Data center electricity consumption could double to roughly 1,000 terawatt-hours a year by 2030, the International Energy Agency (IEA) warned on November 19, with the global expansion of artificial intelligence accounting for three-quarters of that growth1,3.
The IEA's Stated Policies Scenario projects that datacenters will produce 70% more carbon emissions within the next four years as the United States and China turn to fossil fuels to power their AI buildouts. Global data center investment is expected to reach $4 trillion by 2030, according to the agency4.
The surge threatens to stretch power grids to their limits2. Tight supply chains, local opposition, and limited grid capacity could slow new data center construction, the IEA said. The agency cited large project pipelines in Texas, Virginia, Finland, and the U.K. as particular pressure points.
Fewer than 100 major AI projects would require a combined 50 gigawatts of power, matching peak U.K. electricity demand. To keep pace, the IEA's Stated Policies Scenario calls for 26 million kilometers of new grid lines by 2035 and the replacement or refurbishment of about 16 million kilometers of existing lines. Under the agency's Current Policies Scenario, grid spending rises 75% over the next decade, making power networks a major energy investment theme.
The carbon trajectory the IEA describes follows a pattern already visible on the ground. U.S. data center operators have been pulling retired jet engines from aircraft boneyards, rebuilding them to burn natural gas, and using them to generate electricity on-site, a workaround that can deliver power in as little as 24 months while grid-connection queues stretch years longer[1].
ANALYSIS The IEA's projections frame AI infrastructure not as a future planning challenge but as an immediate constraint on global electricity systems, with grid investment now competing for capital alongside the data centers themselves.