Monday, April 1st 2024
US Government Wants Nuclear Plants to Offload AI Data Center Expansion
The expansion of AI technology affects not only the production and demand for graphics cards but also the electricity grid that powers them. Data centers hosting thousands of GPUs are becoming more common, and the industry has been building new facilities for GPU-enhanced servers to serve the need for more AI. However, these powerful GPUs often consume over 500 Watts per single card, and NVIDIA's latest Blackwell B200 GPU has a TGP of 1000 Watts or a single kilowatt. These kilowatt GPUs will be present in data centers with 10s of thousands of cards, resulting in multi-megawatt facilities. To combat the load on the national electricity grid, US President Joe Biden's administration has been discussing with big tech to re-evaluate their power sources, possibly using smaller nuclear plants. According to an Axios interview with Energy Secretary Jennifer Granholm, she has noted that "AI itself isn't a problem because AI could help to solve the problem." However, the problem is the load-bearing of the national electricity grid, which can't sustain the rapid expansion of the AI data centers.
The Department of Energy (DOE) has been reportedly talking with firms, most notably hyperscalers like Microsoft, Google, and Amazon, to start considering nuclear fusion and fission power plants to satisfy the need for AI expansion. We have already discussed the plan by Microsoft to embed a nuclear reactor near its data center facility and help manage the load of thousands of GPUs running AI training/inference. However, this time, it is not just Microsoft. Other tech giants are reportedly thinking about nuclear as well. They all need to offload their AI expansion from the US national power grid and develop a nuclear solution. Nuclear power is a mere 20% of the US power sourcing, and DOE is currently financing a Holtec Palisades 800-MW electric nuclear generating station with $1.52 billion in funds for restoration and resumption of service. Microsoft is investing in a Small Modular Reactors (SMRs) microreactor energy strategy, which could be an example for other big tech companies to follow.
Source:
Axios
The Department of Energy (DOE) has been reportedly talking with firms, most notably hyperscalers like Microsoft, Google, and Amazon, to start considering nuclear fusion and fission power plants to satisfy the need for AI expansion. We have already discussed the plan by Microsoft to embed a nuclear reactor near its data center facility and help manage the load of thousands of GPUs running AI training/inference. However, this time, it is not just Microsoft. Other tech giants are reportedly thinking about nuclear as well. They all need to offload their AI expansion from the US national power grid and develop a nuclear solution. Nuclear power is a mere 20% of the US power sourcing, and DOE is currently financing a Holtec Palisades 800-MW electric nuclear generating station with $1.52 billion in funds for restoration and resumption of service. Microsoft is investing in a Small Modular Reactors (SMRs) microreactor energy strategy, which could be an example for other big tech companies to follow.
98 Comments on US Government Wants Nuclear Plants to Offload AI Data Center Expansion
Hell just adding solar panels on individuals houses was a butt show waiting game attempting to get on the grid to sell back a little to subsidize what they cost/ install/.. monthly
Hell these were people signing 20+ year contract plans that took months to get completed if they ever do.
Also they were not adding batteries for storing the power they generated either
Instead these companies were also selling natural gas generator systems if the power goes off hehe
But for commercial entities, on site power delivery is more pragmatic (if not very expensive). But then, these massive companies can afford to invest in it.
I have family there. California is much more meme than reality, and outdated one at that. At most today, electricity is expensive, but not unreliable.
To take it back, my point was Texas isolates itself grid wise and that's why its worse off than even California. Nothing more. The isolation itself is factual.Sadly, it seems to be these days. Case in point how everyone lit up at the mention of Texas.
More like people listed States, Cali for one that black outs are an event that happens quite often at the least monthly and some genius "in their own mind" named a rare event Texas had as some sort of proof lol
Details of the Texas single blackout are troubling no doubt though but before had the lowest electric rate in the US and likely still does.
www.energybot.com/electricity-rates/
It's 28th on my count.
Nebraska is the cheapest.
Not sure how you got this site to give you a list but kudos hehe
Before the crazy ass freeze deal happened I just signed a 36 month fixed rate contact for 10 cents kwh
People that weren't on fixed rate plans got hosed for sure.
You should start mining. :D
Yeah I shopped in winter when rates are lower hehe
I almost waited to long seeing the freeze hit just after the contact kicked in so couple more weeks and I'd been hosed as well :fear:
Heck I get more blackouts than Cali or Texas (I live in a forest with above ground wires).
Cali has very poor forest management with line clearing likely electric provider and state ignorance combined
With so many forest fires I can't say they are willing to do a lot of fire breaks even to this date knowing how common they are yearly.
Either way seems Texas single event gets pushed as some sort of proof that utility isolation is bad but this freeze event was pretty infrequent and hopefully never happen again.
Frankly being in Louisiana atm there aren't any shopping around for providers here the price is the price so pretty much a monopoly system lol
Well yeah but neither does Texas references but it was lol
Nuclear is so out of touch with reality it's baffling how it came up when wind and solar are clear a million times easier/ faster to get going with the AI dog whistle hehe
Sorry this begs the question did they ask AI what type of power source to use for the AI data centers lol
Not much more productive dialog here to be had. I'll be going.
In that list the USA is #126. There are ~70 countries doing 50% or more and you choose to be fixated in the 100%.