than 2 billion US dollars, and has more than 300 employees.

They are currently building the SPARC tokamak. SPARC is presumably an acronym but try as I might, I haven’t been able to figure out what it stands for. Depending on whom you trust it’s either the soonest possible or smallest privately funded affordable robust compact, and is lacking a noun.  I believe its real name is Sparky McSparcface.

Whatever its name means, the device also uses deuterium and tritium as fuel. It isn’t expected to produce net power, it’s just a demonstration machine. Saying that it’s small is maybe somewhat of an overstatement. While it’s considerably smaller than ITER, it’s still about 5 meters tall and 15 meters wide.

According to a 2020 paper the device will use magnetic fields of over 12 Tesla to heat the plasma to about 80 million Kelvin and they, too, use high-temperature superconductors for the magnets. SPARC is supposed to begin operation in 2025. The company says their goal with this machine is to get a reaction gain larger than two.

After SPARC, they plan to build their next machine which will lose two letters and just be called ARC. It’s supposed to start running in the early 2030s and generate around 250 MegaWatts of net power. The company hopes it’ll be the world’s first fusion power plant to put power into the grid.

The Chinese startup ENN Fusion Technology is also working on a small tokamak, and a whole bunch of governmental tokamak projects are under construction, too.

Then there are the stellarators. The idea dates back to the 1950s, when American astrophysicist Lyman Spitzer showed that magnetic fields could be configured into a twisted loop to make a “magnetic bottle” that could be filled with plasma. This configuration makes it harder for the particles to escape. At least theoretically, Spitzer claimed, this would make the plasma more stable and the device more compact.

But in the 1950s, they couldn’t solve the equations, so the idea wasn’t pursued until the 1980s, when supercomputers became powerful enough. Still today it’s difficult to do computer simulations for stellarators.

The world’s largest existing stellarator is currently Wendelstein 7-X, at the Max Planck Institute for Plasma Physics in Greifswald, Germany. It’s a test reactor that has been in operation since 2014. It cost more than 1 billion euros, but some private startups hope to get it done more cost-efficiently.

One of the stellarator startups is Type One Energy, based in Madison, Wisconsin. It was founded in 2019 by scientists who had already built a smaller stellarators for demonstration purposes, so they bring the expertise. The company has attracted almost 52 million US dollars in funding, but is fairly small with currently about 10 employees.

Their machine is called STARBLAZER, which does not seem to be an acronym but it nevertheless must be capitalized. It too will use deuterium and tritium as fuel. They claim that STARBLAZER I will get energy out of the reaction for several hours by 2030, but not output net power.

It’ll be followed by STARBLAZER II. According to their website the second machine is supposed to run continuously and reach Q equals infinity by 2034. Technically, this means they either produce an infinite amount of power, or produce power without any input whatsoever, or more likely, someone’s tried to divide infinity by infinity.

Another startup banking on stellarators is Renaissance Fusion based in Fontaine, France. It was founded in 2020, has attracted more than 16 million dollars, and has 14 employees.

They also plan to use deuterium and tritium in a stellarator with high temperature superconducting magnets. Their special idea is that they claim to have found a way to simplify the complicated stellarator design. Rather than using difficult shapes for the magnets, they use simple shapes, but then they use laser engravings on the outside of the vessel that direct the currents and create the required magnetic fields.

They say that this is cheaper and faster but has the same result. They also want to surround the vessel with a circulating fluid that contains lithium. This will serve both to transport off the heat and better shield the radioactivity that’s created in the reaction.

They plan to have a small-scale demonstrator that should produce energy gain in the reaction by 2027. By 2032, they want to have a full-size net-electricity reactor connected to the grid.

Other companies that work on stellarators are Helical Fusion in Japan, N.T. TAO in Israel and Princeton Stellarators in the United States.

A very different approach to fusion is to create beams of the fusion plasma and use magnetic fields to collide them. The company TAE Technologies, which is based in California, uses this approach. It was founded in 1998 and originally called Tri Alpha Energy, which is where the TAE comes from, not that this explains anything. They have over 400 employees and have raised about 1 point 2 billion US dollars. Funders include Google and Goldman Sachs.

