For me it’s the “seemingly” crazy unrealistic deadlines, which are achievable if they don’t actually try and hit net energy gain.

Use 20MW of energy to add 10MW into plasma, get 20MW of fusion, convert 30MW of heat into 10MW of electricity and they have reached their stated goal without actually achieving anything useful. And that’s assuming steady state operation rather than a fraction of a second pulse that briefly reaches their goals.

It’s exactly the same thing as a startup selling dollars for pennies and saying yea we’re going to make it up in volume.

ARC will be about the size of JET, and JET was built in four years, with most of that being just for the buildings. The test reactor, SPARC, will be about half that size.

Tokamak scaling is very well established. The output scales with the square of plasma volume, and the fourth power of magnetic field strength. Stronger magnetic fields also make the plasma more stable. JET already demonstrated a five-second plasma, which they only had to shut down because they have copper coils that would melt if operated longer than that.

Because of all this, many independent fusion researchers think SPARC will succeed in getting 10X gain in 2025. After that, the larger ARC should easily reach commercial levels.

Of all the commercial fusion companies, CFS is the most conservative one. Tokamak Energy is a close second, with a very similar approach. The other fusion startups are attempting approaches that have more physics risk, though many of them would have fewer engineering and economic difficulties if the physics does work out.

Simply scaling SPAR or ITER doesn’t result in commercial operation.

First you need fuel, global Tritium supplies are tiny and DD fusion is much harder.

Next stability is an open question, no Tokamak has ever operated near maximum capacity for even 1 hour.

Add to that serious material science questions, etc etc and even Q>100 alone just doesn’t actually mean much.

Now let’s just assume all of that is solved, you still need to actually ensure your design is economically viable. Simply producing energy from fusion alone isn’t enough which means you need to cheaply solve not only all the above but do so cheaply.

These are valid points. I was just responding to your statement that their time frame is "achievable if they don’t actually try and hit net energy gain." They are trying, and are likely to accomplish net gain about as fast as they say they will.

They actually have a fairly practical-looking design though. ARC will use FLiBe for tritium breeding and coolant. That will surround an inner reactor wall that's 3D-printed and replaced annually. MIT already demonstrated joints in the superconducting tape, allowing the reactor to be opened up to replace the core.

Economic viability is an open question for all tokamaks. Some of the other designs would have a better chance of achieving low costs if things work out, but there's a much bigger question mark on the plasma physics.

The other problem is that the thermal power density of ARC is about 1/40th that of a PWR reactor vessel. It's a much larger, much more massive/complex/expensive way to make heat.

Removing the need for large numbers of armed guards, giant concrete containment vessels, many many safety systems etc, should save some money.

It’s really not clear which would end up cheaper in the long run.

Spilling a thousand tons of melted lithium (and beryllium?) would be disruptive enough that containment, safety systems, and armed guards would still be needed. Also, not spending to protect your $100B facility from harm might make your insurance company nervous. (You did find an insurer, didn't you?)

If it actually cost 100B to built then your never going to break even.

As to containment, the lack of high pressure steam makes much thinner walls completely viable. You still need shielding around the vessel, but not a completely redundant system capable of containing highly energetic steam explosions.

In terms of risks, sabotage at a nuclear reactor can be vastly more expensive than just the cost of equipment. Modern reactors are reasonably safe, but not if people where actually trying to break them. Especially when you consider what someone could do with access to the fuel. It’s not weapons grade, but dirty bombs are horrific.

You do still need a very strong structure to resist outward pressure. It's isn't steam pressure, it's magnetic pressure, from JxB forces in the magnets. For ARC, the stainless steel supports for the magnets weigh 5300 tonnes, and comprise 3/4 of the mass of the entire reactor. The energy stored in that magnetic field is (I think) larger than would be stored in the steam of a PWR of the same thermal power output.

The building housing an ARC will be very large. The reactor itself is 20 meters tall, and the entire top half of it has to be lifted off and moved aside when changing out the vacuum vessel. All that lifting and moving will have to be done remotely because of radiation from the activated vacuum vessel, which then will need yet another shielded area where it can be broken down for disposal and the debris from that cleaned up or at least contained.

