Global electricity consumption from data centers is climbing steeply as artificial intelligence workloads multiply, according to data from the International Energy Agency and industry forecasters. Data center demand reached about 485 terawatt hours in 2025, a 17 percent increase over the prior year, with AI-specific facilities growing far faster at roughly 50 percent.

The near-term outlook shows the trend accelerating. Global data center electricity consumption is forecast to reach 565 terawatt hours in 2026, while worldwide data center power capacity is expected to rise 27 percent to 132 gigawatts, up from 104 gigawatts in 2025. In the United States, more than 4,500 facilities already draw about 176 terawatt hours a year, equal to 4.4 percent of national electricity use.

The longer horizon points to a doubling. The agency base case projects global data center electricity consumption reaching around 945 terawatt hours by 2030, just under 3 percent of total global electricity, with installed capacity near 290 gigawatts. Accelerated servers running AI workloads are the main engine, projected to grow about 30 percent a year, well ahead of the 9 percent annual growth for conventional servers.

The figures underscore why utilities and grid operators are scrambling for new generation and transmission. The gap between how quickly compute capacity can be built and how slowly power infrastructure is added has become the central constraint on data center expansion heading into the second half of the decade.

Source: International Energy Agency - https://www.iea.org/reports/energy-and-ai/energy-demand-from-ai