Oracle Is Designing a Data Center That Would Be Powered By Three Small Nuclear Reactors 96
With electricity demand from AI becoming so "crazy," Oracle's Larry Ellison announced the company is designing a data center that will be powered by three small nuclear reactors capable of providing more than a gigawatt of electricity. "The location and the power place we've located, they've already got building permits for three nuclear reactors," Ellison said. "These are the small modular nuclear reactors to power the data center. This is how crazy it's getting. This is what's going on." CNBC reports: Small modular nuclear reactors are new designs that promise to speed the deployment of reliable, carbon-free energy as power demand rises from data centers, manufacturing and the broader electrification of the economy. Generally, these reactors are 300 megawatts or less, about a third the size of the typical reactor in the current U.S. fleet. They would be prefabricated in several pieces and then assembled on the site, reducing the capital costs that stymie larger plants.
Right now, small modular reactors are a technology of the future, with executives in the nuclear industry generally agreeing that they won't be commercialized in the U.S. until the 2030s. There are currently three operational small modular reactors in the world, according to the Nuclear Energy Agency. Two are in China and Russia, the central geopolitical adversaries of the U.S. A test reactor is also operational in Japan.
Right now, small modular reactors are a technology of the future, with executives in the nuclear industry generally agreeing that they won't be commercialized in the U.S. until the 2030s. There are currently three operational small modular reactors in the world, according to the Nuclear Energy Agency. Two are in China and Russia, the central geopolitical adversaries of the U.S. A test reactor is also operational in Japan.
What are we even doing? (Score:1)
Re: (Score:2)
Funny, that's what we were saying about cryptocurrency.
Re: (Score:2)
AI, Capto, NFTs, Quantum Computing, flying cars, etc., take your pick. Always some hallucination about to change the world with the clueless.
Re: (Score:3)
They're changing the world, all right. They're making it unlivable sooner.
Glow In The Dark pr0n (Score:2)
>> glow in the dark while watching your pr0n
Correction: AI generated "normal pr0n", or pr0n made by people who glow in the dark, two new categories.
There will be a shortage of normal pr0n once we are all contaminated and glowing in the dark.
Re: (Score:2)
Re: (Score:3)
Re: What are we even doing? (Score:2)
TFS said it is for AI.
But un general AI workload need a much higher power density than pretty much any other workload we have seen. This is causing many places to redesign their datacenter to take higher power in and heat out.
New generations of GPU are essentially liquid cooled only forcing redesign of cooling infrastructures. And this is only happening because of how flop intensive machine learning models are compared to previous mostly IO driven web workloads.
Re: (Score:2)
What Oracle does. (Score:3)
There's no way 'AI' can be worth this.
Its absolutely worth it, when ‘AI’ can somehow sidestep decades worth of NIMBY red tape and political handouts.
Nuclear generators, create power. Which can be sold to anyone who requires it for any reason. If ‘AI’ is used for anything with Oracle, it’ll be calculating the litigation costs for a consumer breaking their Oracle power license for using the neutral wire during unlicensed amp-hours or some shit.
Re: (Score:2)
You can't get around NIMBY by attaching your nuclear power station to a data center. That just means you have to build them both somewhere where you can build a nuclear power station. The advantage of a building a large data center is that you have a guaranteed 24x7 market for power you generate, even during off-peak hours.
Re: (Score:2)
You can't get around NIMBY by attaching your nuclear power station to a data center. That just means you have to build them both somewhere where you can build a nuclear power station. The advantage of a building a large data center is that you have a guaranteed 24x7 market for power you generate, even during off-peak hours.
No. The advantage of even being able to build a nuclear reactor in under two decades, is exactly what I said it was. If it were easy to build the damn things, we’d have plenty of new ones to shut down the old ones just waiting to create the next disaster. Hard to even dismiss NIMBY when the Government owns/preserves most of that land where humans aren’t.
And the advantage of building a nuclear powered data center, is whatever bullshit they’re selling you. You, I, or anyone else don
Re: (Score:2)
Apologies about the Amazon reference. For a moment there I got my corrupt billionaires lying about their ultimate plans and goals mixed up. As if there’s any real difference between an Amazon nuclear reactor or an Oracle one. Both are likely coming.
