Google Inks Deal With Nuclear Company As Data Center Power Demand Surges (cnbc.com) 51
Google announced it will purchase power from Kairos Power's small modular reactors (SMRs) to support its clean energy goals and data center demands. The company did not disclose the financial terms of the deal. CNBC reports: There are only three SMRs that are operating in the world, and none in the U.S. The hope is that SMRs are a more cost-effective way to scale up nuclear power. In the past, large, commercial-scale nuclear reactor projects have run over budget and behind schedule, and many hope SMRs won't suffer that same fate. But it is uncharted territory to some extent. Kairos Power, which is backed by the Department of Energy, was founded in 2016. In July, the company began construction on its Hermes Low-Power Demonstration Reactor in Oak Ridge, Tennessee. Rather than use water as the reactor coolant -- as is used in traditional nuclear reactors -- Kairos Power uses molten fluoride salt.
Google said the first reactor will be online by 2030, with more reactors going live through 2035. In total, 500 megawatts will be added to the grid. That's much smaller than commercial reactors -- Unit 4 at Plant Vogtle, which came online this year, is 1.1 gigawatts, for example -- but there's a lot of momentum behind SMRs. Advocates point to lower costs, faster completion times, as well as location flexibility as reasons. Monday's announcement is another example of the growing partnership between tech companies and nuclear power. Data centers need 24/7 reliable power, and right now nuclear is the only source of emissions-free baseload power. Many hyperscalers have ambitious emissions-reduction targets, which is why they're turning to nuclear power.
Google said the first reactor will be online by 2030, with more reactors going live through 2035. In total, 500 megawatts will be added to the grid. That's much smaller than commercial reactors -- Unit 4 at Plant Vogtle, which came online this year, is 1.1 gigawatts, for example -- but there's a lot of momentum behind SMRs. Advocates point to lower costs, faster completion times, as well as location flexibility as reasons. Monday's announcement is another example of the growing partnership between tech companies and nuclear power. Data centers need 24/7 reliable power, and right now nuclear is the only source of emissions-free baseload power. Many hyperscalers have ambitious emissions-reduction targets, which is why they're turning to nuclear power.
Makes little sense, except for... (Score:3)
The strange part is that a normal large power plant would make more sense for data centers. They want a lot of power, they want it cheap, and stable. There's really no need to scale it down.
I'm guessing they're trying to side step punitive regulation put in place by anti-nuclear activists with goal of preventing new nuclear power plants from being built. Which was quite successful across many Western nations.
Re:Makes little sense, except for... (Score:4, Insightful)
Yup!
One of the impacts of regulation is that it changes. Building a large plant takes years and while it is being built regulations change. While the plant is still being built, the plant has to make changes to align with the new regulations. There is a long post on Medium or Substack about this. The article tries to nail down why the US is one of the most expensive places to build nuclear power. Changing regulations is a major cause.
If one could bang out a dozen modular reactors in the time it takes to build a single large one, it is a huge win.
Re: (Score:2)
Why waste time with these little chickenshit reactors, we need gigawatts of nuclear
Their goal is to be able to build reactors quickly to one standard in factories like aircraft, rather than slowly site-built with every one being different, like airports. But 2030 as a starting date? China will be cranking SMRs out long before that. Those won't be approved in he US or EU, but the data centers can easily be built in hungry countries that do approve them.
Obscenely high like literal hookers. (Score:2)
>> China will be cranking SMRs out long before that.
China is not that dumb. They did the math, and seen that the kWh price is obscenely high like literal hookers.
(try to heat yourself with hookers, each one giving you 250W of heat, and calculate the kWh cost. Now that is comparable to SMR.)
China does conventional nuclear power, but the only reason is to get enough plutonium for their exponentially growing weapon production.
The expensive power coming out is just a bonus.
Re: (Score:2)
The actual reality is of course that it's by far the cheapest power per lifecycle. You just need to not have all the punitive regulations making it expensive.
Re: (Score:1)
citation that nuclear in south korea is 1/3 the cost of USA because of less beaurocracy and fear? you can look that up AC with search engine, or get an account and reply and THEN I'll do the work of links.
Nuclear is cheap reliable 24x7 power that smarter countries are tooling up. And a few are getting into thorium and breeders so they have literal millennia of fuel supply.
Re: (Score:2)
Why waste time with these little chickenshit reactors, we need gigawatts of nuclear
Because the big reactors take 20 years to finish and have a track record of massive cost overruns.
The SMRs will be built on an assembly line. You just go online and order what you need, and it shows up a few days later, ready to plug in and operate.
Need a gigawatt? Just order twenty 50-Mw SMRs.
SMR is a scam. (Score:2)
That is a hell of a wet dream you have.
It's a pity the economics don't work.
SMRs are just a means of sucking investor and subvention money, a scam, nothing else.
