Fission impossible? Meta wants up to 4GW of American atomic power for AI

Meta believes it will need one to four gigawatts of nuclear power, in additional to the energy it already consumes, to fuel its AI ambitions. As such, it will put out a request for proposals (RFP) to find developers capable of supplying that level of electricity in the United States by early 2030. "Advancing the technologies …

[thames]

There''s SMRs, and then there's SMRs

There are four SMRs already under construction just east of Toronto. These are 300 MW Hitachi designs, being constructed by a long established Canadian nuclear component supplier for a major utility. They are about the same size as the first generation of commercial reactors which were built in Canada (near by to these ones). They will use standard fuel from commercial suppliers. The economics and technology are not really an issue.

What is questionable is the very small SMRs being promoted by companies with no track record in the nuclear industry, no manufacturing facilities of their own, and which use specialized highly enriched fuel not available from existing commercial suppliers. These very small SMRs (100 MW or ever smaller) have very questionable economics as there are significant economies of scale in most power plants. They are simply too small and use very expensive non-conventional fuel.

If Meta want 1,000 to 4,000 MW, then three to a dozen or so 300 MW reactors could do the job for probably much less cost than the very small SMRs being promoted by some companies.

The whole point of a "small modular reactor" is the modularity, the "small" is just a way to get there. The idea is to build as much as possible in a factory and do the least amount of assembly on site. The company which can do that with the largest reactor will have the lowest operating costs.

What Meta should be doing is building their data centres in places which already have a good supply of electricity, instead of putting them in places with no electricity and then looking for someone to build power plants to serve them.

I expect that most of these micro-SMR companies will go out of business without having ever built an actual commercial reactor, while the companies offering ones in 300 MW and up sizes will be selling plenty to utilities.

[UnknownUnknown]

Re: There''s SMRs, and then there's SMRs

I suppose the argument/hope is that they maybe disruptors in the same sense Space X have been in Rocketry.

Perhaps more likely is that incumbents like Westinghouse, General Electric, Combustion Engineering, Bechtel, Rolls-Royce, Framatome, Technicatome, Hitachi, Toshiba manage to diversify into this.

I’m sure a SMR from Russia or China would be banned or maybe attract a 100%+ Orange Jesus tariff..

Geothermal anyone ??? Go see Iceland.

[Headley_Grange]

Re: There''s SMRs, and then there's SMRs

What if the new companies are disruptors in the manner of Elon's and his Cybertruck?

[UnknownUnknown]

Re: There''s SMRs, and then there's SMRs

https://www.gov.uk/government/news/rolls-royce-small-modular-reactor-design-completes-second-step-of-regulatory-assessment

Rolls-Royce’s seems to be making good regulatory progress, and despite a wobble during COVID primarily at the aero-engines business, the company is back to very good health…. which is more than can be said for Westinghouse or Toshiba.

[thames]

Re: There''s SMRs, and then there's SMRs

It's notable that Rolls Royce's SMR is a 470 MW design, about the size of a typical late 1970s reactor, and it uses conventional fuel from commercial suppliers. Again, the emphasis was on the "modular" rather than the "small" aspect of SMR. The design is centred around how to assemble large factory built components with the minimum of time and work on site.It is an entirely different beast from the unconventional micro-SMR designs.

[Vikingforties]

Re: There''s SMRs, and then there's SMRs

Talk to France about reactors going rusty.

[Sgt_Oddball]

Re: There''s SMRs, and then there's SMRs

The last thing we need with Nuclear is a company that's 'disruptive' with a fix fast mentality - something's are better when they never break in the first place.

[UnknownUnknown]

Re: There''s SMRs, and then there's SMRs

Agile + Nuclear - ;-)

[cyberdemon]

Re: Geothermal anyone ???

Take a few dozen barrels of high level nuclear waste

Dig a really deep hole

Throw barrels down hole

Hey presto - artificial 'Geothermal'.

[Vikingforties]

Re: Geothermal anyone ???

No joke icon needed. James Lovelock was asking for exactly that so he could go off grid.

[user555]

LMRs maybe?

Maybe those being built in Canada should be called Large Modular Reactors then.

[collinsl]

Re: LMRs maybe?

I'd go with MMR - Midsize Modular Reactor.

[nautica]

Perhaps there's a ready solution.

There exists a manufacturer which is the first and only to have its design certified by the U.S. Nuclear Regulatory Commission: Nuscale Power, founded in 2007. It is not a 'johnny-come-lately'; it has been working on a viable solution for a very long time.

Read about Nuscale here.

[Albripi]

Re: Perhaps there's a ready solution.

NuScale was a scam in the end. They even lied to their investors. And costs skyrocketed. Furthermore they don't want to built the licensed version anymore, but another variant yet to license.

[thames]

Re: Perhaps there's a ready solution.