Re: (Score:2)
Re: What are we even doing? (Score:2)
Dude have you heard this
https://youtu.be/XEa_kj2btS4?s... [youtu.be]
Or
https://youtu.be/fY1EIArdsto?s... [youtu.be]
Totally worth hastening our climate demise
Re: What are we even doing? (Score:2)
If you believe in modern reactor designs making nuclear power inherently safe and the only scalable carbon-neutral energy choice - there is an argument that accelerating commercialization of modular reactors is worth the AI bubble.
Re: (Score:2)
"Small" modular nuclear reactors is just a buzzword and "hot" new trend in the nuclear industry. There's nothing about them that makes them significantly "safer" than a 1970's nuclear plant design. Instead of concentrating nuclear production to a small area that will suffer pools of nuclear rod waste, you've spread the radioactive material problem across every nook & cranny in the country. The real reason why "small" modular nuke plants came about was so that instead of one mega nuke plant, we'd have
Re: (Score:2)
Instead of concentrating nuclear production to a small area that will suffer pools of nuclear rod waste, you've spread the radioactive material problem across every nook & cranny in the country.
No, you haven't, but only because their claims are bullshit. They claim you'll be able to do that, but you're not getting approval to build any of their shit near any people. Consequently if they ever actually build any (they have built 0 so far, not even a prototype) they will just wind up installed in groups on the same kind of sites bigger reactors would have been installed in. This offers the benefit that some can be serviced while others run, but it also multiplies the number of points of failure which
Re: (Score:2)
By then, Larry Ellison will be in control of SkyNet & the president... permanently. If only Jeff Bezos had thought of this first, muahahaha! https://www.youtube.com/watch?... [youtube.com]
Larry Ellison (Score:5, Funny)
Re: (Score:2)
Re: (Score:3)
Re: (Score:2)
Re: (Score:2)
Re: (Score:1)
Crowd size!
Re: (Score:2)
And how do they plan to get these SMRs? (Score:1, Troll)
Because SMRs have been a hallucination by the nuclear fanatics to keep their delusion alive, nothing else.
Re: (Score:1, Funny)
Just the sheer amount of deaths per TWh make nuclear something that nobody can consider. All one has to do is bring that number up, and nuclear is always taken off the table, even if it means going back to burning natural gas, biowaste, and peat.
Re: (Score:2)
Just the sheer amount of deaths per TWh make nuclear something that nobody can consider. All one has to do is bring that number up, and nuclear is always taken off the table, even if it means going back to burning natural gas, biowaste, and peat.
Sorry, I'm confused. Are you saying deaths per TWh for nuclear are worse than the alternatives? I'm not aware of anyone dying due to an incident at a US commercial nuclear reactor, not even TMI. You can't say that about, say, coal.
Re:And how do they plan to get these SMRs? (Score:5, Interesting)
There's possibly deaths indirectly associated with the 1959 Santa Susana Lab meltdown (from cancer). To your point, this was an experimental reactor (not a commercial reactor). Unfortunately there's no way to corroborate or verify the potential impact of the meltdown, since the actual meltdown was kept secret at the time, I'm assuming the tally of possible deaths attributable to the accident was derived statistically.
https://www.nbcnews.com/id/wbn... [nbcnews.com]
"The report by an independent advisory panel estimated it was likely that radiation released during the meltdown at the Santa Susana Field Laboratory caused about 260 cases of cancer within a 60-square-mile area around the reactor.
The labâ(TM)s former owner, Rocketdyne, has said for years that no significant radiation was released. But the independent advisory panel said the incident released nearly 459 times more radiation than a similar one at Pennsylvaniaâ(TM)s Three Mile Island in 1979."
https://www.engineering.com/am... [engineering.com]
"On Monday July 13, 1959, an experimental nuclear reactor in Area IV was showing skyrocketing temperatures that would indicate a runaway reaction. Workers plunged control rods into the core to stop the reaction. It didnâ(TM)t work, so they shut it down manually. They released radioactive gases into the air to relieve pressure and prevent an explosion. Inexplicably, they turned the reactor back on. They had no idea a partial meltdown was occurring.
