Google adopts small nuclear power reactors at unprecedented scale — inks deal for seven reactors to feed AI data centers
Google adopts small nuclear power reactors at unprecedented scale — inks deal for seven reactors to feed AI data centers
Nuclear power startup Kairos Power is its partner.
You're viewing a single thread.
Crazy how quickly we've gone from "Nuclear is a dead technology, it can't work and its simply too expensive to build more of. Y'all have to use fossil fuels instead" to "We're building nuclear plants as quickly as our contractors can draft them, but only for doing experiments in high end algorithmic brute-forcing".
Would be nice if some of that dirt-cheap, low-emission, industrial capacity electricity was available for the rest of us.
One of the things with AI is that it's a largely constant load factor. Nuclear is really good for that.
However, I highly doubt any of these new nuclear plants are finished before the AI bubble bursts. SMRs haven't even been proven in practice yet, and this is the first good news they've had in a while. Restarting Three Mile Island isn't expected to work before 2028. The hype bubble could easily burst in the next year, and even if it doesn't, keeping it going to 2028 is highly unlikely.
So we'll probably have some new nuclear around that isn't going into AI, because those datacenters will be dead when the hype passes. Might as well use them, I guess.
However, I highly doubt any of these new nuclear plants are finished before the AI bubble bursts.
Given the military applications of the technology, I don't think it is ever really going away. Consumer AI (those user facing image generators and chatbots, for instance) might lose funding. But Israel's Lavender AI is going to become a permanent fixture in our lives, as it's rolled out for the policing of more and more territory.
You don't think there's any chance that AI as it exists today might be just the start of a huge industry?
As it exists now, no. The models are reaching their limit, and they aren't good enough. They can't absorb any more information than they have, and more training iterations aren't making them better. They'll do some useful things; a recent find of the longest black hole jet ever found was done in part from AI classification of astronomy data. It's going to get implemented into existing tools and that's about it. It won't be enough to justify the money that's already been dumped in.
Historically, the field has been very bursty. Lots of money gets dumped into it, it makes some big improvements, and then hits a wall. Funding dries up because it's not meeting goals anymore, and the whole thing goes into slumber for a decade or two. A new breakthrough eventually comes, and then money gets dumped in again. We've about maxed out what the last breakthrough can give us. I expect we'll need at least one more cycle of this before AGI works out.
I still think it's too expensive, and this contract doesn't change my position. Google is committing to buying power from reactors, at certain prices, as those reactors are built.
Great, having a customer lined up makes it a lot easier to secure financing for a project. This is basically where NuScale failed last year in Idaho, being unable to line up customers who could agree to pay a sufficiently high price to be worth the development risk (even with government subsidies from the Department of Energy).
But now Google has committed and said "if you get it working, we'll buy power from you." That isn't itself a strong endorsement that the project itself will be successful, or come in under budget. The risk/uncertainty is still there.
Well, once the AI hype calms down and people realize the current approach won't lead to actual intelligence or "The Singularity", there may be quite some nuclear plants left over. That or they will be used to mine shitcoins.
-
Tax them enough that they don't have the cash to just up and build their own personal-use nuclear powered, nation spanning infrastructure.
-
Use those taxes to build a nation spanning nuclear infrastructure that everyone can use.
I've got so many ads so far for how adding new taxes is bad even if it pays for good things, and all of the issues they are arguing about aren't even adding any taxes. Actually adding taxes seems like a great way to make political enemies, even though it's often the best tool there is for a thing.
Eh, I would say investment into R&D should be encouraged and maybe allow tax write offs. Even of the end goal is a private power source. Once that R&D turns into workable, operable, sellable products, then tax the fuck out of them. Perhaps disallow making things that can be a boon to public infrastructure from being deem proprietary, so that it can be more easily adapted to public use.
I dunno, I'm typing from my couch after a few beers.
-
Fun Times! Because everyone pays for the waste and when something goes wrong. Privatizing Profits while Socializing Losses. The core motor of capitalism.
The cleanup for fossil fuels is an order of magnitude more expensive, and an order of magnitude more difficult. It also impacts so many things that its true cost is impossible to calculate.
I'm aware of the issues with nuclear, but for a lot of places it's the only low/zero emission tech we can do until we have a serious improvement in batteries.
