AI lifts energy demand

Artificial intelligence is pushing U.S. power demand high enough that gas pipelines, gas turbines, and utility pollution plans are being rewritten around data centers. (eia.gov, fortune.com, turbomachinerymag.com) Data centers are warehouse-sized buildings full of servers, and those servers need steady electricity every hour, not just when the wind blows or the sun is out. The U.S. Department of Energy said in December 2024 that data-center load growth had tripled over the prior decade and could double or triple again by 2028. (energy.gov, iea.org) That demand is now showing up in old-line gas infrastructure. Williams Companies said it will break ground on April 14, 2026, on its Northeast Supply Enhancement project, an expansion of the Transco system across Pennsylvania, New Jersey, and New York that executives described as the first new pipeline into New York in more than a decade. (fortune.com, williams.com) Power producers are also chasing gas turbines, which are jet-engine-like machines that burn fuel to make electricity and can ramp up faster than many large power plants. Turbomachinery Magazine reported that aeroderivative turbines are in demand because data-center operators want fast, reliable generation near new campuses. (turbomachinerymag.com) The rush is colliding with manufacturing limits. Wood Mackenzie said global gas-turbine orders reached 110 gigawatts by the end of 2025 against annual manufacturing capacity of 60 to 70 gigawatts, and it expects prices to hit about $600 per kilowatt by the end of 2027, up 195% from 2019. (woodmac.com, publicpower.org) Utilities are changing their planning, too. Reporting from The Cool Down, based on recent filings and Associated Press reporting, said NextEra Energy dropped its goal of reaching zero emissions by 2045, while Nevada utility NV Energy said it may need more fossil generation to serve proposed data-center loads. (thecooldown.com, techxplore.com) Bills are part of the fight. Brookings said new data centers can raise residential electricity costs because utilities must build new generation and grid equipment, and CNBC reported that states and regulators are debating whether large technology companies or ordinary ratepayers should absorb those costs. (brookings.edu, cnbc.com) The federal forecast has shifted with that backdrop. The Energy Information Administration said on April 8, 2026, that U.S. electricity consumption has risen 2.1% a year on average over the last five years after roughly 15 years of flat demand, with data-center server use now a major driver in its long-term outlook. (eia.gov) The argument is no longer whether artificial intelligence needs more power. It is whether the grid can add that power fast enough without locking in years of higher emissions, higher equipment costs, and higher utility bills. (eia.gov, brookings.edu, woodmac.com)