The high-energy use of new technologies could delay the coal phase-out in the US. AI is driving this trend.
Some US states are delaying plans to close coal-fired power plants because of the huge demand for electricity from new technologies, the Financial Times reports. The growing demand for electricity from AI, crypto-mining and cloud services is reportedly forcing the US to rely on the old energy source coal for longer.
Operators are worried about the stability of the grid and see demand for electricity rising. By the late 2020s, 54 gigawatts of coal are expected to go offline, about 4% of the country's total electricity capacity and 40% less than predicted last year.
Power demand for data centers could double by 2030, accounting for about 9% of total U.S. electricity demand, according to a study by the Electric Power Research Institute. ChatGPT alone could consume nearly 10 times as much electricity as Google searches, the International Energy Agency estimates.
Simply delaying a shutdown date doesn't mean the plants will be used, notes Seth Feaster of the Institute for Energy Economics and Financial Analysis. Coal's share of U.S. electricity has dropped from 40% in 2014 to 16% in 2023, according to the US Energy Information Administration.
OpenAI CEO hopes for nuclear innovation
Training and running generative AI models takes a lot of computing power. Data centers for generative AI systems also need lots of fresh water to cool the processors and make electricity, says Kate Crawford, a researcher at USC Annenberg and Microsoft Research. By 2027, global water use for AI could equal half of the UK's use.
OpenAI CEO Sam Altman doesn't even try to hide the fact that developing ever more capable AI models will require huge amounts of energy and a breakthrough in energy production. He sees fusion energy, specifically nuclear fusion, as a possible answer.