The design of the 300 MW Hitachi SMRs which are being built in Canada at this time are also approved by the US NRC. The Canadian Nuclear Safety Commission (CNSC) and the US NRC shared information during the safety design review process. The US plan on building the same SMR in the US and there are MOUs between the two countries with regards to sharing of Canadian experience with this design with the US.

[John Smith 19]

So who *are* "Institute for Energy Economics and Financial Analysis" ?

IDK.

I'm not saying they are wrong, but I'd like to know whose funding them.

Just as I'd like to know who's behind any "independent non-profit," like the ones on Tufton Street in London that were so vigorous in their support of Liz Truss's unfunded tax cuts,*

before in her 49days in office she managed to multiply the BoE interest rate from 0.1% to 5%.

I'll note that PWR's (of which NuScale is one) are the dominant reactor design on the planet not because they are the best (in any given sense), but because Westinghouse's entire R&D budget was paid for by the US Navy. They have design features that will ensure that they are never easy or cheap to construct and will always have very poor thermal efficiency relative to any FPP, from the late 1950's onward.

Despite decades of talk about the benefits of SMR's there remains a gap in the market for something built using existing materials certified in multiple countries to create a design that's much easier to construct, certify and ultimately decommission (many reactors were built with little or no regard for this, making their dismantling a massive PITA) while also being cheaper to operate. Something, perhaps without the 40 000 "Elbow weld" joints (roughly 6000 of which are under lifetime inspection at about $5K/joint/inspection)?

*Which coincidentally seem to be bankrolled by billionaires whose tax bills stood to drop quite substantially.

[Albripi]

Just call Putin if you want reliable, cost effective and above all completed in time and budget nuclear power plants :P

[Anonymous Coward]

Wasted effort?

If any of these ever actually get built, the AI fad and perhaps even Meta will be over.

[IGotOut]

Re: Wasted effort?

yeah, but it's a great way to improve supply without the demand.

Let the tech bros pay.

[Alan Brown]

SMRs are a scam

There, I said it - they're snake oil, just like hydrogen or Drax are

SMRs aren't worth pursuing because the economics on the NON-nuclear side are all wrong

In short, steam turbines(*) need to be as large as possible to be economic and that means 1GW+ size power plants

SMR plants are TOO SMALL to be commercially viable and will end up as money pits if pursued

Regardless of the heat source(**) for thermal power plants, the fundamental requirements of the input to generators dictate how large a design needs to be. Nuclear power plants aren't 900-1200MW because that's as large as the reactors can be built, but because that's what the generators require

We absolutely and unconidtionally REQUIRE nuclear power, but SMRs are about the most expensive and inefficient way of achieving it I can think of - Over the next 20-30 years electrical demands in developed nations will increase by a factor of 6-8 as decarbonisation proceeds (we're already seeing the start of the domestic gas/oil heating phaseout in Europe) and demands in developing nations will increase to match ours (a factor of 20-30 times nigher than many nations are currently producing)

Water moderated nuclear power is problematic, not least of which because it's based on the unwanted waste products of the weaponsmaking process (Enriched uranium was kicking around taking up space and being a nuisance at 1950s separation plants until Alvin Weinberg came along. Depleted urankum is what's used to make plutonium at a ratio of 9kg to every 1kg of 3% "reactor grade" that's put into the civil program).

The late lamented Lester was a big fan of Thorium and LFTRs here in Vultureland back in the 2010s. China's currently testing a rebuild of the ORNL MSRE (TMSR-LF1) and have resumed where Alvin Weinberg left off at Oak Ridge in 1969 by testing directly thorium fuelled operation (the first 2 year run is a few months off completion and there appear to be no complications so far). Hopefully this is sucessful and scales without issue to GW size plants, as it divorces civil nuclear power entirely from its dependency on the weaponsmaking cycle in addition to being a couple of orders of magnitude safer than existing water based technology. At a projected 80% cost reduction over building and operating nuke plantsm, plus 99% less waste production AND the abliity to burn existing nuclear waste/depleted uranium, this would be cheaper than renewables and kickstart a new renaissance in energy availability, 60 years after Richard Nixon attempted to erase Molten Salt Liquid Fuelled systems form history

(*) At the end of the day, we still use steam to drive turbines to spin generators. Supercritical CO2 would be nice but it's not a commercial entity yet (TMSR-LF2 is slated to test a small 10-100MW S-CO2 turbine) and will probably be another couple of decades before it scales to useful sizes. Steam turbines are expensive to maintain and the relative cost of maintenance compared with income goes down as sizes go up - meaning the only current economic size is "as large as materials science enables them to be". That may change with S-CO2 as it's vastly less corrosive than water(***), but I doubt the economic fundamentals will be altered

