AI Expansion Outpaces U.S. Electric Grid Development
The rapid growth of AI is straining the U.S. electric grid, as data centers consume increasing amounts of electricity faster than infrastructure can be built to support them. Experts warn that tech companies like OpenAI, Google, and Meta are expanding their operations at a pace utilities cannot match due to lengthy planning processes required for grid upgrades. By 2030, AI-related data centers could account for nearly 12% of U.S. electricity use—six times the pre-AI level in 2018. Electricity demand is rising faster than expected across North America, with summer peak demand projected to grow by over 224 GW in the next decade due largely to AI data centers. In some Western regions, planned data center loads already represent up to 40% of forecasted electricity needs. The grid was already under strain before this surge, as utilities had not built significant new infrastructure for decades. Industry experts note that securing power is now a major bottleneck in developing new data centers. A gigawatt of electricity—enough for 750,000 homes—can take years to generate and deliver, while companies often request it within two years. The Department of Energy has acknowledged the urgent need for more infrastructure due to rising demand from AI and other sectors.