Re: (Score:1)
In places that are foolish reactors take 20 years and cost 2-3 times what a reactor in smart country does. South Korea can do it in 10 years at 1/3 the cost.
A one or two gigawatt reactor is well worth the wait.
SMR need the periodic refueling and that is disaster of terrorism or accident waiting to happen
Re: (Score:2)
I was wondering why this got modded down and then I realized that it wasn't. The poster just has awful
Makes perfect sense, because corruption. (Score:2)
Yup!
One of the impacts of regulation is that it changes. Building a large plant takes years and while it is being built regulations change. While the plant is still being built, the plant has to make changes to align with the new regulations.
An idiotic process that we have come to not only accept, but expect. Why fix it when it’s horrifically broken, right?
There is a long post on Medium or Substack about this. The article tries to nail down why the US is one of the most expensive places to build nuclear power. Changing regulations is a major cause.
Funny how it’s “changing regulations” in America, when we just call it “fucking corruption” everywhere else. There’s a reason nuclear builds have been met with every form of red tape in America. Because we allow that corrupt shit to happen. Every fucking time.
Re: (Score:2)
Building a large plant takes years and while it is being built regulations change. While the plant is still being built, the plant has to make changes to align with the new regulations.
Having worked on nuclear projects before I have to say, no this is false. Not only do most regulations get frozen for project development at the approval stage, the nuclear industry is infamous with its insanely INSANELY slow update for regulations. There's very little recycle back to engineering phase. Not zero, but not significant.
But then the entire story begs a question.
The article tries to nail down why the US is one of the most expensive places to build nuclear power.
The US isn't one of the most expensive places to build nuclear power. There's this fantasy that everyone else is building them cheaper,
Re: (Score:3)
It's an attempt at reassuring investors of continued growth while also saying they're being real about global warming.
Google are hedging their bets is all. After all, this AI thing is seriously wasteful on electricity. Worse than Bitcoin.
The reality is renewables are going to supersede SMRs for baseload capacity before they can be realised. Renewables will even push hydro offline in the end.
Re: Makes little sense, except for... (Score:1)
Re: (Score:2)
The strange part is that a normal large power plant would make more sense for data centers. They want a lot of power, they want it cheap, and stable. There's really no need to scale it down.
The theory behind SMRs is they can be built on factory production lines rather than custom building them in situ. That promises to reduce the production costs by a lot and improve reliability, enough that it's worth it to give up some economies of operational scale.
Another assertion is risk mitigation is a lot easier with a flock of smaller reactors instead of one big one. The mess you need to clean up is a lot smaller if one reactor fails. Hopefully this makes licensing easier and faster.
Finally, the SMR a
Not Polywell? (Score:2)
Hydro (Score:5, Informative)
nuclear is the only source of emissions-free baseload power
Hydro isn't emission-free baseload power?
Bill Gates, Sam Altman and Jeff Bezos have all backed nuclear companies.
There is money to be made if it works. But a power purchase agreement, assuming that's what this is, doesn't guarantee it will ever produce any power. The larger issue with all these plants is who is responsible if anything goes wrong. With plants attached to the grid the ratepayers are responsible, but I doubt Google or Microsoft are taking on that liability here.
Re:Hydro (Score:4, Informative)
Re: (Score:2)
Tidal power is not baseload.
Power peaks twice a day, dropping to zero in between.
Re: (Score:3)
The larger issue with all these plants is who is responsible if anything goes wrong. With plants attached to the grid the ratepayers are responsible, but I doubt Google or Microsoft are taking on that liability here.
I would love to feel confident enough in the American people as a whole to not tolerate someone merely walking away from a Chernobyl event. But American citizens can’t even fire a senile leader who went Retired On Active Duty (ROAD) a year ago.
I guarantee Google or Microsoft are not taking on that liability. Greed N. Corruption doesn’t follow rules. It merely pays penalties that are always worth it, by design. A mega-corp nuclear reactor, will become Too Big To Fail. Guaranteed. Because ci
Re: (Score:2)
But American citizens can’t even fire a senile leader who went Retired On Active Duty (ROAD) a year ago.
I can't even figure out what you are talking about here. And I have no idea how this got modded up. In 2020, Americans fired a weird, senile, unhinged leader. The current (potentially senile) leader has already been fired and won't be returning to office next year. So he's been fired in 2024. The only question is whether we will bring back the previous senile leader or hand the torch to a younger generation. It's not clear which way that election will go, but American citizens have certainly demonstrat
Re: (Score:2)
But American citizens can’t even fire a senile leader who went Retired On Active Duty (ROAD) a year ago.
I can't even figure out what you are talking about here. And I have no idea how this got modded up. In 2020, Americans fired a weird, senile, unhinged leader. The current (potentially senile) leader has already been fired and won't be returning to office next year. So he's been fired in 2024.
