The Energy Institute reports that U.S. data centers now account for nearly 40% of global data-center electricity consumption.

The first-ever global tracking of data-center power demand reveals that the United States drove nearly half of worldwide growth in 2025, with artificial intelligence workloads pushing total consumption to 788 terawatt-hours (TWh).

According to the Energy Institute’s 2026 Statistical Review of World Energy, produced in collaboration with Ember and supported by S&P Global Energy, this marks the first inclusion of data-center power demand tracking in the publication’s 75-year history. Global data-center electricity consumption surged 19.7% year-over-year, climbing from 658.2 TWh in 2024 to 788 TWh in 2025. Looking further back, the trajectory is even steeper: consumption has roughly doubled since 2020, when data centers drew 410.8 TWh.

The United States sits at the center of this acceleration. American data centers consumed 312.6 TWh in 2025, accounting for 39.7% of the global total. The country also contributed approximately 49% of worldwide growth, adding 63.5 TWh of new demand in a single year. China ranked second, capturing 26.1% of global data-center electricity demand, followed by Europe at 18.4%.

While overall U.S. electricity demand grew by 3% in 2025—matching the global average—data centers now represent about 2% of global electricity generation. Forecasts suggest data-center power draw could double again by 2030. This rapid expansion underscores a structural imbalance: total energy supply growth stands at 1.7%, while electricity demand growth reaches 3%.

The report also identifies electric vehicles alongside data centers as primary drivers of rising electricity demand. In response, major technology firms are increasingly securing carbon-free baseload power. Early examples include Microsoft’s agreement to restart a unit at the Three Mile Island nuclear facility and Amazon’s investments in nuclear development. Additionally, companies such as Hut 8, Core Scientific, and Iris Energy have announced strategies to dedicate portions of their power capacity specifically to AI and high-performance computing workloads.