Artificial intelligence has officially become an energy problem. The insatiable amount of energy required to power data centers is huge and expected to double by 2030. In fact, expects anticipate that energy consumption by AI will increase by 200%. Other forecasts suggest that AI energy consumption will increase sixfold.
Dec. 5, 2024 in Paris. The first Global Conference on Energy & AI will take place in December, announced Fatih Birol, the director of the International Energy Agency (IEA). This global summit aims to gather industry experts to discuss potential measures to address the rising energy demand to power AI.
“Although their energy use today is small in the context of the global energy system, demand has increased rapidly in recent years and is expected to keep growing. Considering the implications of the AI revolution is therefore critical to understanding the future of energy,” the IEA stated in a press release.
More than 1,000 TWh by 2026. The latest IEA report predicts that by 2026, data center energy consumption will exceed 1,000 terawatt-hours, equivalent to the energy consumption of Japan. To put this in perspective, in 2022, global data center consumption was around 460 TWh, accounting for approximately 2% of all electrical energy use.
In comparison, it’s estimated that cryptocurrencies–which also require a large amount of energy–will consume about 160 TWh by 2026, with a projected growth of over 40%. However, this is still significantly less than the expected impact of AI.
Energy drought-bound. Tesla CEO Elon Musk has recently suggested that we’re approaching an “energy drought” because the AI will run out of transformers by 2025. In Musk's words: “The artificial intelligence compute coming online appears to be increasing by a factor of 10 every six months... Obviously that cannot continue at such a high rate forever, or it’ll exceed the mass of the universe, but I’ve never seen anything like it.”
Nuclear power comes around. The debate about whether to turn to nuclear power always resurfaces whenever significant amounts of energy are required. U.S. energy secretary Jennifer Granholm believes that nuclear power will be necessary to meet AI energy demands. OpenAI CEO Sam Altman shares this view and predicts that AI will consume more energy than anticipated.
There are concerns about whether the energy sector will be prepared for this impending revolution. At least the IEA is already taking steps to make it a priority.
Image | Federico Beccari via Unsplash
View 0 comments