Very few countries can have a large stable base load of renewable energy. Not every country has the geography for dams (which have their own massive ecological and environmental impacts) or geothermal energy.
Seriously, we need to cut emissions now. So what's the option that anti-nuclear people want? Continue to use fossil fuels and hope battery tech gets good enough, then expand renewables? That will take decades. Probably 30+ years at the minimum.
Nuclear should only be done by the state. Any commercial company doing nuclear HAS TO CARE FOR THE WASTE. It has to be in the calculation, but no on ecan guarantee 10000 years of anything. Same with fossils... execute the fossil fuel industry. They destroyed so much, they don't deserve to earn a single cent.
That funky startup is producing waste. Imagine a startup selling Asbestos as the new hot shit in 2024.
We’re talking 11 years for 7 “small” reactors. The first decade just to establish a business, but no real difference in the overall picture. How many years, decades after that to make a noticeable difference?
Meanwhile we’re building out more power generation in renewables every year. Renewables are already well developed, can be deployed quickly, and are already scaling up, renewables make a difference NOW.
Renewables cannot provide a reliable base load. Not unless you can have your solar panels in space where the sun always shines, we figure out tidal power, or you're lucky in terms of geography and either hydroelectric or geothermal work for you.
Solar power doesn't produce energy at night, wind doesn't always blow. You know the drill.
You completely sidestepped the entire crux of my comment.
We need a base load of energy to fill that gap, because batteries currently can't, and likely won't be for decades. Here are the options we have available:
-
nuclear power, which produces a waste that while trivial to store far away from people, will be radioactive for hundreds of years.
-
fossil fuels, which cause massive damage not only to the local environment, but to the planet, and cleanup is effectively impossible.
-
we put society on unpredictable energy curfews. At night the population can't use much energy. When there's a drop in wind or solar production, we cut people's energy off. Both political parties must commit wholeheartedly to this in order to make it viable. Our lives would become worse, but we'd not have either of the above problems.
Of those 3 options, I'd rather go with nuclear. What's your choice?
Don't forget to add that nuclear waste created is absurdly small in volume.
More renewables.
We’re at the beginnings of having useful levels of storage and can keep building out renewables while we develop storage. At the current rates of adoption, we’ll need true grid storage in about ten years.
However, note that one option for “grid” storage is a battery in every home. Another is a battery in every vehicle. Neither is the best option but those are options we already know and just need to scale up
Ok, you've added more solar panels and wind turbines.
It's nighttime. There isn't much wind. An extremely common thing to happen I'm sure you'll agree.
There now isn't enough power, places have constant blackouts, electricity prices skyrocket because demand far outstrips supply.
Grid storage large enough to replace fossil fuels + nuclear is far, far, far, far, far, far further than 10 years off.
I'll ask again:
-
Nuclear base load that assists renewables
-
Continued fossil fuels for multiple decades that assists renewables, and hope that we can reverse some of the damage done in the meantime through some kind of carbon capture tech
-
regular blackouts, energy rationing, but 100% renewable
What do you choose? Saying that you'll magic up some batteries in a capacity that currently isn't possible isn't an answer.
I want 100% renewables too, but it's currently not feasible. Our choice is between having a fossil fuel base load or a nuclear base load. Other options aren't available yet.
And here’s the magic choice …… “time of use metering”. As we electrify everything and add “smart” controls, we can be much more dynamic with time of use metering to adjust the load.
When the sun doesn’t shine at night, already has much lower electrical load than daytime. Early analog efforts at time of use metering tried to shift more load to the night so “base load” wouldn’t have to adjust, and max load wouldn’t be as high
Now we can develop smart time of use metering to shift more load to “when the sun shines”. I’m not aware of anything to quantify this so let me just make shit up: if the load “when the sun doesn’t shine” is half what it is when solar is producing, that’s a crap load of grid storage or base load that magically never has to exist
-
-
Right but how about actually addressing the question?
What about base load then. It's all well and good building shit tons of solar panels and wind farms but sometimes you need energy and the sun isn't shining and it isn't windy. What do you do then?
That's why we need base load and I'd rather the base load came from nuclear than from fossil fuels, as I'm sure you would too, but you seem to be anti-nuclear as well, so what do you want?
I'm so sick of you eco warrior types with absolutely no understanding of the problem. It's not as if the internet doesn't exist it's not as if you couldn't educate yourself if you wanted to. People are out here trying to educate you all about it, and you cope by ignoring them.
