RE: https://mstdn.social/@stshank/117162806916390598
This is of course where the private equity bros are pivoting to next, to hide their stolen loot (once the data centre bubble pops).
20
Replies
1
Boost
@cstross I hope so. Unlike all the prior schemes that just steal money from the general public (i.e. power/water/computers for AI and bitcoin), there is not really any public impact I immediately see for failure here. Maybe a drain on already rare metals?
If they happen to succeed and make energy generation trivial then that is...not really a bad thing?
For me, they cannot switch fast enough 😅
@zombie042 Non-aneutronic fusion reactors (ie. D+T or D+D reactions, which means all current designs) are a proliferation risk: they emit neutrons which cause secondary activation of reactor components, turning them into radioactive waste, *and* they can be captured by 238U cladding, creating 239Pu (aka bomb juice).
@cstross Ah, so more nuclear waste...ew (I worked out at Hanford for a while). I was thinking of the end goal of clean energy so yeah that would be bad too. 😫
@zombie042 There *are* better fusion reactions: if we could push past the idiotic space cadet obsession with 3He and get to a viable 11B + P reaction (at temperatures an order of magnitude higher than anything achieved today) we could have fusion without surplus neutrons using a fuel that is available in heaps and drifts on Earth (boron). But it's hard to make an economic case for it that can outcompete photovoltaic panels and batteries. Maybe for long-haul container shipping?
@cstross the sad part is I’d be a lot less bothered by the data center expansion if clean energy was a solved problem first.
(And yeah, I agree the current direction of most work on fusion isn’t promising, but maybe it’ll lead to something aneutronic)
@peter I disapprove of ALL economic bubbles on general principles—they're a grift, and the worst people make a killing at the expense of public trust. And real estate bubbles are just plain toxic. This is both, in one bundle!
@peter Clean energy is getting there: it's been reported that the combined cost of solar with battery (to provide 24x7 power) has dropped below natural gas in the US for the first time, which means, well, *clean* fusion power is already here.
@cstross yeah, not the cleanest reaction, but keeping the reactor 1AU away provides enough shielding. Plus it’s already running, and the total power output is more than reasonable. Now if we could just harness enough of it to replace all the other energy sources and have sufficient leftover to power the data centers.
Unfortunately we already ate most of the energy gains with crypto and are on track to make the planet uninhabitable (by humans) before we even finish the data centers on the books.
@peter The data centres are a construction industry/real estate investment bubble leveraging Nvidia's circular NPU order-book scam, and they're already alienating the public—badly enough that, for example, here in Scotland the government is considering an immediate moratorium on all new data centre developments: public opposition to local DC projects in the US is now over 70%. And there isn't enough power to run them anyway. So I expect the bubble to burst in the next few months.
@cstross that’s my expectation too, the signs are there (and it can’t pop soon enough for me). The real question for me now is does this cycle crash the global economy or do they prop it up again to make the next crash that much worse.
@cstross
It's the wrong approach anyway.
Just like fossil fuels, it unbalances the global energy budget - though much more slowly.
Solar does not. Neither does wind.
CFS seems quite real. I know a number of people working there. Looks like a good mix of plasma physicists and engineers. (About the other companies, I know little or nothing, and thus have no opinion.)
The physics is straightforwardly in CFS's favor: tokamak size scales with magnetic field like 1/𝐵³ (and the power density like the inverse 4th power). They've demonstrated novel high-temp superconductor magnets (yttrium/barium/copper oxide) at 20 Tesla field strength. They have built the manufacturing capacity to make a bunch of them, and are doing so now. Apparently they supply such magnets to other sites, as a little side business.
The test reactor is being built in Danvers Massachusetts. It's designed for a Q of 11, i.e., 11x more power out than in.
Details on reactor size vs field strength:
https://iopscience.iop.org/article/10.1088/1741-4326/ad2425
A recent talk by Clauser on tokomaks including the CFS SPARC being built now; scaling equations on slides 26-35:
https://suli.pppl.gov/2025/course/Clauser-tokamaks-2025-v2.pdf
@cstross I thought they were all about the quantum computing bubble next. Someone should start some FOMO rumour for them about AGI’s imminent invention of warp drive 🤣🤣🤣
@cstross I was so sure it'd be quantum computing.
That with all the talk about how quantum effects and consciousness might be connected.
That with all the talk about how quantum effects and consciousness might be connected.
@markus Right now they're in throw-money-at-ALL-the-walls-and-see-what-sticks mode. The stench of panic hanging over the markets (aggravated by Trump's latest Canada tariffs hitting toilet paper prices).
@cstross
I'm less angry about that. Some physics and engineering may come out of it.
Whereas using staggering amounts of energy to produce BS...
@iinavpov This is a complete 180 from my position a very few years ago, but right now I don't believe we need another nuclear power source. Fusion isn't inherently clean, can be used to breed plutonium for weapons, and relies on large centralized generators—heavy industry, in other words, with its toxic political and economic side-effects.
@cstross
Now, I don't have a strong view on that, except I have little belief that there will be any machine actually operating coming out of any of these startups.
I just think that it's a vastly less destructive way of spending money when compared to AI.
I think that if any breakthrough happens, it'll serve in some form many decades from now...
There have been some legit advances in fusion lately, most notably at the National Ignition Facility.
I'm OK with fusion reactors. They've been working on it so long they've earned a win.
It would be galling of course for present day crooks to take all the credit. That's what happened with recent neural net breakthroughs, after decades of work by real scientists.
@cstross I have a friend who works for CFS and I have learned about so maybe different ways that welding two pieces of metal can go wrong in the past couple years. I also learned about the Massachusetts Board of Boilers, which I think would be a great name for a villainous organization in a novel (not to tell you how to do your job)