> would be stored in the steam of a PWR of the same thermal power output.

Safety system don’t just need to handle normal operations. The energy in a fusion reactors magnetic fields is very well known, a fission reactor steam explosion or potentially hydrogen explosions can bet vastly more violent.

Using Fukushima as a baseline. ~1,000 kg of hydrogen * 142 MJ/kg is a lot of energy and that was vented outside the primary containment vessel before detonating.

Granted this is not an inherent requirement for fission, but good luck convincing regulators it’s unnecessary.

The ARC reactor uses 380 tonnes of TiH2 as a neutron shield for the magnets. This fully decomposes at the temperature of the molten salt, which could yield up to 15 tonnes of hydrogen.

https://www.sciencedirect.com/science/article/abs/pii/S09203... Sorbom 2015 has ARC listed at 3.3 meter major radius. It probably won't be exactly that in the end, but it also won't be 20m tall. Something closer to 8 meters probably?

20m may be an overestimate, but 15m looks quite reasonable. Look at figure 1 in https://arxiv.org/pdf/1409.3540.pdf

> If it actually cost $100B to build then you're never going to break even.

IOW, we're never going to break even. You and I are paying now, and would in any case never get any of the revenue, if in fact any could be had.

But we are supporting the careers and research of plasma fluid dynamics physicists and their students, and a few of them might do other, actually useful, and anyway wholly unpredictable things, later. With a good measure of luck, none of those things will build up to any world-spanning catastrophes the way the steam engine did.

I give it 10% chance of being better than nuclear power in 30 years. That’s worth the investment even if the odds of a solid payoff are low, the upside is significant.

How? Better than fission is a very low bar.

p-B or D-He3 fusion, if achieved, would have application in the outer solar system even if not competitive here.

> How?

Lower risks should mean fewer NIMBY issues, which means wider adoption. Nuclear advocates miss that if fission played a larger role in energy generation it would also see more major incidents. Which then risks a backlash etc.

It sounds like you might know the answer to this: I vaguely recall hearing them say the tritium is created in the reactor from duterium? Is that a thing?

It is, deuterium fusion's final product is half tritium (and half He3).

But D-T reactors would mainly breed tritium from lithium. The CFS reactor, for example, will immerse the reactor in a pool of FLiBe salt, which also functions as coolant. Each beryllium atom hit with a neutron emits two neutrons, and when neutrons hit lithium atoms they breed tritium. General Fusion does something similar with lead in place of beryllium.

Thank you. So do they need to source tritium externally at all? I’ve seen people claim CFS will have to source tritium, making it expensive to operate. Can they breed all they need?

They'll need tritium for startup. After that they shouldn't need more if they can get the breeding ratio high enough, but that might not be easy.

In practice it means the mix of Li-6 and Li-7 in your coolant pipes has to capture more than half of the neutrons emitted, and not too many of the slow neutrons released in intermediate reactions get stolen.

Interesting thank you!

Unless a DT reactor can breed its own tritium it will be totally economically hopeless.

Yeah. Someone the other day said in a comment that commercial fusion will never be viable because tritium is so expensive. But it does sound like breeding is the plan at least with ARC so I don't know what that commenter was thinking.

Doesn't Moon have immense supplies of tritium? I bet Musk, with his landing rockets, has bigger plans than just launching a bunch of sats.

Tritium has only a 12-year half-life, and decays to He3. The Moon has large supplies of He3, which could be fuel for more advanced fusion reactors.

He3 barely exists on Earth. However, pure deuterium fusion is easier than He3 fusion, and its waste is half He3, half tritium.

3He on the moon occurs in regolith at a concentration of maybe 10 ppb. At that concentration, simply heating the regolith to drive out the 3He uses more energy than the 3He would produce in fusion, unless extreme heat recycling can be achieved (and good luck transferring heat into/out of powders in vacuum.)

Even if it could be recovered, all the 3He on the entire moon might power the world for maybe 100 years.

Thanks, didn't realize the energy balance was so bad.