Base load? Oh you mean the kind of power only the industry needs but wouldn't be able to pay for if it wouldn't be shifted towards the public? Don't try to fool people by just not talking about this little fact.
Solar and small scale power buffering could easily be decentralized for the publics overall power need, including charging and utilizing cars as buffers. A private person isn't "the base load"... but we all pay for "the base load".
Base load err... educate yourself nuclear boy.
Apart from that: Your arguments didn't change, they are still wrong, that's why "we" stopped listening. You reproduce Industry talking points without checking. (e.g. "bAsE loAD") like an angry little LLM.
Who needs the power needs to pay for it. Including the waste. I don't see why I should clean up Google's micro nuclear waste.
This reply is both unintelligible, and unhinged. You also seem to be berating someone for not knowing what baseload is, while simultaneously showing (I think, it’s hard to tell honestly) that you have no idea what it means.
I don't berate. He is right, but again I don't see how the containment of nuclear waste, Google is producing for LLM training for their profits, should be a public concern. Even on a global scale, "base load" is the continuous need of power .... so mostly industries. You don't need Nuclear Power Plants to run street lights and Hospitals, you need them to run steel mills and manufacturing plants.
My point is exactly: Why should the industrial need for reliable power be priced on our bill without a fair share on the profits for society? And this isn't even touching the impossibility of putting a price tag on something that has to be stored for 1000ns of years.
Unhinged? I just replied in the same tone. He didn't even reply to any of my points. Come clear, what's your point?
You need to be on pills.
Base load is the amount of charge that you need in the system to deal with just basic usage. This includes powering your computer so you can post incoherent rents on the internet. Something I assume you think is very important.
Without base load when it's night and not windy all the power goes out, I assume you would think that was inconvenient even though you are not a mega corporation.
Now rather than trying to deflect answer the question how do we supply fundamental power when the sources are renewable are not operating and don't say we can store it in batteries because we can't not at that capacity.
That's not completely wrong but in parts. I can easily buffer solar energy to cover 80% of my energy needs. You have to understand that most of the base load isn't "our" power consumption. It's mostly commercial.
And again. Google training LLMs is not Base-Load and nether deflection. It's the subject of the Article!
You are totally ignoring their arguments. Not every place can do wind or solar or hydro. Like it’s simply not an option.
Time doesn’t care. Neither does the rate of climate change.
Everyone pays for not using nuclear too, a thousand fold more so.
It's almost like the brand spanking new tech to make small nuclear reactors are extremely cost prohibitive and risky, and to lower the cost someone needs to spend money to increase supply.
If only that was the government that invested in the R&D and tech to make it happen.
Gaining funds from taxes (meaningful taxes), and investing that money in making their country better.Hopefully this decision is because carbon taxes that will make consumer products representative of the actual cost of the item (not the exploitative cost). >
No no, let the free market decide.
Fucking AI threatening to replace basic jobs (when it's more suited to replace the C-Suite) gobling up energy and money, too-big-to-fail bailouts and loophole tax rules bullshit.So yeh, someone needs to spend the money and that should be the government.
Because they should realise that carbon fuel sources are a death sentence.I'm glad you don't make the decisions because I don't want my taxes, that I work hard for and pay money into, to be spent by the government on highly-likely dogshit experimental brand new nuke tech that may eventually cost more money later on to maintain, and I prefer they spend it renovating existing infrastructure or building tried/true legacy nuke plant designs.
Your taxes already go towards this.
That's how governments leverage capitalism to placate the people. Grants for green energy initiatives.
Private companies get free money for taking some amount of risk because they are likely to profit massively from it.
https://www.canarymedia.com/articles/nuclear/google-agrees-to-multi-reactor-power-deal-with-nuclear-startup-kairos
Kairos is getting free money (grants & tax breaks) and profits from this. Google is extremely likely (can't find a source) to be getting free money for thisCompanies EXIST to extract profit.
Of one of the worlds most successful companies is doing this, it's because "line goes up".I'd prefer this happend so that "humans survive".
But "humans don't die faster" is fine for now.(I guess "humans" means "poor humans". As in anyone that doesn't outright own 2 homes.)
The day I lost all respect for my father was when he told me in all seriousness that the fundamental purpose of capitalism was to make people happy.
