Utilities cite AI data‑centre strain
U.S. utility companies are reportedly stepping back from some pollution‑reduction commitments and pointing to rising electricity demand from AI data centres as a factor. The reporting links the capacity and cost pressures from large compute loads to shifting utility goals and higher bills. (thecooldown.com)
Some United States utilities are loosening pollution-cutting plans as electricity demand from artificial-intelligence data centers surges. (apnews.com) The clearest example is Nevada, where NV Energy told regulators it may miss the state’s requirement to supply 50% renewable electricity by 2030 because of proposed data-center load. The utility said the new requests could require roughly three times the electricity used by Las Vegas. (apnews.com) In North Carolina, Senate Bill 266 became law on July 29, 2025 after lawmakers overrode Governor Josh Stein’s veto. The law removed the interim requirement to cut carbon dioxide emissions 70% from 2005 levels by 2030, while leaving the 2050 carbon-neutrality target in place. (mcguirewoods.com) The pressure starts with the size of the load. A July 2024 Department of Energy advisory report said hyperscale data centers are seeking connections of 300 to 1,000 megawatts, with lead times of one to three years, stretching local grids. (energy.gov) Federal estimates show why utilities are scrambling. Lawrence Berkeley National Laboratory said United States data centers used 176 terawatt-hours in 2023, or 4.4% of national electricity, and could reach 325 to 580 terawatt-hours by 2028, or as much as 12%. (energy.gov) That growth is colliding with state climate rules and utility planning cycles. Rocky Mountain Institute said in its April 2025 review that utilities were already revising integrated resource plans around new large-load customers such as data centers. (rmi.org) The money fight is over who pays for the extra wires, substations, and power plants. Brookings wrote in March 2026 that new data centers can raise residential rates by forcing new generation and transmission costs into the broader system. (brookings.edu) California’s Little Hoover Commission made the same warning in March 2026, saying lawmakers should make technology companies, not households, cover grid-upgrade costs tied to artificial-intelligence data centers. (calmatters.org) Utilities and technology companies say the buildout also brings jobs, tax revenue, and a stronger domestic computing base. The Department of Energy’s 2024 advisory report said the challenge is to add that capacity “reliably and affordably” without shifting costs to existing customers or increasing greenhouse-gas emissions. (energy.gov) The next test is whether regulators let utilities add fossil-fuel generation, rewrite clean-energy timelines, or force large data-center customers to pay more of the bill. Nevada and North Carolina have already shown how fast the artificial-intelligence power boom can move from server rooms into state energy policy. (apnews.com)
