this post was submitted on 12 Oct 2025
76 points (100.0% liked)

technology

24040 readers
489 users here now

On the road to fully automated luxury gay space communism.

Spreading Linux propaganda since 2020

Rules:

founded 5 years ago
MODERATORS
top 11 comments
sorted by: hot top controversial new old
[–] invo_rt@hexbear.net 31 points 2 days ago (3 children)

It would be more hilarious if the data centers WEREN'T going up instead of going up anyway and raising everyone else's power bills by up to 50%.

That being said, cheap access to electricity has always been a competitive advantage of the US broadly. If that becomes not the case (which iirc China is supposed to be generating as much power as the entire US in just renewables by 2030), it'll probably be a bigger limiting factor on AI than raw compute power.

[–] yogthos@lemmygrad.ml 25 points 2 days ago (1 children)

Energy prices in China are also far lower than in the states. In a way, the AI race is starting to mirror the Cold War arms race. Back then, the US used its economic might to outspend the USSR on military tech, forcing Moscow to divert scarce resources from areas that improved the standard of living, such as light industry, towards the military. Now, China's lower energy costs and state-backed infrastructure investments are pressuring the US to pour their limited resources into AI infrastructure just to keep up. While LLMs get most of the hype, AI encompasses everything from autonomous systems to logistics planning. People in Washington are terrified that if China pulls ahead, they're never going to catch up. From a strategic perspective, they have no choice but to do what they're doing.

[–] invalidusernamelol@hexbear.net 11 points 1 day ago* (last edited 1 day ago) (2 children)

Yep, a Markov chain solver that's the size of a city is massively useful for a lot of specific planning cases. Especially once you're looking at large quantities of data. The actual use case for the US is kinda moot though, since the markets aren't structured. They're controlled entirely by the whims of a couple people which isn't predictable by a Markov chain.

China will see a lot of utilization and benefit though, Gosplan used hand computed Markov chains, but an already centrally planned economy with tons of historical data and predictable, mostly bubble free history will actually be able to provide useful planning outputs.

The US I think knows this, and isn't planning on using this machine to stabilize the economy, but instead as a weapon of war. Domestically, they'll use it to pre-crime and mass arrest people that they don't like. Abroad they'll use it as an inverse planner to give them targets that will specifically destabilize the economy.

[–] yogthos@lemmygrad.ml 10 points 1 day ago (1 children)

The real story is the looming energy crisis as opposed to what the US plans to do with AI models. The US power grid is already running on a razor's edge with reserve margins sitting at a scary low 15%. As more data centres keep getting deployed, the emergency buffer for heat waves, unexpected outages, and so on, is about to get swallowed whole.

The problem is you can't just wish new power plants into existence. Building new generation, whether it's a gas plant, nuclear, or renewables with their storage needs, takes the better part of a decade and has gotten brutally expensive thanks to the trade war. So we're about to see a classic supply and demand shock. A massive, constant, and inflexible new demand from AI data centers is going to slam into a power supply that can't keep up. The inevitable result is that electricity prices are going to go parabolic for everyone, not just the tech companies.

As a result, we'll see a cascading effect on the broader economy. First, grid stability becomes a national issue. Those rolling blackouts that were a regional problem in Texas or California could become a lot more common as the system gets overloaded. Second, and more dangerously, it torpedoes US manufacturing competitiveness. The US spent the last decade leveraging cheap energy to keep domestic manufacturing viable. Now, the US is about to have Europe's problem, where prohibitively high energy costs make it unprofitable to run a factory. Essentially, whatever is left of the physical industrial base will have to be sacrificed to power the AI bubble, and the economic hangover is going to be brutal.

[–] Des@hexbear.net 5 points 1 day ago* (last edited 1 day ago)

can't even get a decent home solar unit with battery for the inevitable either because they are like three times more expensive then they used to be (which was still too damn much). yay tariffs + no more subsidies

so get ready for everyone to have gas guzzling backup generators kicking on multiple times a day (if they are lucky to have one). oh yeah and wet bulb events as well

[–] vovchik_ilich@hexbear.net 7 points 1 day ago (1 children)

Paul Cockshott has some videos talking about the idea of computational costs of planning an economy using the most modern methods and they can be done at home with a powerful consumer-grade PC in a matter of hours, we don't need city-sized supercomputers for that

Yeah, it's really easy to accomplish with some basic tensor flow models. The only upshot of large hardware is being able to have a plethora of models.

Plus looking at the Chinese developments, they're focusing on model distillations which are orders of magnitude more efficient than generalized LLMs.

[–] miz@hexbear.net 16 points 2 days ago

the electricity rate in China is about 7.5 cents per kWh, the lowest in the USA I can find is Idaho at 63% higher

[–] sefra1@lemmy.zip 12 points 1 day ago (2 children)

By 2030 the AI bubble should be long bursts

[–] juniper@hexbear.net 9 points 1 day ago

please dont get my hopes up

[–] kotak_doost@hexbear.net 5 points 1 day ago

Yeah but it'll be just in time for the next bubble. Place your bets, I'm going with quantum computing.