That strengthens the point I was trying to get at, which was that there's no reason to go to the moon for He3 when, if we can get net power from He3, we can just make it in a D-D reactor and gain energy in the process.

My favorite 3He source would be "Planet X", if it exists and is an Earth (or perhaps even Mars) size planet out far enough that the temperature at the exobase is low enough to retain 3He in its atmosphere. Sure, getting there would be difficult, but this scenario assumes we have D3He fusion, which we can handwave would get the miners out there in a reasonable amount of time.

If it takes long enough to get there and back, folks back home might perfect p-B fusion in the meantime.

I don't think p-11B fusion will be possible. The groups attempting it look really sketchy (as in, talking about doing things that clearly don't work.)

It will take fundamental advances to get to p-11B fusion. A startup working today is anyway in no position to achieve such fundamental advances; that is more in the line of national laboratories, or would be if they were not fooling with ITER instead. Maybe some of the ITER work will be useful for other things.

For D-3He fusion to be useful, we have to hope that it will be economically viable to synthesize 3He, in the meantime. Arguably, we ought to be making and stockpiling tritium now so there will be enough 3He when we need it; but that might be too optimistic.

I don't think they'll have SPARC by 2025 but the physics work out. It's basically a bog-standard tokamak with superconducting magnets but with much more current capacity and therefore magnetic field strength from newer HTS. It's the fabrication of the magnets at scale that's totally new engineering and manufacturing; as we saw with Tesla that's quite hard. Plus the supply chains of HTS tapes aren't exactly mature.

I think both Tokamak and CFS have roughly the same strategy of using bigger magnetic fields. Given the scalings here https://royalsocietypublishing.org/doi/10.1098/rsta.2017.043... + JET working well it's a lot less risk than whatever most others are doing. Make a JET sized tokamak but have 5x the magnetic field strength gives a 625x gain in power, ideally.

I agree with what your saying, but that’s doesn’t mean the end result is even close to an economically viable fusion reactor. Look back in 1990 people could have spend a lot of money building a truly monumental reactor and hitting Q>100 briefly. Nobody suggested that was a good idea because all the other unsolved problems which is why ITER was being built in the first place. The point wasn’t simply to scale it was to test blanket materials etc in an environment very close to a working reactor.

SPARK is undoubtedly a more efficient design but it’s not a solution for any of the major outstanding problems.

Of course. They are basically a privately funded research group at this point. No one has any idea how to operate a profitable fusion plant.

By those kinds of metrics, Tokamak Energy haven't "achiev[ed] anything useful" either.

Note that both companies are aiming to build a tokamak using high-temperature superconducting coils, which will facilitate energy gain in a small, high-field tokamak. CFS has built the coils, but no tokamak yet. Tokamak Energy has built a tokamak, but it doesn't use HTS coils yet, it uses standard/super-inefficient copper ones (they're developing the HTS coils as a separate project and expect to marry them in a future prototype). This prototype has the same "make it up on volume" problem, it's just lossy in a different way.

Both have made important advances that demonstrate critical components that will be necessary to do the full thing, but different ones, and neither of them has demonstrated everything. If anything, CFS's piece is arguably more important, because other copper-magnet high-field tokamaks have been built before (e.g., Alcator C-Mod at MIT, essentially the predecessor to CFS's prototype), but nobody has built a full-sized HTS coil before.

I am all for fusion research, in the context of research. But CFS’s short term goals comes off as a publicity stunt which rubs me the wrong way. Like IBM’s Watson playing Jeopardy etc, where what’s achieved seems like a breakthrough for non exports.

It’s possible they could succeed long term, but I don’t think it’s particularly likely.

Sure, I think that's valid. I just don't think Tokamak Energy's announcement is less of a stunt. They've both made progress on a well-defined but ultimately not-that-big piece of a big pie. Just different pieces. If the act of making big press pronouncements around incremental progress is a stunt, they all do it.

I hardly view "build a machine to understanding burning plasmas and steady-state tokamak operation" as a publicity stunt. These are hard machines to build in the best of conditions. The design of any power plant must be informed of results from machines like SPARC.