I was hoping the article would go into the current state of boron reactors out there such as [0], but it kept reiterating the disadvantages of D-T reactiors.
Also interesting is using AI for plasma control. It didn't exactly say how they're using AI, which is nice to know when talking about real-time code that handles temperatures in the millions of degrees.
> Artificial intelligence software, recently developed by TAE with help from Google, should keep the fusion fuel stable even at the 3-billion-degree temperatures needed in a commercial-scale power plant. “A year ago I was concerned about the ability to control the machine,” Binderbauer says, “but no longer.”Courtesy of TAE Technologies
which implies that it's being used for plasma control, is a lie.
No, it's used for control. But you should be thinking more like "process control" than "intelligent" control. Or to be more specific, deriving an optimal process control model via pretty hard-core plasma density reconstructions; my understanding is that they track plasma density fluctuations and tweak parameters in real-time, but I would be surprised if the TF model is in the control path.
It seems a bit overly-aggressive to accuse them of lying because some pop-sci journalist heard tensorflow and went straight to "artificial intelligence." And even there, it's an understandable and common hype.
It doesn't take away from the point that they're doing some really interesting and novel computational physics to make this thing work.
The thing I never understand about these fusion reactors is how they plan on generating energy from them, assuming they exceed break-even. In a conventional nuclear reactor the coolant is piped to a steam generator which then drives steam turbines to generate power. This is a relatively obvious design and it was pretty well understood from the start of development how the power generation would work. Of course, there were a lot of engineering challenges to overcome to make it work in practice. While fusion itself is a harder problem (as the lack of success for decades has demonstrated), it seems like the power generation aspect has many new challenges as well.
The typical design for a fusion power plant that runs on deuterium tritium fuel is to place a lithium "blanket" around the plasma. 80% of the energy released in the deutrium - tritium fusion reaction comes out in the energy of a neutron which would be absorbed in the blanket, heating it up and also generating tritium fuel which could then be fed back in as half of the fuel (the other half being deuterium which is abundant in seawater).
You would then run a heat exchanger from the hot lithium to create steam to then turn a turbine and make electricity.
There is also direct-capture in some fusion designs where the charged by-products are electrically decelerated, producing current. I don't believe this is possible in the standard Tokamak design, but the efficiency gain is a selling point for those designs which do allow it.
That’s right, there are other fusion fuels that have only charged fusion products which could be directly converted to electricity. This is in contrast to deuterium - tritium which releases 80% of its energy in a neutron which has zero charge.
Examples of fusion fuels whose main reaction produces only charged products are deuterium helium-3 and proton boron-11. These reactions however require higher temperatures and better confinement characteristics.
The reason why deuterium tritium is the major focus of most (though certainly not all) research is that it has the highest reactivity at the lowest temperature compared to other fuels. Unfortunately it produces a high energy neutron which makes the conversion to electricity more complex.
Direct capture cannot capture a neutron, as it is not a charged particle. Charged particles are relatively trivial to block with the lithium blanket. Not sure how direct capture could possible be more efficient. Do you have a reference that explains what you are talking about?
There's just the matter of getting the temperature up to 10x of an ordinary D-T fusion plasma. Non-equilibrium reactors like the polywell try to bypass the problem entirely, but (to my knowledge) it's very hard to maintain a non-thermal state.
The people working on fusion generally think it is a solved problem. Namely, fusion is a heat source and we have numerous technologies which can convert a heat source to generate electricity a number of which can be connected to the fusion generator with "some engineering". That's why they talk about the parts which aren't yet solved.
I used to work with under a guy who worked on ITER, they were all aware of that part of the problem. It just wasn't part of the research they were engaged in.
The power density of ITER will be a horribly low 0.05 MW/m^3. Of course getting heat out is easier if there's so little heat being produced. But even then, the divertor at ITER will be challenging, as will dealing with heating from disruptions.
As mentioned elsewhere in these comments, the usual plan is to convert the fusion energy to heat in a blanket, through which coolant flows.
This sounds simple, but actually shows one of the biggest problems with fusion. The wall can handle only a certain power/area before the engineering becomes too difficult. And because the wall has to be outside the plasma, the ratio of the wall area to plasma volume will be quite low. In contrast, in a fission reactor, the "wall" is the walls of the fuel rods, which are only about 1 cm in diameter. There is lots of surface area through which to convey the heat.
