Al is changing how humans operate in the digital world, but it might also change how things work in the physical world.
Former Google CEO Eric Schmidt has said that AI could end up using as much as 99% of the world’s energy. He said this would be up from the current 3% of energy that AI was currently using. Schmidt made his comments while speaking to US lawmakers in a testimony on Artificial Intelligence.
“What we need from you, if I may say that directly, is need energy in all forms,” Schmidt said. “Renewable, non-renewable, whatever, it needs to be there and it needs to be quickly. I and others are investing in things like fusion, which are incredible, but they’re not going to arrive soon enough for the need,” he said.
“People are planning 10 gigawatt data centers. Now, just to do the translation. An average nuclear power plant in the United States is one gigawatt. How many nuclear power plants can we make in one year where we’re planning this 10 gigawatt data center gives you a sense of how big this crisis is,” he added.
“Many people think that the demand in of energy part that our industry (tech) takes will go from 3% to 99% of total generation. One of the estimates that I think is most likely is that data centers will require an additional 29 gigawatts of power by 2027 and 67 more gigawatts by 2030. Gives you a sense of the scale that we’re talking,” Schmidt said.
“These things are industrial at a scale I have never seen in my life. We have a bunch of regulatory issues around fixing the energy grid. It takes on average 18 years to get the power transmissions and so forth to put these things in place. We need to find federal ways to preempt that and make it happen faster in order to deal with the needs,” Schmidt said.
Schmidt seems to be saying that if AI does become commonplace, and AGI and Superintelligence is achieved, it would need an enormous amount of energy to power itself. All AI operations consume significantly more power than normal computer operations, with estimates saying that a single ChatGPT query takes 10 times as energy as a similar Google query. More intensive AI applications, like video and image generation will consume even more power. As such, as AI becomes more capable and its usage spreads, the amount of electricity it would require would increase exponentially. Schmidt seemed to be suggesting that at some point, AI could end up using 99 percent of all generated power. This makes it imperative that the US — and other countries — dramatically scale up their electricity generation efforts. Electricity is currently taken for granted, but access to large amounts of power might become the variable that decides the balance of power in a post-AGI world.