Are AI data centers eating the world’s electricity?

The evidence

Worldwide. The IEA says data centers used about 415 TWh of electricity in 2024, around 1.5% of the world’s total. Its projection is about 945 TWh by 2030, slightly more than Japan’s total electricity consumption today.¹

In the U.S. LBNL estimates U.S. data centers used about 176 TWh in 2023, which is 4.4% of U.S. electricity, up from about 58 TWh in 2014.^2,3 Its range for 2028 is 6.7% to 12.0% of U.S. electricity.^2,3 DOE says data-center load growth has tripled over the past decade.³

The trend was not always up. LBNL says U.S. data-center energy use was fairly stable at about 60 TWh from 2014 to 2016, and reached about 76 TWh (1.9% of U.S. electricity) in 2018.²

Per question, it can fall. Google says the energy of a median Gemini text prompt fell 33-fold over a recent 12-month period. That is a company claim for one product, and Google says it has not been independently verified.⁴

These figures cover all data centers, not only those running AI.

What we don’t know

• The 2030 and 2028 numbers are projections. They depend on assumptions about demand, hardware and efficiency, and the sources give them as ranges or scenarios, not as measurements. LBNL says the lack of direct energy data in a fast-changing sector limits its analysis, especially for future demand.²
• This page does not separate AI from other data-center work (email, video, cloud storage), because the sources cited here report data centers as a whole.
• We have no sourced answer on how much new power plants, local grids or electricity bills will change in any particular town. That depends on local decisions this page does not cover.

Related

• Is data-center water use just cooling?
• Does each AI question use a bottle of water?
• Glossary: kilowatt-hour and terawatt-hour
• Glossary: power usage effectiveness
• How AI works: energy
• Myths vs Facts: “always exploding?”