The square/cube law comes into play here. Fusion reactors will have terrible volumetric power density, compared to fission reactors (or, even worse, internal combustion engines like gas turbines.) As a result, fusion reactors will be massive, expensive things, compared to the competing source of heat.
The figures for existing reactor proposals illustrate this. The volumetric power density of a PWR primary reactor vessel is ~20 MW/m^3. In constrast, the power density of the ARC proposal from MIT (counting the volume of the reactor + blanket + magnet shield + magnets + support structure) is around 0.5 MW/m^3.
If someone breathlessly hypes some fusion reactor design, ask them what its volumetric power density is.
This is underappreciated. Achieving sufficient confinement pushes toward larger volume/surface area, but getting high power density pushes the other way. In a way, the fact that 80% of the DT power output is in fast neutrons becomes beneficial -- that fraction is absorbed volumetrically, so only 20% must be conducted through the surface (or convected, if it's a liquid).
The neutrons cause their own problems, though, and these problems interact with the purely thermal ones.
For example: the plasma facing surfaces on ITER are backed by a CuCrZr alloy with high thermal conductivity. The amount of direct thermal power this innermost wall can withstand, from the plasma and photons, is proportional to this thermal conductivity, since stress there is proportional to the thermal gradient.
But in an actual reactor, designed to operate for long periods of time, this alloy cannot be used, as it becomes too activated. Instead, DEMO is going to use a kind of reduced activation steel (Eurofer 97) in which the thermal conductivity is an order of magnitude lower. This is one of the things pushing DEMO far outside the range of potentially competitive cost.
Yup. There's also the need to maintain transparency of the first wall to neutrons so the tritium breeding ratio stays above 1. However, a very thin wall is easy to breach in the event of plasma eruptions, or even long-term erosion due to ion bombardment.
It seems to me that a thick (~ cm at least) flowing liquid first wall is probably the way to go for achieving high power density. The liquid has its own problems though, such as corrosion compatibility with the underlying structural material, the need to have low vapor pressure at high temperature, the possibility of splashing, and MHD effects if the liquid is conductive.
I agree. Also, liquid surfaced walls are probably required for any pulsed fusion approach, which at a given average power/area will have much higher instantaneous power/area. This drives one toward LINUS-like approaches.
Compact Fusion Systems recently got funding to push this approach. On LinkedIn they announced purchase of surplus high-energy capacitors (from the old Shiva Star at AFRL) to build a prototype.
They seem to have taken to hear some lessons learned from previous experiments (cannot assume lossless compression, must pay attention to RT instabilities, must think about vaporization of first-wall).
One downside to a fully liquid blanket: the peak temperature is limited by the vapor pressure of the first wall, which in turn is limited by the impurity tolerance of the plasma. In a non-imploding device, it would be possible to run the majority of the blanket at a higher temperature & separate that from the cooler first-wall liquid. Of course, this puts some solid material closer to the neutrons, so YMMV.
I've seen their slides before, and I was impressed by their engineering-oriented approach. The engineering obstacles to fusion are so great that (IMO) one should use them to drive the design, rather than taking a physics-oriented approach and worry about the engineering later.
From the article
"The plants would convert the raw fusion energy into heat, then use the heat to boil water, spin a turbine, and ultimately to generate electricity that can light your home and charge your phone."
Yeah, but raw fusion energy to heat is not nearly as trivial as it sounds. And given how tricky it is to sustain a fusion reaction it seems like it will be even harder to do it when you're dealing with moving heat from the source of the fusion reaction to wherever your steam generator is.
I believe there are two main ways of cooling: convection and radiation. Convection requires a gas or liquid carrier to be in contact with the hot substance; it seems like it would be very difficult to get water (steam) into direct contact with the plasma to transfer the heat, and then to carry the heat away without the radioactive plasma along with it.
Radiation is a function of temperature (to the 4th power) and surface area. It seems even harder to channel the heat from the plasma to the engine this way.
I don't know how this one in particular is planning to do it, but the Polywell wiffle-ball design uses magnet coolant as a heat transfer medium. I'm sure I've heard of a design that uses charged particles spat out of the fusion reaction directly as charge carriers, too.
