AI Power Demand Outlook Puts US Utilities on Investors' Radar
US electricity forecasts highlight a growing infrastructure challenge as AI data centres increase demand for reliable power.
Reuters reported that the EIA expects record US electricity consumption in 2026 and 2027, with AI-related computing among the sources of growth. The forecast is not a direct measurement of the power required by any single AI company. Weather, homes, factories and other large computing facilities also influence the outlook, so a national projection cannot be treated as a data-centre utilization report.
A server campus needs an actual grid connection as well as land, buildings and installed accelerators. The IEA notes that cooling, networking and other equipment account for part of a centre's load. A project can be built before it is fully energized, and available megawatts do not automatically imply that all machines are earning revenue.
Before valuing an AI infrastructure investment, distinguish permitted capacity, energized capacity and equipment deployed for billable workloads. Review utility interconnection queues and contracted electricity pricing. Even if national power demand rises, a particular operator's margins could suffer if local infrastructure or financing costs rise faster than its customer revenue.
Reporting sources & references
These links identify the reporting or public materials on which the article is based; they do not imply our newsroom witnessed the events.