(**) Coal, oil, gas, geothermal, nuclear, concentrated sunlight or magic unicorn poo - at the end of the day it's all about heat generation and extracting useful work from the difference between that heat energy and ambient temperatures

(***) Water at 350C and 20-100atm is extremely corrosive to the insides of nuclear reactors. "Wet Steam" is a major wear factor in nuclear power turbines - it's not hot enough to be dry and this is why most industry stopped planning around nuclear power since the 1960s - existing nuclear technology is more expensive than burning coal! Compared to water-moderated nuclear, the corrosion factors of molten salts are trivial to almost non-existent

[thames]

Re: SMRs are a scam

There are economies of scale to most electrical generating plants, regardless of the technology used. For nuclear reactors many of these are associated with the civil works such as site prep, roads, cooling water channels, grid connections, etc. This is why so many power plants have multiple units.

New "conventional" SMR designs have optimized their layouts to have reduced footprints in order to require smaller civil works, and also to use less concrete by making the reactor building more compact (so there is less to enclose).

Existing heavy water moderated reactors, such as the CANDU, which are in commercial operation around the world, can use thorium as fuel. India uses thorium as part of their fuel load in their CANDU derivatives. India do this mainly because they have lots of thorium but much less uranium, and they want to reduce their dependence on non-domestic supplies.

For most countries though, uranium is so cheap that it simply isn't economic to use the more complex and expensive thorium fuel. Reactors fuelled entirely with thorium will not "self-start". They need either enriched uranium or plutonium (reactor grade, not the special isotopes used in bombs) to start the reaction and create U-233 from thorium to run the main reaction.

At this time, it is cheaper to simply use abundant supplies of uranium in a once through fuel cycle than to set up the chemical reprocessing facilities to recycle spent fuel into mixed-oxide (MOX) uranium or thorium based fuel. France do it (for uranium based MOX), but that was a decision made based on national security to reduce reliance on uranium imports.

If uranium prices rise high enough, then there are already reactor designs that can use uranium or thorium based MOX fuels, and these reactors have been in large scale commercial operation in multiple countries around the world for decades. They just don't happen to be used in the US, which is why the US-centric media don't talk about them much.

[John Smith 19]

"They just don't happen to be used in the US, "

And TBF the odds-on bet is they never will be, despite never having experienced a Chernobyl or Three Mile Island event (Fukishima was more a once-in-a-millennium weather event)

CANDU's "crime" is that it has a (slight) positive coefficient of reactivity. At 100% if the power increases it continues to rise, very slowly.

The NRC is terrified of this because in the PWR (whose coefficient is negative) if it does go positive things go to runaway very quickly.

The fact CANDU keeps most of its moderator at <70c and atmospheric pressure (creating a huge cushion to absorb a lot of that energy) cuts no ice with the NRC, most of whose staff know nothing but PWR's (and some BWR's).

Unfortunately they share the same poor thermal efficiency as the PWR and BWR and the need for special build steam turbines.

[bazza]

Re: "They just don't happen to be used in the US, "

CANDU's use of heavy water was (is?) a bit of a bonus for the UK's Rutherford Appleton Lab's ISIS particle accelerator. Heavy water being disposed of by the Canadians was really good for target cooling in the ISIS particle accelerator.

Trouble was, after a tour in a CANDU reactor, the heavy water would be lousy with tritium (T2O, or DTO, or THO). So, the ISIS facility had to install a load of Tritium detectors in case of leaks (you really don't want to ingest Tritium), and because (in theory) leaks didn't happen these were on something of a hair trigger.

The trouble was that, at the time, next door, there was the UKAEA and its reactors, and because they had reactors of a particular sort they were prone to guffing off vast clouds of tritium such as you'd never believe (and were licensed to do so). And it'd waft over the fence, blow in through gaps in windows, etc. and set of the tritium detectors in ISIS. Every single time there'd have to be a check, search, etc in case, but of course nothing was ever found to be amiss.

This was way back in - I guess - the late 1980s, very early 1990s. I don't know if they use heavy water today in ISIS; they're not using uranium targets like they used to, so perhaps the D2O has gone too.

[Anonymous Coward]

Even a large company as entitled and self-important as Meta don't simply get to order up a nuclear reactor and have it installed when they feel like it. This will likely never happen given the over-regulation they will face.

[bazza]

Is it just me, or does anyone else think that the expenditure of 4GW on "AI" (for any purpose) is a complete and utter ****ing waste of resources?

To put that into context, so far as I can tell a ton of H2O produced from seawater using reverse osmosis requires 2.98kWh (see this presentation slide 19. 4GWh could produce 1.3million tons of fresh water, every hour. That would be quite a lot of irrigation water for farmers, or an awful lof of clean water for houses.

[IGotOut]

Yeah, but these are narcissistic asshole snake oil peddlers we are talking about, so "yeah fuck society, look at me!"