I have yet to read an article or paper which has convinced me that energy generating fusion is possible at low costs and with materials that we currently know of. I don't see why anyone (including governments) should invest in these types of projects. Not only is the science extremely complex, the implementation is a huge engineering challenge. We already have a huge fusion power plant at the heart of our solar system and we can harness its power with some pretty cheap solar panels.
There are a few reasons why the investment is flowing.
1) New enabling technologies, including high temperature superconducting tape, algorithms for plasma control and diagnostics which take advantage of new hardware (GPUs), and advanced manufacturing techniques are now available.
2) Optimism that private companies can synthesize the past 70 years of plasma physics research with these enabling technologies to develop transformative approaches to fusion.
If you're interested I wrote a short article about this topic a few months ago,
> algorithms for plasma control and diagnostics which take advantage of new hardware (GPUs)
I don't know about that. Here's a few bullet points from [1] (which someone else linked to in this thread; it describes the approach taken by these guys at TAE) that don't inspire a whole lot of confidence:
"Our prior is “reasonable”, but is it really the marginal distribution over all possible plasmas? hahahahhahahaha. We model many effects, but plasmas are complex beasts and we do not model all. We only have one measurement, of much smaller dimension than our unknowns. We never sample from the tails. takes too long to get samples. by definition you can’t really validate them. Will we ever know we’re right about anything? we have zero golden data"
I remember reading recently that a Nobel Prize was won a few years ago for work with chirped pulse laser amplification using titanium sapphire lasers that can apparently achieve nano or microsecond energy pulses in the terawatt range. A potential contender for a non-fission fusion spark, but still does not solve the containment problem. The article says the laser could generate a magnetic field somehow?
And a startup hoping to try it, run by the guy who came up with the idea decades ago: https://www.hb11.energy/
There are several groups doing experiments with it, and it seems to be going really well.
There are two lasers. One hits a target that generates a magnetic field; it'd be hard to describe without a picture but see the articles at the first link. Basically the laser blasts electrons off a metal surface, they hit another surface and flow through a coil. For a nanosecond there's a 4000 tesla field. (An MRI machine generates around 3 tesla.)
The second laser is faster and more powerful: 10 petawatts or more, for only a picosecond. That hits the fuel. It's enough to kick off fusion by itself, but the magnetic containment creates an avalanche effect that multiplies output. Then it all blows up, you harvest the energy and cycle in another target.
Thanks for the link. It is a much more thorough explanation.
How fast is the fuel used up within the field? Would there be a way to inject the actively fusing reaction with a steady fuel input rate for long term generation (neutron bombardment embrittles superconducting metal containment with the D/T reaction, unlike boron encased in supposed laser induced magnetic field?)
I'd imagine this would occur in a sphere (closed and contained). Tokamak designs aren't spheres, but also closed relying on magnetism to push back against a reaction that is pushing out as fusion occurs:
To produce thrust - what if it was a half sphere somehow? Propellant implies ejection of something, and a fusion reaction ball is magnetically interactive, with no radioactive material byproduct? What if a fusion thruster harvested some energy from the reaction to "push" back against an actively fusing pellet feed rate? Could this propel a craft or am I missing something fundamental here?
The magnetic field disappears in a nanosecond, the fuel pellet gets used up, and it explodes, destroying the coil that generated the field. So you just send in another target assembly and fire the laser again, every second or two.
There's nothing wrong with a pulsed system like that. Lots of fusion designs are pulsed. A gasoline generator with an internal combustion engine is a pulsed system too.
Add a magnetic nozzle and you could definitely turn this into a rocket. Thrust would be low but efficiency very high, so it'd be useless for launch but great for long-distance travel.
There is a very general problem with pulsed systems for fusion. The issue is that plasma-facing surfaces are confronted with extreme instantaneous power levels. The depth to which heat can diffuse is proportional to (pulse length)^(-1/2). A nanosecond pulse will deposit heat in a tenth of a micron thickness, or less.
This forces any fusion reactor that uses pulses to have a sacrificial ablative layer on these surfaces that must be renewed (and to deal with the forces from the explosive vaporization of this thin layer). This is problematic if the reactor also requires high vacuum. The scheme for p-11B fusion that this subthread was talking about, for example, has been presented with a direct conversion scheme that uses a megavolt level vacuum capacity. Imagine what happens to such a capacitor when its surfaces flash superheated vapor.