Really? Here’s a question for all those needing reminding of who’s carrying around a fucking nuclear football still. If we fired that (definitely) senile guy already, then answer one question there genius. Who the hell do YOU think the President of the United States is right now?
Like I said. Retired On Active Duty. And you’re too infected too see the danger in that.
The only question is whether we will bring back the previous senile leader or hand the torch to a younger generation. It's not clear which way that election will go, but American citizens have certainly demonstrated that they can fire senile leaders.
It's a shame that the first part of your post contains irrelevant nonsense because the second part makes valid points but they are buried within the noise.
No. it’s a fucking shame you don’t see the danger of your TDS infection. Makes you assume you can STILL bla
Re: (Score:2)
It's a good deal for Google. Someone else already paid to build and operate the plant through the expensive period. Now it's just subsidised energy from an operator that is probably finding the market increasingly difficult at renewables ramp up. All the risk is externalized too.
Re: (Score:2)
Hydro (and geothermal) are emission-free baseload power.
Both of them have geological requirements that make them hard to scale up much more than they are currently being used:
Hydro means damming a river. Good luck getting approval, considering we're celebrating the removal of dams in California [ca.gov].
Geothermal means drilling for access to hot strata. I don't know much about it, but if it were cheap, easy, and available in lots of places, we would have more of them already, right?
Re: (Score:2)
Hydro (and geothermal) are emission-free baseload power.
Geothermal is not emissions-free. Radioactive elements come out of those vents. When the vents are enlarged and production eventually wanes then they add water (usually primary treated sewage water) to the system in order to restore it. This causes more emissions as more matter is flushed loose. Most of them escape into the atmosphere, but some collect on the turbine blades. At The Geysers geothermal field near Calistoga, CA, the world's largest operating geothermal plant, what was washed off of the blades
Re: (Score:1)
nuclear is the only source of emissions-free baseload power
Hydro isn't emission-free baseload power?
No, neither nuclear nor hydro is emissions-free. The only measurement that matters when talking AGW is cradle to grave, and both nuclear and hydro have higher emissions per Wh than solar or wind. When you flood an area for a reservoir you create anaerobic conditions which cause CO2 and methane release, and the building of the dam including the CO2 emissions from the concrete (and of making it in the first place) is a CO2 emitting process.
Hydro actually has ongoing emissions due to the decomposition process
Re: (Score:2)
Hydro isn't emission-free baseload power?
It largely emission free is once operating. Construction might be another issue. How much CO2 was emitted making the concrete for the Three Gorges dam? Hydro, however, comes with it's own unique set of environmental costs. And we seem to have dammed all the rivers we're willing to dam so it's difficult to envision scaling it up by much. Damn.
Re: (Score:2)
Re:"...the first reactor will be online by 2030" (Score:4, Interesting)
To get any sort of nuclear plant approved in the US is going to take a LOT longer than 5ish years. For the US SMR's are a new tech, so even less likely for speedy regulatory approval.
It's nice that they have more location options than traditional nuclear. So the option of placing these things in places people won't get overly offended may help the process.
The article does not state that they intend to deploy in the US. It actually reads as if they were being very careful about not mentioning this. So Google may be aiming to put SMR's in a more friendly country. There is talk about various US agencies and existing nuclear but nothing about deploying SMR's in the US. I suspect this is a test article to gauge reaction. Allowing Google to "clarify" certain points in such a way that they can spin it in a more positive light.
My personal view on this is Google will deploy in a more friendly country first. Use that as proof that they and the operator can delivery and use them efficiently and safely. In order to accelerate the US regulatory process. Also since SMR's are not American they don't need US approval to operate them outside the US. An American style reactor would require US approval to "export" nuclear technology. This last point is debatable however. I'm sure the US nuclear regulatory authority would have comment.
Besides 2030 for a US deployment even stretches Elon time hopefulness. Also costs associated with SMR's are not well understood. The long term costs are definitely not known.
SMR's are not the solution to all things bad about Nuclear. SMR's are their own waste storage systems. Meaning that where they are built is forever a nuclear waste site. That must be maintained and protected. The only thing that SMR's bring to the table is big changes to the cooling stack. So a dramatically reduced requirement for access to vast amounts of clean water. Which has been a weak point for current reactors types.
SMR's also require a permanent connection to a stable power grid. They can not operate isolated from the grid. The grid provides the emergency power to power the pumps for cooling in the advent of some incident in the reactor. Example if an SMR scrams and it's not connected to the grid it has a high chance of melting down without an active powered cooling system. SMR's are not nuclear power plants in a box.
Also Note Hydro electric provides baseload. So does geo-thermal and Solar tower. It can be argued that renewables + storage also provide the same.
Again it feels like someone is testing the water here to gain public reaction. Favourable and Google puts out a hand for federal money to help "mature" the tech. Unfavourable deploy in another country and use experience there to drive US acceptance.
