The US will face an electricity shortage in two years, provided that demand from AI data centers continues to grow. Business Insider writes about this with reference to Bernstein Research.
Excess demand over supply in the electric power industry threatens to increase prices for computing power, analysts said. At the same time, any assessment of energy consumption by AI should be perceived with a degree of skepticism due to the rapid change in variables: the number of outputs over a certain period of time and their complexity, the size of the model, the performance of chips and servers.
Bernstein estimates that by 2030, the number of AI search queries will reach 10 billion per day, and responses will consist of an average of 375 words. The figures are comparable to Google, while requests to ChatGPT require about 10 times more computing power, experts say.
Why is AI so voracious?
Business Insider draws attention to the fact that the graphics processors used for AI training and operation emit more heat compared to conventional CPUs, and the server rack for neural networks requires three times more energy.
The development of hardware and software for artificial intelligence can lead to increased productivity — more computing with less power. Bernstein's estimates are based on a steady improvement in the ratio of computing power to energy consumption.
Bill Gates urged not to overdo it with concerns
Microsoft co-founder Bill Gates urged environmentalists and governments "not to overdo it" in their concerns about the huge amounts of electricity consumption from generative artificial intelligence. According to him, data centers will lead to an increase in global energy consumption by 2-6%.
In May, Microsoft said that greenhouse gas emissions have increased by almost a third since 2020, largely due to the construction of data centers. Gates noted that technology companies will contribute to the development and implementation of clean energy because they will pay a "green premium" — a higher price for it.
"Technology companies are people who are willing to pay more and help develop green energy capacities," he said.
Recall, according to Bitwise expert Juan Leon, the infrastructure of miners and their access to cheap electricity can satisfy the growing demands of artificial intelligence.
In June, Core Scientific signed a 12-year contract with AI firm CoreWeave to provide infrastructure with a capacity of 200 MW for hosting Nvidia GPUs.