Interesting. But if the direct conversion works, then the magnetic field removes most of their kinetic energy from the charged particles before they get to the walls. All it gets is the x-rays. There has to be some radius where that's no longer a problem. Each pulse in this design would be about 300 kWh; offhand I don't know what percentage is x-rays.
If it's too hard to maintain vacuum, then reverting to a plain ol' thermal cooling could be a backup plan.
A magnetic field leaves the kinetic energy of a charged particle unchanged.
The electric field is supposed to reduce the energy of the alpha particles, but (1) the alphas from p-11B are not monoenergetic, and (2) what is keeping the electrons (that are inevitably liberated in the extremely energetic explosion of the target, the impact of the alphas with the collecting electrode, and photoelectric emission from all surfaces exposed to photons from the plasma) from shorting the whole thing out?
To produce thrust - what if it was a half sphere somehow? Propellant implies ejection of something, and a fusion reaction ball is magnetically interactive, with no radioactive material byproduct? What if a fusion thruster harvested some energy from the reaction to "push" back against an actively fusing pellet feed rate? Could this propel a craft or am I missing something fundamental here?
Sadly, that huge fusion plant is offline half the time for any given location on the Earth, and is severely degraded by cloudy weather. Also using it to power the world will require large areas of land to be dedicated to collecting its output.
Space advocates seem to have been going off the idea of that over the past few years. I'm not sure how much of that is due to a better appreciation of the downsides, and how much is due to the very rapid improvements we've been seeing in ground-based solar.
A couple of recent (late 2019) articles from Casey Handmer:
It's worth reading the comments too; there are some interesting points there, although I think the game-changing ones rely on new tech and/or well-developed lunar industry.
Space solar is a way to get the most out of expensive PV cells. But PV is no longer expensive. It's cheaper to just overbuild on Earth and transmit energy through time (storage) rather than through space (microwaves).
Yes. I'm aware of that. But while economically usable fusion remains SciFi so far, space based solar just lacks some bootstrapping. Which seems to be more attainable as time passes.
If "we" really wanted this, we could do it now because the technology is there. There is no what if like in any path to fusion.
edit: especially considering all the hype about colonizing Mars, or a Moonbase.
That seems like it could have more difficult engineering problems than a fusion reactor! Seems like a quick way to accidentally microwave a desert into a sea of glass. Hope you don't put anything remotely near the power receiver on earth.
Power density on the ground would be way lower than you're thinking.
The leading proposals, like SPS-Alpha, put a gigawatt-scale power station in geostationary orbit. Power is beamed to ground via microwave. The receiver station has to be kilometers wide, which puts the power density low enough so birds wouldn't be harmed; even that amount of focus is only doable by sending a pilot signal from the receiver, so the beam can't trivially be redirected to other targets. Despite its size, the receiver would be inexpensive, since it's mostly wire antennae.
Hey, think positive! If microwaves prove too dangerous, and they finally do manage to build a cable from the orbital power station down to earth, they have a space elevator as well!
There are other options for using the sun's energy. We can grow trees to burn, we can optimise windows and window coverings to use solar energy to heat and cool buildings, we can put solar panels in the dessert, or in the ocean, or above roads or in space (but that's not very cost effective).
My go to book on energy is this one http://www.withouthotair.com/ It gives good numbers for all the energy that we could generate from renewables and all the energy we currently use.
There is a lot of land in the world. And definitely have the capacity to build the supply chains. It's just that it's slow and expensive while oil, gas and nuclear is fast and cheap.
We need policy changes. Or in a very ideal world, fusion.
It's startling how quickly this happened. Even ten years ago it wasn't true. Many people have not updated their priors yet and still think nuclear is a realistic option. At this point, nuclear is a long shot, which is why I refer to SMRs as HMRs, "Hail Mary Reactors".
They spin a large ball of molten lead using pumps and shoot hot plasma into the vortex that forms in the middle, then strike the walls of the lead chamber with carefully timed steam pistons to make an implosion pressure wave not entirely unlike Fat Man’s design, except reusable and for fusion.
The fusion heat and radiation gets absorbed by the lead and they run a heat exchanger on the pump loop for power generation. They also think they can breed fuel with some lithium in the lead.
I love their plan aesthetically but I’m not qualified to judge how feasible it is. It’s very satisfyingly physical though.
I was withholding judgment on General Fusion. They had a concept that, at least in principle, would allow engineering limits on the power density at the first wall to be evaded (since the first wall would be thick liquid metal, not a solid that could sustain damage.)
