Elon Musk has sounded the alarm on a possible power shortfall that could hit AI data centres as soon as next year. Speaking from Tesla’s Texas headquarters, he revealed that his AI venture, xAI, is building a gigawatt-scale facility near Memphis, Tennessee—a project expected to finish within six to nine months. To put that in perspective, a gigawatt is roughly equivalent to the output of an average U.S. nuclear plant, according to the Department of Energy.
Musk pointed to three main hurdles in scaling AI: chips, transformers, and power generation itself. “As we solve the transformer shortage, there will be the fundamental electricity generation shortage,” he told CNBC’s David Faber. Even as tech obstacles are tackled, power generation could soon emerge as a critical bottleneck.
This concern mirrors Google’s February warning about a looming power capacity crunch in the United States—a situation intensified by stiff competition with China’s rapid AI advances. At a New York conference, Caroline Golin noted that Google’s exploration into nuclear energy reflects an urgent need for more reliable power, especially as renewable energy struggles with grid stability.
Musk also highlighted the stark contrast between China’s swift expansion in power generation and the relatively static output in the U.S. Meanwhile, xAI’s plan to use natural gas turbines at its Memphis centre has drawn ire from environmental groups, who argue that this approach sidesteps Clean Air Act regulations.
Industry opinions remain mixed. While Dominion Energy reports steady demand for data centres, Constellation Energy cautions against overestimating the load. Constellation CEO Joe Dominguez suggested that the current forecasts might be overstated, urging a measured approach to future capacity planning.
