Global electricity consumption by data centers is on track to exceed 1,000 terawatt-hours in 2026, roughly equal to the total electricity use of Japan, according to the International Energy Agency. The projection reflects a rapid acceleration driven largely by artificial intelligence workloads and the power-hungry processors that run them.

The United States accounts for a large share of the total. The IEA projected US data center electricity demand would rise above 250 terawatt-hours in 2026, making the country the single largest source of data center power consumption. Much of that growth traces to a surge in investment in graphics processing units used to train and run AI models.

The agency described the pace of growth as steep. AI-related computing demand has expanded far faster than earlier forecasts anticipated, and the sector's electricity appetite is expected to keep climbing as new facilities come online. Cryptocurrency mining adds further load, though AI has become the dominant driver of new demand.

The figures carry direct implications for electricity systems. Concentrated data center growth strains local grids, lifts wholesale power prices, and pressures utilities to add generation and transmission quickly. The IEA has urged planners to account for the geographic clustering of data centers, since the load is not spread evenly across regions. For the US grid, the projections underscore how a single fast-growing sector is reshaping national electricity demand.

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