Amazon makes $500M bet on itty-bitty nuclear reactors to fuel cloud empire

With energy scarcity threatening to derail datacenter ambitions, cloud providers are looking for salvation in the atom. On Wednesday, Amazon announced plans to support the development of three new nuclear energy projects, which it says will see the construction of several new small modular reactors (SMRs). Just as their name …

[Dr. Ellen]

If a reactor is small, modular, and made in a factory, the government will only have to approve one set of designs. Large reactors are designed for particular circumstances, so each must be examined and approved separately. Cuts out a lot of red tape.

[DS999]

They should have done that decades ago

Building a bespoke design for every project was always ridiculous.

[NoneSuch]

Re: They should have done that decades ago

Expect the Sierra Club et al. to object in 3... 2... 1...

[MachDiamond]

Protests in 3,2,1.....

The anti-nuke crowed isn't protesting size, but that a power plant has wicked nasty radioactive materials at all. Each and every one of these smaller ones will have a corresponding lawsuit attached over every aspect of the plans. The fees from all of the blood sucking lawyers won't scale with the output of the reactor so it just means more BSL's/billable hours.

I'm still anxious to see much more modern designs. The father of the PWR, Alvin Weinberg, soured on them once there were better alternatives and the shortcomings came to light. In the US, the regulatory agency is notorious for the use of the word "no". If they say yes, their asses are on the line, saying "no" has no repercussions so good luck getting approvals for a new design.

Just what we need, scores of small reactors dotting the landscape to power AI, crypto mining and cloud services.

[Anonymous Coward]

Re: Just what we need, scores of small reactors...

and each one is a viable target for those who hate the current (or all) governments.

Just say'in...

[Andy_bolt]

Re: Just what we need, scores of small reactors...

There are many other facilities that could be damaged in a way that could damage adjacent properties. It’s the invisible bogeyman of nuclear that tabs attention. The cities where nuclear bombs were dropped have rebuilt. Many people did suffer from the radiation but it has not left the land permanently unusable.

[MachDiamond]

Re: Just what we need, scores of small reactors...

"Many people did suffer from the radiation but it has not left the land permanently unusable."

Permanently is in the eye of the beholder. If your home that would sell today for a nice price dropped to zero and you had to move means you would lose a massive chunk of money and possibly be made homeless. 80 years from now, you wouldn't care that the area is perfectly fine for habitation. Maybe it's just that the new SMR up the road makes too many people nervous and you now owe more on your mortgage than you can sell the house for as prices in the neighborhood drop. It more about perception than reality.

[steviebuk]

problem is

Letting Amazon and Google "manage" these.

When that side of the business starts costing too much we'll all be reading here, on our Pip-Boys and in our bunkers about the fault that happened during the week when Google/Amazon's small nuke reactor to power their shitty AI, had a melt down due to a failed software update. Because the month before they'd laid off a few 1000 staff at their nuke power sites due to "costs".

[DS999]

Re: problem is

Does being a utility automatically make one more qualified to operate it? It isn't like the regulations won't still apply to an Amazon or Google operated reactor, so they won't be able to apply software updates willy nilly or cut staffing below required levels.

[steviebuk]

Re: problem is

When they have billions to lobby anything is possible. And if you had someone like Trump is in power then all bets are off, look at Joe Manchin for example.

[DS999]

Re: problem is

If Trump is re-elected our problems are far far worse than who is regulating the nuclear industry.

[MachDiamond]

Re: problem is

"Letting Amazon and Google "manage" these."

The requirements for operating a nuclear power plant are vast and tedious. One slip up and the license to operate gets suspended so being casual about it isn't going to fly and a lot of supervision is going to come down on them initially as a new entity in the nuclear generator arena if they manage to get licensed. I'd be very unsurprised if operation is out-sourced to an operator that already has a lot of experience.

[collinsl]

Re: problem is

That's why the article says that Amazon's reactors "will be owned and operated by partner Energy Northwest, which represents numerous state public utilities in the region." No doubt this partner company already has or deals with nuclear reactor organisations which have the requisite skills and certifications.

[Michael Hoffmann]

Whatever happened to the hype of Thorium reactors from a few years ago?

Weren't they supposed to be the safe, quick and cheap alternative?

Wiki says that the investment for development hasn't come forward - well, these companies have cash in spades. Also says that, while safe, it's apparently a bit too easy to skim off weapons-grade uranium.

A nuclear armed Google and Amazon. Kind of like the Great Houses in the Dune universe with their "family atomics".

[MachDiamond]

"Whatever happened to the hype of Thorium reactors from a few years ago?"

China is still developing them. The US gave up decades ago and those that want to make them have hit walls getting approvals. No approvals, no investor money. No investors, no business.

The scary thing is China will register a bunch of new patents that will give them de facto control over the technology for nearly 2 decades if they come up with the best implementation. That's not to say that there won't be a way around, but it might be pointless. I remember seeing some shows about steam patents and how some tried to get around them and wound up with horrible compromises.

[John Smith 19]

"No approvals, no investor money."