Their approach is very different from the ones we have previously discussed, not just because they collide plasma beams, but also because they use a different fuel. They want to fuse normal hydrogen with boron nuclei. The advantage is that these are both abundantly available fuels, and also, the reaction generates no neutrons that could damage the reactor.

But of course, this has a downside, too. Since boron is a bigger nucleus than hydrogen, the electric repulsion is stronger. To get fusion to work, one has to reach temperatures ten times higher than that required for the more commonly used deuterium-tritium fusion. This means they need to reach about one billion Kelvin. And then the energy they get per reaction is lower than what they’d get from deuterium-tritium. However, their device is designed so that it can also run on several other fusion reactions which include the conventional one using deuterium and tritium.

Their current machine is called Norman, which is not a joke, but a name chosen in honor of Norman Rostoker, one of the company’s founders who passed away in 2014. Norman is 7 meters tall and 24 meters long. It consumes up to 750 MegaWatts, which is enough to power a small city. Luckily, Norman doesn’t run continuously, but it fires shots, so they only need that power for half a second at a time. To get that done, they’ve built equipment to store and suddenly release the required amount of energy.

The difficulty with this approach is to get a significant part of the plasma to collide at high enough energy. They inject blobs of plasma on both ends of the machine, and then use magnetic fields to shape it into two rings, similar to smoke rings. and then push them into each other. They need billions of Kelvin but so far they haven’t made it past 70 million.

So there’s some way to go, though this is getting close to the temperature needed for the deuterium-tritium reaction. They expect to get deuterium tritium to work in the next couple of years. By the end of the decade, they hope to get to a billion-Kelvin that’s necessary for the boron reaction.

A completely different approach is pursued by the company ZAP Energy, based in Seattle. It was founded in 2017, has about 200 million US dollars in funding, and 60 employees. They use what’s called the Z-pinch.

For this, one charges huge capacitors and then suddenly releases the energy into a vacuum chamber which contains a gas of the fuel. The electricity turns the gas into a plasma and creates something like lightning bolts. The columns conduct electricity which creates electric fields. The electric fields in return create magnetic fields that pull the lightning bolts towards each other, until the density gets so high that fusion takes place. The direction of the lightning bolts is called the z-direction, which is why the thing is called the z-pinch.

The Z-pinch system doesn’t work continuously; it fires shots. The company says that the main advantage of Z-pinch fusion is its simplicity: It’s small, which brings down costs, and it can be tested and improved quickly. It also avoids the issue that hot plasma destroys the walls of the confinement vessel.

The difficult part of this method is to get the geometry and timing worked out so that the pinch creates a density high enough for fusion. They told us they think it’s possible that this year they will reach a reaction gain larger than one, and think that an operating fusion plant by around 2030 is a realistic goal.

The company MIFTY in the United States uses a similar approach. Yes, that’s also an acronym. It stands for MAGNETO-INERTIAL FUSION TECHNOLOGY INC. I asked ChatGPT for a good name for a fusion startup and it suggested “NovaFusion” which I thought sounds pretty good. Unfortunately, it turns out to be a brand name of a hair care product.

Let’s then talk about inertial confinement, which tries to avoid the often energy-intensive magnetic fields necessary to hold the plasma. Inertial confinement is the method that is used by the national ignition facility that made headlines a few months ago by reporting they got more energy out of the reaction than went in.

The NIF uses a laser to fire at a fuel pellet. This method is good for research purposes, but the NIF lasers eat up a lot of energy that does not go into the reaction to begin with. So, it works, but it doesn’t produce net power. A couple of companies have ideas for how to do it better.

One company which uses inertial confinement is First Light Fusion, founded in 2011, has attracted almost 100 million US dollars, and currently has about 70 employees. They’re located in, drums please, Oxford.

First Light Fusion uses what they call the Big Friendly Gun. I don’t know about friendly, but it is without doubt big. It’s 22 meters long and weighs 25 tons. To fire it, you fill in three kilogram of gunpowder. If you ignite the gunpowder which launches a tungsten projectile which hits a fuel pellet that contains hydrogen isotopes. A lot of the design that makes this method work is actually not in the projectile but in the target. The target is designed to focus and thereby amplify the pressure generated by the shot.