The radioactive gases they released could have been as much as 13,000 curies of iodine-131 and 2600 curies of cesium-1371. By contrast, Three Mile Island released 17 curies of iodine-131 and no cesium.
The reactor was off and on for the next two weeks, then repaired in 1960. It continued to operate until it was shut down permanently in 1964."
"Nuclear reactors normally are cooled by water. Modern nuclear reactors are enclosed in thick concrete domes to contain radioactivity in the event of a runaway reaction. The SRE had no concrete containment and was using molten sodium, not water, to cool the reactor cores. "
And yes, the total impact of radioactivity emitted from coal burning plants is greater on the total population.
https://www.scientificamerican... [scientificamerican.com]
"In a 1978 paper for Science, J. P. McBride at Oak Ridge National Laboratory (ORNL) and his colleagues looked at the uranium and thorium content of fly ash from coal-fired power plants in Tennessee and Alabama. To answer the question of just how harmful leaching could be, the scientists estimated radiation exposure around the coal plants and compared it with exposure levels around boiling-water reactor and pressurized-water nuclear power plants.
The result: estimated radiation doses ingested by people living near the coal plants were equal to or higher than doses for people living around the nuclear facilities. At one extreme, the scientists estimated fly ash radiation in individuals' bones at around 18 millirems (thousandths of a rem, a unit for measuring doses of ionizing radiation) a year. Doses for the two nuclear plants, by contrast, ranged from between three and six millirems for the same period. And when all food was grown in the area, radiation doses were 50 to 200 percent higher around the coal plants."
Re: (Score:2)
We have actual deaths with research reactors, including some pretty gruesome ones - imagine being impaled by a control rod you were improperly removing, for example. Research wise we also have Madam Curie, the demon core scientists, etc...
But that isn't power. We have deaths even from solar and wind power. Natural gas kills workers semi-regularly. Coal kills people on the regular, both from pollution and accident.
Consider though, that you have to go back to the 1950s to get fatalities.
As opposed to, say
Re: (Score:2)
Thank you for the informative reply. I'd never heard of this incident.
"In a 1978 paper for Science, J. P. McBride at Oak Ridge National Laboratory (ORNL) and his colleagues looked at the uranium and thorium content of fly ash..."
LIght bulb: we mine and burn coal, then extract uranium and thorium from the fly ash to run nukes, completely pissing off every solar/wind energy promoter in the process. Win win win!
Re: (Score:2)
Re: (Score:2, Funny)
For what it's worth, statistics show Nuclear has the second lowest death rate. It's behind only solar.
You are only counting human deaths, right?
I hear that solar installations are the biggest non-food producing fryers of avian species in the US.
Re: (Score:2)
Re: (Score:2)
I heard it was tall, glass buildings. They have staff to go around picking the dead birds up every day.
Tall, glass buildings? You mean all of that old office space that could become low-rent, low-income, low amenities, high-rise residential flats?
Re: (Score:2)
Re: (Score:1)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
SMRs are an expensive pipe dream (Score:1)
The real metric is $/MWh, and it is far worse for any of the non-existent SMRs than traditional big nuke, which are worse than any other electricity nuke.
In other words, SMRs are an expensive pipe dream designed by startups just to siphon funds.
Re: (Score:2)
Re: (Score:2)
These are not suitable for civilian use, for numerous reasons. One very expensive Russian stunt notwithstanding. Yes, you really do not know anything.
Re: (Score:2)
Re: (Score:2)
Because SMRs have been a hallucination by the nuclear fanatics to keep their delusion alive, nothing else.
I don't want to hear "hallucination" about nuclear from people that believe everything can run on wind and solar.
Nukes require their own cooling (Score:2)
Just sayin'
Re: (Score:2)
Just sayin'
Yes. What's your point? Every combustion based power source needs cooling. Wind, solar, and hydro stations probably also need cooling somewhere. I don't think Ellison was saying these currently-nonexistent SMRs would not need cooling, was he? Nor was he saying a gigawatt data center wouldn't need some serious heat management.
I wonder if anyone has done the math on recapturing datacenter heat. It's hot in the building, less hot outside, so it seems you ought to be able to generate power from the temperature
Re: Nukes require their own cooling (Score:2)
Re: (Score:2)
" done the math on recapturing datacenter heat." The big brains say there is not enough heat differential from data centers. It needs to be much hotter to make steam to turn a turbine. Cold climates could heat water for a town.