But they gave up on their original "acoustic" compression scheme, and now will compress the plasma via subsonic motion of the liquid metal. This scheme involves a solid conductor going down the middle of the chamber. It will be exposed to orders of magnitude higher radiation flux than the first wall of mainstream fusion concepts, as well as pulsed loads from magnetic fields up to 100T (which correspond to pressures far higher than the chamber of a gun, and higher than the deepest point in the oceans.) Getting this conductor to survive even one shot would require heroic engineering; keeping it cooled and together as its material properties rapidly degraded would require superheroic engineering.
I heard that the shock wave was spalling off the front few cm of the liquid metal wall as a spray of jets/droplets -- the shock reflects from the vacuum interface & where the wave interfered destructively with itself, the pressure dropped to zero, and cavitation occurred. This happens in a random & uneven fashion.
The Turchi mentioned there was associated with Los Alamos, where the LINUS concept was investigated. General Fusion seems to be moving in that direction, so this slide from Turchi at Compact Fusion Systems could be of interest. It has a focus on practical engineering and cost issues that I find encouraging.
Do you end up with a huge ball of radioactive lead at the end of the day? From what little I know of fusion, it seems you still end up with irradiated lead or other shielding.
(And yes, I know that coal releases radiation, and that the installation of solar panels and wind turbines is, currently, a dangerous job, etc.)
Huh, I suppose you do! I had a naive short-circuit in my head that "neutron flux is bad because it destroys your your materials. If your material is molten lead anyway who cares?"
I also didn't know lead has such a low neutron absorption cross section.
For what it's worth lead with extra neutrons seems a lot less scary than eg uranium fission waste. Stable isotopes 206, 207, and 208 represent 98% by abundance, 208 has the lowest cross section, and 209 has a 3 hour half-life into Bismuth-209 which is nearly stable (2e19 year half-life). So it seems almost all of your neutron captures just make other stable lead isotopes or briefly-terrifying 209 that's totally safe after a couple days. You only get real scary stuff if the trace amounts of undecayed 209 manage a second capture.
Edit: I should add they want to mix lithium into the lead to absorb neutrons and regenerate fuel
Lead has a low neutron absorption cross section, but it has a fairly hefty inelastic scattering cross section for fusion neutrons (for fission neutrons, which are less energetic, it just elastically scatters them, with very little moderation). It also causes some (n,2n) reactions with fusion neutrons, which would be nice to make up for tritium losses.
The one that gave me most hope from this point of view was the Polywell design. It made some waves a few years ago then got classified and sputtered out, then this happened: https://en.wikipedia.org/wiki/Polywell#%22Final_nail%22?
I'm entirely with you as far as tokamaks go - they're relying on unobtainium for a significant component - but I think other designs are possible, just under-researched.
Polywell was 100% BS, I'm afraid. The whole concept of forming a "virtual cathode" made no sense, due to plasma conductivity shorting it out. The group at Sydney who did experiments with similar devices concluded that maintaining the virtual cathode in a reactor-scale Polywell would require a 20 gigawatt electron beam.
Are you aware of https://en.wikipedia.org/wiki/Wendelstein_7-X? It is well on its way to produce sustained plasma discharges of multiple minutes. The stellarator design has some decisive advantages over Tokamaks.
I like how the chamber and magnet design looks like something made by Cthulhu in cooperation with Salvador Dali himself. Even just due to that it should work. :)
That sounds like engineering and scaling challenges, not inherent limitations... I mean, they clearly have the magnets figured out and so far there doesn't seem to have been a lot in the way of unforeseen roadblocks.
That sounds like the dismissive attitude a plasma physicist, for whom mere engineering issues are trivial and of lesser importance. But engineering issues are perfectly capable of ruining the prospects for a technology. Fission, for example, faces nothing BUT engineering issues (where cost is considered a part of engineering.) The equivalent of breakeven and ignition was achieved for fission in 1942.
DT fusion faces fundamental engineering obstacles, obstacles that are not solvable by tweaking the confinement scheme. Scaling up makes the main problem, power density, worse, not better (see a later comment by me on this issue.)
> Even if renewables come to generate the majority of our electricity, they need to be combined with so-called dispatchable power that can be switched on at any time.