Nick Hawker who founded the company says he was inspired by the pistol shrimp, which fights off much bigger fish with shock waves that it creates by snapping its claw.

According to their estimates they need to fire the projectile at 50 kilometers per second to achieve appreciable rates of fusion. Their current experiments reach a *few kilometers per second, but it does work. In 2022 they demonstrated that they could create fusion reactions this way, if not many. Their achievement was independently validated by the UK Atomic Energy Authority. So, there’s some way to go, but the approach seems to be working. And, assuming their estimate is correct, they’re not orders of magnitude away.

They also have plans for a power plant. For this, the target would be surrounded by a vessel. In the vessel, liquid lithium is circulated which absorbs the energy and heats up. The heat is then transferred to water that boils and turns a turbine which creates electricity. They say they should be able to produce 150 MegaWatts of net electricity by firing once every 30 seconds, and it would cost less than 1 billion USD.

There are some other, smaller, startups who work on inertial confinement, for example HB11 in Australia. And then there are the companies that combine field confinement with inertial confinement.

This approach is pursued for example by General Fusion based in Burnaby, Canada. They have a demonstration plant near Oxford which will soon be renamed to Oxfuse. General Fusion has been in the business for a while. It was founded in 2002 and has attracted more than 300 million US dollars, much of it by Jeff Bezos. They currently have somewhat more than 200 employees.

General Fusion also uses deuterium and tritium, but they use magnets to hold the plasma and then compress it with steam pistons.

Michel Laberge from General Fusion told us that the biggest advantage of this approach is that it’s cheap and that it’s sustainable. Their compressed plasma is completely surrounded by a fluid of lead and lithium, which absorbs almost all the neutrons. This prevents damage to the surrounding material. Their machine is also comparably small, about the size of a truck.

The company announced in December last year they’d reached 3 million kelvin for about 10 milliseconds. They need to reach about 100 Kelvin. Laberge told us that they plan to have a machine which produces electricity in the early 2030s.

Another startup that combines fields with inertial confinement is the Seattle-based company Helion Energy. It was founded in 2013. They have more than 570 million US dollars in funding, investors such as Peter Thiel and Sam Altman, and about 90 employees.

This company wants to use deuterium and helium-3 as fuel. Helium-3 is a rare isotope of helium and for this method to make sense they must use another reaction to produce the Helium-3. This reaction has the advantage that it doesn’t produce neutrons directly. However, if you mix those two gases and heat them up, a second reaction takes place which is the fusion of two Deuterium nuclei. And that does create a neutron. So in terms of radioactivity using Helium may not be much of a benefit. It has the disadvantage that the electric repulsion between helium and deuterium is stronger than between deuterium and tritium, so one needs to put considerably more energy into the collision.    

Their technique is to fire pulses of plasma at each other in a linear accelerator. The machine for this is about 40-foot long and six-foot tall. At both ends, a mix of deuterium and helium-3 is heated to plasma conditions, magnetically confined, and then accelerated to about 450 kilometers per second. At least that’s what they say on their website. Charged particles in magnetic fields tend to move in spirals, and if they do that, they emit radiation, and I’m not sure where that energy is going.

Be that as it may, the fuel collides in the center of the reactor where it’s further compressed by more magnetic fields. When fusion happens, the plasma expands and shifts the magnetic fields, which creates electricity.

The nice thing about this system is that it would create electric energy directly, which is more efficient than going through various heat conversions and turbines. The company is presently building a demonstration reactor called Polaris. They hope they will get energy out of the reaction by 2024. I’m a little skeptical about that.

In summary, there are about two dozen startups working on nuclear fusion, and they are trying a variety of different approaches. Many of them have quite advanced machines and have demonstrated that their equipment works as desired, but at the moment these are all demonstration machines. For now, none of them is anywhere close to feeding power into the grid. Most of them hope to build a commercially viable machine by the early 2030s.

I came out of doing this video being more optimistic about nuclear fusion than before. It seems likely to me that at least one of these approaches should work out in the end, though I haven’t been able to make up my mind which one’s the most promising. What do you think? Let me know in the comments.