Thanks, that's not a surprising conclusion. It's certainly been cheaper to just vent the heat than try to capture it. I've seen datacenter designs where they just circulate ambient air through the building rather than use an A/C system. That's really cheap when it's cold and dry enough outside.
However, A/Cs work by concentrating and moving heat. It seems one ought to be able to concentrate the waste heat enough to create steam, or at least, pre-heat the water going to the boilers in the SMRs. As you say tho
Re: (Score:2)
Yes. What's your point?
Cooling generally requires water. Data centers that use a water cooling cycle use a lot of it. Which causes problems for some people. So now we are going to bundle two water-hogging uses together. Lets see who signs off on that land use permit.
Re: (Score:2)
SMRs don't exist. (Score:2)
True.
SMRs don't need cooling, because they don't exist.
Those are an economic aberration, designed to lure investors.
Seems to work, no need to cool the overheated investors.
Re: (Score:2)
Re: (Score:2)
There are lots of SMRs out there, mostly military, which is why you are not in the know.
Provide a citation that shows someone, anyone, has actually built and not just imagined a SMR with full passive cooling. NuScale claims they will have it, but they have built 0 units despite having approval to build one for testing for ages now. I can find a list of proposed designs, some of which it is claimed construction was starting on, but nothing actually built.
Re: (Score:2)
Re: (Score:2)
I fail to see the economics of SMR for military, you need a hotwar to actually make use of them, when not in use arent they just decaying?
In short, yes that is accurate. This is why we only actually use them for missile subs and for a couple of aircraft carriers. Nuclear power makes long missions in the former possible, and solves massive logistics problems for the latter as large carriers use immense amounts of fuel.
Re: (Score:2)
Re: (Score:2)
Those use case do not exist because .... economics.
Re: (Score:2)
But we already have nuclear power systems for large naval vessels
Yes, those are SMRs. Feel free to catch up to the conversation any time.
Re: (Score:2)
Re: (Score:2)
SMR = small modular reactor, none exist. Feel free to catch up with the acronyms.
https://en.wikipedia.org/wiki/... [wikipedia.org]
HTH HAND
Maybe he can design some cool stuff... (Score:2)
We have heard about new gen reactors, SMR reactors, and thorium reactors for a long time, not to mention small reactors which can do megawatts of power and just sit there, perhaps reprocessing fuel so the spent stuff can be kept in use until it is effectively dead lead. If it takes a billionaire to shove nuclear energy forward, so much the better.
Of course, it would be nice to have energy storage technology (doesn't have to be chemical batteries, other methods like phase changes, flywheels, supercaps, etc.
Re: (Score:2)
Maybe. But Oracle hasn't been the power-house of innovation since... well, a long time - and even then I'd say it was a bit debatable.
Nuclear power stations? Good if he can do anything useful in that space, but I won't hold my breath. If you can make such things using lawyers strong-arming people into things, then maybe it'll happen, otherwise I doubt it.
Excellent (Score:2)
Hopefully Oracle can become the SpaceX of nuclear power.
Or would you rather have China cornering the market and us buying from them?
Re: (Score:2)
we can run with out an t447 safety command console (Score:2)
we can run with out an t447 safety command console
Re: (Score:2)
If Oracle could do nuclear power "right", great for them. It means more people in the field, perhaps more money thrown that way. The more power we have available, the easier it is to transition to EVs, where power for the EV just might be carbon neutral.
Plus, nuclear power has a ton of applications. A ship sifting all the stuff from the Pacific Gyre can throw that into a large thermal depolymerization machine and get the short chain oil back out, which can be reused for plastics, or for other things with
Re: (Score:2)
Hopefully Oracle can become the SpaceX of nuclear power.
This is your first time hearing the name "Oracle" right?
Right?
company instead of government? (Score:2)
i thought nuclear reactors were a long-term thing. i know companies aren’t.
what would stop any company from shrugging and walking away if something went horribly wrong?
