US grid planners are being asked to do something they have not been asked to do since the postwar buildout: add a decade's worth of generation and transmission on a five-year clock. The North American Electric Reliability Corporation in a single year lifted its ten-year build target from 80 to 132 gigawatts, a 65 percent jump the permitting and construction cycle cannot absorb. The decade-ahead math has stopped being a planning exercise and become a procurement race.
The Las Vegas Sun's reporting, drawing on Lawrence Berkeley National Laboratory data, traces the source of the new demand. Data centers now account for 55 percent of all new US electricity demand. The cluster of AI power headlines is mostly a story about chips and training runs; underneath it sits a more ordinary pattern. Forecast revision cadence has decoupled from build cadence.
The International Energy Agency now projects global data center electricity doubling from 415 terawatt-hours in 2024 to roughly 945 by 2030, growing 12 percent a year, four times faster than any other sector. The Electric Power Research Institute's US range, 9 to 17 percent of US electricity by 2030, more than doubles today's share even on the low end. Revisions land in months; new plants take years to permit.
The reusable frame: any new data center announcement is now a question about whether the local interconnection queue, siting, and rate base can absorb it, not a question about compute. The mechanism sorts who gains and who pays. Utilities with permitted sites and headroom collect the load. Hyperscalers with capital and patience capture the cheapest electrons. Ratepayers in the few states absorbing the build absorb the rest. The next two planning cycles decide whether reliability holds and where new lines and plants go.
Reported by Sky for Type0, from Data centers drive 55% of US electricity demand growth. Read the original: lasvegassun.com