True.

And the less like a PWR* that reactor looks the more the NRC will struggle to analyse and license it. BTW modern licences are to both build and operate.

*The NRC has bitched for years about reactor development in the US being "Stagnant" ignoring the fact that a hell of a lot of its reg are prescriptive to PWR's and have little or no relevance to other reactor architectures. This also means they have a largish team of staff who know lots about PWR failure modes but fu**all about any other design. TRISO fuelled GCR's are a long way out of their knowledge base.

[MachDiamond]

Re: "No approvals, no investor money."

"This also means they have a largish team of staff who know lots about PWR failure modes but fu**all about any other design. "

In the US, I'd be happy to see the Department of Energy working with companies to test and get new technologies licensed so the NRC has less to say "no" about. It would be taxpayer money well spent as opposed to building more fighter jets that cost $40,000/hour to operate.

[John Smith 19]

Re: "No approvals, no investor money."

Well INL has been doing a lot of work on this for years. It is NRC that sets the ground rules. The best thing to do is start talking to NRC early and keep talking to them.

But that still means they are going to have to demonstrate their new tech.

[Dagg]

Cost!

The SMR are all well and good, but what about the cost! I have not seen any valid numbers from anyone. Maybe someone from El reg can provide a Cost Benefit Analysis for an SMR covering construction, operation and disposal?

The Australian conservative opposition (the coalition) have been pushing these as a low carbon solution. The issue is they can't/won't provide costings.

They also cannot get any private companies to commit to building them so they when they get into power they (a conservative government) will build them.

A couple of things stand out.

1/ If these are that good/great and will make money why aren't the private sector interested?

2/ And why would a conservative government that like small government and privatise everything suddenly what to commit to spending money on public infrastructure.

The only answer is that the SMRs are a non starter because they will never be financially viable, hence no private interest and it means the the coalition can say oops we tried and then keep their promise to the fossil fuel industry to keep burning coal and gas.

[Wexford]

Re: Cost!

You're likely aware of this Dagg, but for the benefit of others the Coalition's nuclear policy is nothing more than to throw uncertainty into the energy market. This in turn allows their fossil fuel sponsors to keep earning from that tasty tasty coal.

[Jellied Eel]

Re: Cost!

Throwing uncertainty into the mix is the job of the shadey 'renewables' lobbyist El Reg keeps citing in most nuclear articles. But then uncertainty is a fundamental feature of wind and solar vs dependable power from reactors. The 'renewables' industry is desperate to keep the billions clueless politicians keep throwing their way, even as it becomes ever more obvious that 'renewables' just can't deliver.

[MachDiamond]

Re: Cost!

"1/ If these are that good/great and will make money why aren't the private sector interested?"

There's private sector interest, but there are some serious regulatory and legal hurdles to get over.

Even breaking even on power costs still means a private and exclusive supply so the money continues to be made with AI/Cloud/Crypto regardless of whether the wind is blowing or government wants to institute a ban on using grid power for a particular application. It can also mean there's excess power to sell on the spot market at times and margin to add another DC/computing center where with grid power, it could be a decade for the local utility to increase capacity at a particular location.

[Dagg]

Re: Cost!

There's private sector interest, but there are some serious regulatory and legal hurdles to get over.

Here in Australia the Coalition has stated that they get rid of these to make it easier for nuclear. But still no private sector interest... Read what you want into that!

[Jellied Eel]

Re: Cost!

Doesn't Australia still have a law banning nuclear? But the AUKUS deal might change that and Australia seems well positioned to benefit. Albeit with a bit of a Catch-22. No nuclear policy probably means a shortage of people trained to work in any nascent nuclear industry.

[Dagg]

Re: Cost!

They have a running functional nuclear reactor at Lucas Heights.

[collinsl]

Re: Cost!

It's New Zealand that has declared itself a nuclear-free zone. To the point where they won't let military vessels in unless they categorically state that they are not carrying nuclear reactors or nuclear weapons. This has frequently caused problems with the US Navy who have a policy of never commenting on whether or not their ships are carrying nuclear weapons (these could include nuclear-tipped torpedoes or bombs don't forget, it's not just limited to nuclear-armed subs and aircraft carriers)

[MachDiamond]

Re: Cost!

"Read what you want into that!"

My best guess on that is regulatory barriers to entry. If the likelihood is 98% of being denied permits to go "hot" in development and operate once all of the front end work is down, investors can find many more places to put their money with a higher chance of a good return.

Energy underpins everything in a modern economy so it's an obvious investment. That doesn't mean that every form of generation is an equally good place to invest. Given coal's reputation, outside of China, it's a very poor place to invest since the chance of new bans and restrictions is very high if bans and restrictions aren't already in place.

[Anonymous Coward]

It’s just a smokescreen for their real plans

Amazon’s secret nuclear weapons program.

[cyberdemon]

Re: It’s just a smokescreen for their real plans

Same-hour delivery on Prime Air Ballistic!