At first I wanted to reply about how one can HOPE that capitalism’s never ending extraction of value from labor might build a better future and enable more happiness.
But there’s a deeper assumption in that statement, and in my limited personal experience it affects conservatives the most. That is thinking that happiness is caused by external factors: money, toys, status, power, etc.
His thinking was that you can never sell anything unless you have something they want, so the fundamental idea of capitalism was that it tries to give people what they want.
To be fair here, no one's certain this will be cost-effective either. The new techs make it worth trying though.
no one’s certain this will be cost-effective either
One of the great sins of nuclear energy programs implemented during the 50s, 60s, and 70s was that it was too cost effective. Very difficult to turn a profit on electricity when you're practically giving it away. Nuclear energy functions great as a kind-of loss-leader, a spur to your economy in the form of ultra-low-cost utilities that can incentivize high-energy consumption activities (like steel manufacturing and bulk shipping and commercial grade city-wide climate control). But its miserable as a profit center, because you can't easily regulate the rate of power generation to gouge the market during periods of relatively high demand. Nuclear has enormous up-front costs and a long payoff window. It can take over a decade to break even on operation, assuming you're operating at market rates.
By contrast, natural gas generators are perfect for profit-maximzing. Turning the electric generation on or off is not much more difficult than operating a gas stove. You can form a cartel with your friends, then wait for electric price-demand to peak, and command thousands of dollars a MWh to fill the sudden acute need for electricity. Natural gas plants can pay for themselves in a matter of months, under ideal conditions.
So I wouldn't say the problem is that we don't know their cost-efficiency. I'd say the problem is that we do know. And for consumer electricity, nuclear doesn't make investment sense. But for internally consumed electricity on the scale of industrial data centers, it is exactly what a profit-motivated power consumer wants.
One of the great sins of nuclear energy programs implemented during the 50s, 60s, and 70s was that it was too cost effective.
I don't see how any of this has any bearing on financial feasibility of power plants.
For what it's worth, before the late 90's there was no such thing as market pricing for electricity, as prices were set by tariff, approved by the Federal Energy Regulatory Commission. FERC opened the door to market pricing with its Order 888 (hugely controversial, heavily litigated). And there were growing pains there: California experienced rolling blackouts, Enron was able to hide immense accounting fraud, etc. By the end of the 2000's decade, pretty much every major generator and distributor in the market managed to offload the risk of price volatility on willing speculators, by negotiating long term power purchase agreements that actually stabilize long term prices regardless of short term fluctuations on the spot markets.
So now nuclear needs to survive in an environment that actually isn't functionally all that different from the 1960's: they need to project costs to see if they can turn a profit on the electricity market, even while paying interest on loans for their immense up front costs, through guaranteed pricing. It's just that they have to persuade buyers to pay those guaranteed prices, rather than persuading FERC to approve the tariff.
As a matter of business model, it's the same result, just through a different path. A nuclear plant can't get financing without a path to profit, and that path to profit needs to come from long term commitments.
It can take over a decade to break even on operation, assuming you're operating at market rates.
Shit, it can take over a decade to start operations, and several decades after that to break even. Vogtle reactors 3 and 4 in Georgia took something like 20 years between planning and actual operational status.
Now maybe small modular reactors will be faster and cheaper to build. But in this particular case, this is cutting edge technology that will probably have some hurdles to clear, both anticipated and unanticipated. Molten fluoride salt cooling and pebble bed design are exciting because of the novelty, but that swings both ways.
I don’t see how any of this has any bearing on financial feasibility of power plants
If you don't get a high ROI, you're not going to have lots of investors offering up their cash at low interest rates.
Plus time. My perspective was that building a new nuclear power industry and any significant number of reactors would take too long: we need to have fixed climate change in less time.
So seven “small” reactors over the next eleven years ….. faster than I expected but still takes decades to make a noticeable difference.
So seven “small” reactors over the next eleven years ……
Is more than we've built in the last 40. And, assuming energy demands continue to accelerate, I doubt they'll be the last seven reactors these companies construct.
It's insane that Google is even allowed to build a nuclear reactor
I don’t think they’re even building many. The article uses the word “adopt” because they’re kinda reviving old power plants. Three Mile Island being one of them.
I don’t think they’re even building many.
You might want to read the article.