Re: (Score:2)
Re: (Score:2)
Erm (Score:5, Informative)
Bullshit detector reading off scale high. Ellison claims they have "building permits" for new reactors at some location they've "found"? Obtaining a site license for operating nuclear reactors, SMR or otherwise, is a Very. Big. Deal. in the US. Ellison can't just reality distortion field those into existence whenever he wishes. If there had been a new reactor site approved by the NRC, the process to achieve that would have been known about, publicly, several years ago at least.
Re:Erm (Score:4, Informative)
Bullshit detector reading off scale high. Ellison claims they have "building permits" for new reactors at some location they've "found"? Obtaining a site license for operating nuclear reactors, SMR or otherwise, is a Very. Big. Deal. in the US. Ellison can't just reality distortion field those into existence whenever he wishes. If there had been a new reactor site approved by the NRC, the process to achieve that would have been known about, publicly, several years ago at least.
Most companies that got permits to build nuclear reactors during the last "nuclear renaissance" before Fukushima held onto them, and just pay the few hundred grand a year it takes to maintain them given the potential value and relatively low cost. Those would include things like environmental site plans, and probably local permissions... the first few years of the process of starting a nuclear reactor done at least. And a lot of those permits were issued for locations that already had things like transmission infrastructure (usually existing coal power plants), and thus would be relatively easy to slot in a new power plant. I'm pretty sure that's what he's referencing. Whatever utility he has negotiated with has a site already queued up, so if and when some version of an SMR is greenlit construction can start very quickly. As to how long it actually takes to build it, that will depend entirely on the regulatory regime for those new reactors.
Re: (Score:2)
Given what we learned from Fukushima and subsequent inspections of other Japanese nuclear plants, it seems wise to re-visit those licences now and do a full re-evaluation of the sites. It turns out that the checks being done previously were not thorough enough.
In any case, even if they have the permit, no SMR is production ready. They are prototypes at best, and for commercial use the new plant design will need to be signed off on as well.
Re: (Score:2)
They don't even have prototypes. NuScale has yet to build a single reactor despite getting type approval! Yet we're still supposed to pretend their technology has been proven out according to the nuclear fanboys.
Re: (Score:2)
As to how long it actually takes to build it, that will depend entirely on the regulatory regime for those new reactors.
They've been approved to build one, but they haven't even started on a prototype. That's a clear sign that they know it won't work. Otherwise they'd be in a hurry to prove it and get to building more of them.
Re: (Score:2)
Re: (Score:2)
Hi from Australia.
It was announced this week that Larry's company Paramount is spending $US8 billion on one of our 3 commercial TV networks.
Opposition leader Peter Dutton is spruiking nuclear power as a solution to the closure of coal. And we don't have a tradition of traditional reactors.
Putting 2 and 2 together, Oracle may seek to build data centers in a 'friendly' nation.
Re: (Score:2)
Putting 2 and 2 together requires more mathematical capability than is presently on display in Australian politics.
It is presently illegal under Federal, State, and Territory laws to do most of the things needed to build and operate a nuclear power plant, fuel a plant, or dispose of its waste: a building permit for such a thing is a non-starter here. For shits and giggles, let's posit the appearance of magical SMR unicorns and native nuclear capability in the requisite timeline, no opposition to compulso
Re: (Score:2)
Re: (Score:2)
You seem to think that the rule of Law is important anymore. The only thing that matters is power and Ellison has billions of it. We shall see how this all works out.
Considering how crazy spend on "Cloud" is (Score:2)
It probably won't matter that their electricity is several times more expensive than solar and wind plus storage. I mean it's just a part of the cost of hosting servers and profit margins on cloud services are insane.
after fact (Score:1)
As much as I love nukes... (Score:2)
It would probably be cheaper to build a data center someplace cold and windy. Power it with windmills and big batteries.
Re: (Score:2)
If it's cold and windy, take out the mechanical components, flood the data centre with oil, and treat the whole data centre as a fully immersed liquid cooling system. Then stick your heatsinks and air conditioning units on the roof.
The outside cold will help take the heat away with far greater efficiency than you could achieve with regular cooling, reducing the power requirements.
so disappointed (Score:2)
i came here for "Beowulf" cluster of nuclear reactors based posts...
Re: (Score:2)
Three is a MOSIX cluster, no sense doing Beowulf on such small scale.
Re: (Score:2)
scale up baby