OpenAI tells Trump to build more power plants or China wins the AI arms race OpenAI wants the Trump administration to build 100 gigawatts of additional electricity generation capacity per annum to avoid the US being overtaken by China in the AI arms race. The highly-valued generative AI biz says electricity is "a strategic asset" critical to building the AI infrastructure that will secure US leadership …
Silicon Valley wants the Washington bureaucracy to build national infrastructure projects to beat the communist's central planned economy?
Obviously a spoof post. If there as a demand for this electricity Private Equity would have already built it quicker, cheaper and more efficiently
Yes, it's another instance of uncapitalist socialism at work in the USA. The USA pretends it is capitalists, but really it is not. With the utterly mind-boggling sums handed out to 'private' companies with no risk to said private companies, endless subsidies for farmers, and so on, the USA is not a symbol of capitalism. The same receivers of this largesse are the first to peer down their noses at 'lazy' people burdened with healthcare costs. They just need to put their back into it, and it will be A OK.
The USA is an oligarchy.
"Private Equity would have already built it quicker, cheaper and more efficiently"
Private Equity isn't stupid (I may be wrong). They aren't going to sign up for infrastructure development unless there's some guarantee that the customer (and revenue) will be there in the long term.
Silicon Valley is more like sports team owners. Build us a new stadium or we'll move your home team across the country. So the politicians build it (to avoid the wrath of the fans) and the team is sold anyway.
build 100 gigawatts of additional electricity generation capacity per annum
that would seem to be about 100 average Nuclear power stations per year - they take a long time to build and commission:
"A typical nuclear reactor produces 1 gigawatt of power per plant on average."
From: https://www.energy.gov/ne/articles/infographic-how-much-power-does-nuclear-reactor-produce
Unless the USA has a lot of old, but still safe reactors 'mothballed' at the moment, I'd be interested to know if this is at all feasible. (And no, I don't think that asking an AI 'entity' how to do it would produce a practical answer.)
100 gigawatts of what though? A quick google indicates that most of that 429 GW of new capacity added in China is wind and solar. Another quick google and some basic math shows that wind and solar in China are running at about a 17% capacity factor. That is, they actually output only the equivalent of 17% of their nameplate capacity when you take into account that they can't run at full capacity all the time (this is pretty normal). This will compare to say a well run nuclear power plant that can run at 95% capacity factor over multiple years.
Let's put it another way, 100 GW of wind and solar is only equivalent to 17 GW of nuclear.
So if they are talking about adding 100 GW of wind and solar capacity in the US, then that amounts to (assuming similar numbers as in China) of 17 nuclear power plants per year. That's still a lot, but nowhere near the same problem.
Of course an electric utility would would be taking on a huge risk to build new generating capacity just to feed the AI bubble in the US as it is quite likely that this capacity would never be used once the bubble pops. If they can bring old plants back on line that is a safer option as it is likely those plants will in fact never be required to operate.
Even if all that 429 GW is wind/solar that's still equal to ~70 GW of nuclear/gas capacity. And their deployments of wind and especially solar are accelerating. They are on pace to install over a terawatt of solar alone per year by the end of the decade, and are accelerating deployment of batteries as well.
Meanwhile the US is run by a moron who kills off all renewable energy projects he can as "woke" and promotes coal power every chance he gets.
Wouldn't matter how much you bribed him, one of his very few personal beliefs is a seething hatred of wind power, because he built a golf course in Scotland once and was unhappy someone put wind turbines where they could be seen from the course and he wasn't able to abuse the UK legal system like he does the US' to stop them.
If the end goal is to provide nuclear power for AI, then this is indeed just a great big pile of debt. China has motives for building nuclear power other than providing billionaires with cheap power.
Of course, OpenAI wants the US tax payer to underwrite the costs. Socialise the costs, walk away with the profits.
If enough reliable, low-carbon generation is built before the bubble bursts, the sudden excess of capacity could meaningfully contribute to a general move to electrification. So perhaps the prize is an inhabitable planet.
But somehow it'll probably still end up that we're all fucked.
Or do you think it will be more likely that government won't manage to build 1000 new nuclear power stations this year and hooking up these data centers to the grid will take too long and cost too much.
So the solution will be 10000s of 'temporary' ie permit exempt, direct-drive gas turbine generators <cough>jet engines<cough> stacked in parking lots ?
If OpenAI needs the power then surely OpenAI can provide the capital investment to build it? This is, after all, how freewheelin' capitalism worked back the Gilded Age. But, instead of constructing it, taking the risk and (maybe) reaping the reward they'd rather either buccaneer style rob everyone else of their power or lean on the 'guvmint' -- aka "We, The People" -- to provide it for them.
Invoking China's is just another of those missile gap lobbying tactics**. Maybe they should copy the Chinese and figure out how to run AI levels of processing while using a lot less power than they do at the moment?
(**The "Missile Gap" was a Really Big Deal in the 1950s, it was floated by the MIC as the reason why we had to spend big on their missiles etc. because it was clear that The Russians were overtaking us, we were falling behind.)(Yes, the self same "Russians" that we all know today.)(The Gap was thought to be by many back then a bunch of BS. Post Cold War we learned that our suspicions were well founded -- the 'Gap' turned out to be a figment of the MIC's Marketing/Lobbying imagination.)
(In a rather ironic twist it appears that these days the "Missile Gap" is now very real but we're so busy chasing casino profits that we're too stingy to invest over the long term in just about anything unless it yields that "twelve and a half percent" -- or more -- that the parrot is always going on about.)
"Yes, the self same "Russians" that we all know today." No, Russians like Sergei Korolev and Valentin Glushko. They launched the first earth satellite, the first living being into orbit (a dog), the first probe to pass by the Moon, the first probe to take a picture of the far side of the Moon, the first man in orbit (surely you've heard of Yuri Gagarin!), the first two-manned orbiter, did the first space walk, and a lot of other firsts.
Was the Missile Gap real? No, it was a mis-analysis by the CIA (who by the way has a web page about how this happened). But that doesn't mean those Russians were dummies, far from it.
The Trump administration is propping up the big AI companies and giving them the illusion of strategic importance and legitimacy. In return, the big AI companies are making a case to justify Trump's push for "Big Beautiful Coal".
It's really not that complicated.
This should be interesting. The only thing I can see working is a "69" position, because it seems to me that Drumpf and the Republican Party are every bit as greedy as OpenAI's management and shareholders.
What they're really saying is "the government should spend tax dollars on AI-specific power generation requirements rather than giving us an outright grant to build the power we need because it's 'bad optics' after Drumpf's largesse to the Uber-rich in the budget."
Last rites before governments (and vendors, and VCs) grok that LLMs aren’t the future of AI and their energy demands are just a vain hope to fund their operations while their business model never will.
Just waiting for real world, reasoning, learning and memory models to emerge from Google Deepmind, Meta et al. That will precipitate the industry adjustments.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