Re: (Score:2)
>> The article does not state that they intend to deploy in the US
But other articles do.
"In a deal that marks the first corporate agreement to deploy multiple small modular reactors (SMRs) in the U.S"
https://www.powermag.com/googl... [powermag.com]
Re:"...the first reactor will be online by 2030", (Score:2)
To get any sort of nuclear plant approved in the US is going to take a LOT longer than 5ish years. For the US SMR's are a new tech, so even less likely for speedy regulatory approval.
It's nice that they have more location options than traditional nuclear. So the option of placing these things in places people won't get overly offended may help the process.
That you for clarifying exactly why building nuclear reactors in America, has become a corrupt cesspool of endless handouts and payouts to the NIMBYs, and has NOTHING to do with “safety”.
Maybe America won’t become too “overly offended” when it becomes a technical shithole by comparison, because being offended defines “the process” to build the future. If America wants it bad enough, it’ll find a way to tell NIMBY to shut the FUCK up.
Re: (Score:2)
NuScale received permission to build a prototype site [nuscalepower.com], but they have terminated the project to build one [apnews.com] despite more than $600 million in DOE grants since 2014 as it appeared "unlikely that the project [would have enough subscription to continue toward deployment" — nuclear booster speak for "it wasn't cost effective" but which can be translated to "this thing won't actually work and it's all just a big cash grab scam".
Re: (Score:3)
https://www.gevernova.com/nucl... [gevernova.com]
Site preparation has already begun, a process simplified by the fact they are being built on an already a licenced nuclear site.
https://www.opg.com/projects-s... [opg.com]
I'm hopeful for the success of this project. OPG's ongoing refurbishment of their existing nuclear fleet has been quite well managed time and budget wise so far.
Bad economics. (Score:2)
It's in bad shape.
The costs doubled lately to 105 USD/MWh
https://www.woodmac.com/blogs/... [woodmac.com]
And that is with subsidies. Without subsidies, and counting all the "externalities" like waste management, it will triple again (6x more than the original promise)
Re: (Score:2)
Re: (Score:2)
each and every country having nuclear power mis-managed waste management, exactly like the USA.
Re: (Score:2)
As problems go it is tiny compared to climate, but life is good here regardless so I'll be fine sitting around here watching you fail as you cherry pick your preferred solutions.
Re: (Score:2)
That is B.S.
Yes, it is not a huge investment now. But you have to care for the stuff for 50 000 Years at best.
Yes, that costs, and it has been mismanaged everywhere.
going nuclear (Score:2)
To slightly more accurately predict the next most likely word in a sentence.
Makes total sense.
Good luck (Score:3)
Good luck to Kairos but they don't even have a pilot plant up and running. The NRC gave them a permit to build a 35-MWth “non-power” demonstration unit last year, and it is "anticipated" in 2027. And sure, if they ever do produce actual power, Google will probably buy some.
https://www.powermag.com/googl... [powermag.com]
Re: (Score:1)
Looking at their website, it's an old concept that they hope will magically be cheaper and better by being smaller: https://kairospower.com/techno... [kairospower.com]
It's using molten salt, which is dangerous and corrosive stuff.
I'd be amazed if they ever got anything working, even a prototype. Investors look at it, a little demonstration non-power device in 3 years time, then some unknown period before they can scale up to something that actually makes money. So much risk of issues with the technology, no clear benefits, a
Re: (Score:2)
Looking at their website, it's an old concept that they hope will magically be cheaper and better by being smaller
Unfortunately what the SMR companies are finding is that being smaller actually makes them significantly more expensive. The industry's economies of scale isn't defined by how many units are produced, but how much power you get per reactor. Even with the best case estimates for SMRs they will never have economies of scale that will drive down construction points, these things will never roll off a factory line en-mass with any foreseeable estimate of current or impeding technological developments.
It leaves
Re: (Score:2)
I almost wonder if it's some kind of scam, because any engineer should have been able to spot this pretty quickly.
Receiving +500 million, results in 2035 or so (Score:2)
Sounds like a trustworthy investment just to piss off Bill Gates.
It's a race! (Score:2)
Hand in the back of the classroom goes up excitedly. "I know, I know! The AI will hallucinate a reactor design that will solve all the problems at the same time, and then do my homework for me and get me a cool job at Google powered by the reactor!!!"
We're completely fucked.
Why two heat exchanger design? (Score:2)
Getting geeky for a bit, the Kairos design [kairospower.com] uses a two stage heat exchanger. Any speculation why? Seems that would add complexity and cost. What is the reason to not create steam directly from the salt coming from the reactor?
I'm also curious how one shuts down and restarts a molten salt system. How do you melt the salts and get them flowing through the pipes? How does one preheat the plumbing so the salts don't freeze when you introduce them? And how do you go about shutting the reactor down, letting it coo