[0laf]

Nuclear apocolypse, MBA style

Small independent nuclear reactors, secretive companies, demanding shareholders and cost cutting beancounters. Match that with a governmental nuclear watchdog that has had it's funding cut to the bone and beyond. What could possible go wrong?

[MachDiamond]

Re: Nuclear apocolypse, MBA style

"Match that with a governmental nuclear watchdog that has had it's funding cut to the bone and beyond."

The outcome of that is very few applications getting processed and easier for them to just say no. No permit, no nuclear power station for you. This isn't a place where a Musk can come in and flout the law and do it any way. They'd have you over a slow fire very quickly. Besides, where would you get the fuel?

[collinsl]

Re: Nuclear apocolypse, MBA style

If Trump wins and Musk gets put in charge of "cutting red tape" or whatever it is that Trump wants him to do then you can bet this regulation will disappear in an instant if it gets in his way.

Also I'm sure that Trump or Musk could do a deal with China or Iran or North Korea to get some enriched uranium for their reactor. I wouldn't be surprised if they try and take it out of a nuclear missile or two (even though that's the wrong type of fuel).

[druck]

Power containerisation

I'd like to see ready to use SMRs shipped in standard 20ft containers. Put couple in the car park, and there's you energy needs sorted. When they need refuelling, a new one would be delivered and the old one taken away to be dealt with off site.

[Dagg]

Re: Power containerisation

Can't they need to be near a large body of water for cooling. Sydney harbour would be perfect. Put them around the harbour at 5 km spacing. Perfect!

[collinsl]

Re: Power containerisation

Much easier to build a hydrogen generating plant then have a hydrogen fuel cell in the 20ft.

[John Smith 19]

"Put couple in the car park, and there's you energy needs sorted. "

Funny you should say that.

The US Army, Navy and Airforce all built prototypes for mobile nuclear reactors in the early 60's. IIRC the spec was more "Moveable on 5t trucks" as ISO containers were not a thing back them. The USAF wanted enough power to run a radar station etc.

The joker in the pack isn't the basic reactor size. It's the shielding or the 400 foot ring around it due to the very limited shielding, meaning either everything is done by remote control or you have a "shadow shield, which is a small sector of the sphere around the reactor you can safely approach on. Outside that you get the full field.

Not to mention anything underground (beneath your car park) can also be cooked from the fission neutrons and gamma emissions which do very bad things to semiconductors, insulators and most organics, like lubricants and plastics, although at shorter range.

[John Smith 19]

Most of these designs are going nowhere.

Here's the thing.

Decades of prescriptive regulation by the NRC mean it knows very little about any other reactor architecture than PWR's. The more a reactor looks less like a PWR the harder it will be to get any kind of license. They couldn't even cope with the Canadian CAND because it has slightly positive reactivity. It's never had a Three Mile Island or Chernobyl incident ever and most of it's water is at room pressure and no more than 70c (giving the operators a lot of time to figure things out), but "Yeah, no." And PWR's are a very poor way to generate steam (they were uncompetitve with coal fired stations of the late 1950's and still produce steam at those conditions).

PWR's are expensive because of the huge amount of concrete and rebar (2x-4x) used compared to a same sized FPP to contain all that steam that flash boils in the event of a major LOCA. In submarines of course this is just dumped in the ocean and flash fries all the fish in the local area. On land this generates a cloud of boric acid (due to the isotopically pure B10 "chemical shim" moderation at $10/gram) with plenty of beta emitting Tritium dissolved in. A containment building big enough to hold all that 177bar water vapour at a reasonable internal pressure (about 10bar IIRC) while it looses all its heat is damm big.

So the real challenge is to design a better reactor using mostly just the bits from a PWR*. That means out of SA 508 steel (maximum inner wall use temperature 350c), preferably not needing an 8 inch thick forged pressure vessel to do so, Uranium dioxide fuel (Melting point ~2500c) and zirconium alloy cladding (Melting point ~1800c). Bonus points for a) (like the CANDU) using natural uranium. b)Generating steam at the pressure and temperature of a modern FPP, c600-630c c)Not using a coolant that needs 2500psi to keep it liquid at the operating temperature of the reactor.

BTW for those who think on-site reprocessing is impossible the first proposal to do so was at the PWR plant in Dresden (Morris, Ill) in 1960. It was expected to need a square patch of ground 35feet on a side (excluding the tanks for high level waste, which PUREX produces a lot of).

*Bit like the scene in "Apollo 13" where they have to build an air scrubber with just the stuff in the capsule.

[bernmeister]

They are missing the point.

The whole point of an SMR is mass production in a purpose built factory. Its all very well having half a dozen reactor designs but without the factory every one is just a hand built prototype. Nobody has yet built a production SMR., they are all lovingly hand built with each part carefully dimensioned and finished by hand as if a luxury motor car was being assembled. I can visualise the logos, "My second SMR is a Rolls-Royce". Dont get me wrong, the concept of mass produced SMRs is a lovely dream but I think it may remain that way for a lot longer than expected. The companies involved dont raise much confidence. Elon Musk companies, Ford and Hitachi, they would be the main runners, not obscure startups.