U.S. electric utilities are turning to artificial intelligence to forecast loads and prioritize maintenance as electricity demand from AI data centers rises, but executives said the industry remains far from letting large language models take actions that affect physical grid infrastructure. AI is supporting utility demand forecasting and maintenance prioritization, but uneven outputs, or those that can't be audited, according to Utility Dive [1].
Federal and state officials are watching the trend closely. Barriers to wider use include AI outputs that are uneven or cannot be audited in a regulated sector with strict safety and reliability standards.
The stakes are rising because data center electricity consumption is climbing. U.S. data centers consumed an estimated 183 terawatt-hours in 2024, more than 4% of the country's total electricity use, and demand is projected to grow 133% by 2030 to 426 terawatt-hours, according to a Kansas Health Institute report cited by Utility Dive