There is plenty of lithium in the oceans — among other places — and no shortage of iron or phosphorus (ignoring agricultural use of P which is much larger), so that it is realistic to store weeks or months of energy in LiFePO4 batteries, even at a global scale. What we lack here is investment.
I agree that solar and wind have disadvantages, and they can't be used to cover 100% of the energy demand, but they are available right now, whereas controlled fusion has been "10-30 years away" for the last 60 (?) years.
whereas controlled fusion has been "10-30 years away" for the last 60 (?) years
One standard internet response to that is this graph[1]. No funding, no results.
The international community decided to pool ressources with ITER. The founding members? The US, EU, Japan and the Soviet Union. The Soviet Union collapsed, the US pulled out for a couple of years, other countries (eg Canada) joined and left gain, Brexit has put the future of JET (where some relevant preparatory experimentation is done) in doubt, etc.
Hm. Personally I wonder who those people are, who allegedly insist on relying solely on wind and solar.
TBH, this sounds like a classical straw man to me. Who'd want to actively dismantle existing hydro power, geothermal-, waste-to-energy, biomass power stations, etc.?
That's not enough to power Germany, for example. To de-carbonize our electricity production completely, we need some yet to be determined additional energy storage mechanism.
Power-to-gas would work very well if we can make it efficient enough (and control potential gas leaks), because a lot of the infrastructure is already in place.
If that doesn't pan out, there's other technology that sounds promising (cryogenic energy storage, perhaps?), but in any case, there's still research to be done...
Germany subsidized renewables when they were expensive, and they are still paying for that. But wind and especially solar costs have crashed since then, having been driven down their experience curves by these subsidies and then the general subsequent growth in demand. You can thank Germany for this act of altruism that made renewables so cheap for the rest of us.
Also, when comparing costs, you must not allow fossil fuels to get away free without being charged the cost the CO2 emissions are imposing on everyone. Among the CO2-free options, wind and solar are now several times cheaper (per kWh produced) than new nuclear.
> Germany subsidized renewables when they were expensive, and they are still paying for that.
Yes, currently 2.05 ct/kWh. (I am German, I should know.) Electricity costs about 29 ct/kWh, constantly getting more expensive. Even if we ignore the subsidy for existing unreliable power generators, 27 ct/kWh is ridiculously expensive (roughly 35 US-ct/kWh). And no, I'm not going to thank anyone.
Isn't this a core demand of the Extinction Rebellion crowd? That we actively dismantle hydro (because it interferes with river ecology), and all the fossil fuel and waste processing power generation? Not sure about geothermal - but I'm sure there's some people out there convinced it's a bad thing.
I totally disagree. Sure, sometimes I just want the technical facts... in those cases, my first stop is Wikipedia. Wikipedia has articles about TAE the company[1] and as well as the Tri-alpha fusion process[2].
But when it comes to fusion energy, there are many possible approaches, yet few are actually being pursued with serious commitment of resources. The future of our species may depend us trying more things harder. And for this reason I find it quite interesting to read the "human interest story" behind the non-mainstream attempts being made, what their motivations are and how they are managing to get the necessary resources and to keep up the commitment throughout the long road to uncertain success.
I replied above but it was meant for you. I'll leave it, perhaps it might help the parent understand why there is much value in knowing about the people behind the science.
Sure, if the title was, "Workings of a boron hydrogen fusion reactor."
It's not, it's a life interest story, obviously so from the title. History of science isn't a review article but it is still history and some scientists like me find history to provide a lot of insight into how we got where we are.
Right? I mean people don't do science, it just happens.
And while we're at it, why do we care about the background of entrepreneurs around here? I mean just walk me through the business model and give me your KPIs already.
> I have zero interest in the man's childhood or what he or his desk looks like; just tell me about the science.
I find this trend has been true of most journalism. Every story has an elaborate narrative built around it to provide "context", most of which has little bearing on the facts.
Someone should create some article writing software that lets you distinguish narrative parts of an article from the factual parts, so readers can selectively show/hide each one.
I recommend learning to speed read. Your brain is capable of working out what the subject of a paragraph is before you've "read" it. Learning to trust that, and skip-read articles only slowing down for the interesting bits, is a great life skill :)
Also interesting is using AI for plasma control. It didn't exactly say how they're using AI, which is nice to know when talking about real-time code that handles temperatures in the millions of degrees.
[0] - https://www.hb11.energy/our-story