The North American Electric Reliability Corporation’s projection for summer peak demand growth over the next decade has gone from 55 gigawatts in 2022 to 80 gigawatts in 2023 to 132 gigawatts in 2024 to 224 gigawatts in 2025, the steepest year-over-year jump in the compound annual growth rates NERC has tracked since it began in 1995. That escalating revision pattern, more than any single forecast, is the clearest evidence that the timing gap between hyperscaler buildout speed and utility grid planning cycles is widening rather than stabilizing.
Rob Gramlich’s description of a “mismatch in timing” between how fast technology companies can finance data centers and how slowly utilities can plan new generation and transmission understates how quickly that mismatch is compounding. NERC’s 2025 Long Term Reliability Assessment, released in January, projects summer peak demand across North America will grow by 224 gigawatts over the next ten years, more than 69% above the 132 gigawatts projected in the prior year’s assessment and a 24% increase over actual 2025 peak demand. Winter peak demand is projected to grow even faster, by 245 gigawatts, up 65% from the previous year’s projection. NERC’s director of reliability assessments, John Moura, described the finding plainly: “The system is changing faster than the infrastructure needed to support it.” Five regions, MISO, PJM, ERCOT, and the WECC-Basin and WECC-Northwest zones of the Western grid, now carry NERC’s “high reliability risk” classification for the years leading up to 2030, meaning planned resources are projected to fall short of adequacy criteria within that window rather than merely under stress in extreme conditions.
The scale of data centers’ contribution to that demand growth is real, though the specific figures in circulation understate how much they have already been revised upward. Lawrence Berkeley National Laboratory’s original December 2024 report estimated data centers would consume between 6.7% and 12% of total US electricity by 2028, up from 4.4% in 2023. An updated LBNL analysis released in 2025 pushed that range higher and extended it to 2030, projecting data centers could account for between 9.5% and 15.3% of US electricity by then, up from a new 2024 baseline of 4.7%, with researchers separately estimating data centers will constitute more than 40% of all US electricity demand growth over the next five years. A commonly cited figure describing data centers reaching “nearly 12%” of electricity by 2030 sits closer to the midpoint of that updated range, around 11.8% by one LBNL estimate, but presenting a single point figure obscures that the same research now puts the upper bound above 15%, a meaningfully larger number than the round figure suggests.
That accelerating demand is landing on a grid that had already fallen out of practice building new capacity. Gramlich, president of Grid Strategies, traced the constraint to a 25-year lull: major transmission and generation expansions in the 1980s and 1990s left utilities with surplus capacity that absorbed demand growth through the early 2000s, a period when electricity consumption grew far more slowly than in the decades when air conditioning and household appliances first proliferated. “But now we have electric vehicles, we have electric space heating, we have new manufacturing, much of which is using electricity more than traditional manufacturing, and then we have data centers,” Gramlich said, estimating that data centers account for roughly half of the new demand growth, with electrification of transport, heating and industry making up the other half arriving on the same timeline.
The practical consequences of that mismatch are already visible in specific bottlenecks rather than remaining a future risk. Lead times for large power transformers, the equipment most directly required to expand grid capacity for new data centers, have stretched to as long as three years in the United States, up from what was historically a matter of weeks. PJM’s own projected capacity shortfall widened from roughly 209 megawatts in its 2026/2027 base residual auction to more than 6.5 gigawatts projected for 2027/2028, and the Department of Energy invoked wartime emergency authority three separate times in 2026 to force data centers within PJM’s territory onto diesel backup generation during grid stress events, a power the department had used only about 26 times combined across the entire country in the 25 years before this one.
The bill impact behind community pushback is also more quantifiable than general concern over “higher electric bills” suggests. Average US residential retail electricity prices tracked roughly with the overall inflation rate from 2019 through 2024, according to LBNL’s updated analysis, before rising 6% in nominal terms in 2025, more than twice the rate of inflation that year. Investor-owned utilities sought a combined $18 billion in rate increases in 2025, the largest total since the mid-1980s. That pattern lines up with survey data on the political backlash the source material references: a Gallup poll found 71% of Americans oppose data center construction in their own area, a higher opposition rate than Americans express toward nuclear power plants, and a separate Reuters poll found 77% of respondents specifically fear data centers will drive up their home electricity costs. Those figures give Gramlich’s “real disconnect” a measurable political dimension: the same demand growth NERC shows accelerating year over year, and the same capacity shortfalls already forcing emergency federal intervention in PJM, are arriving at the same time utilities are seeking rate increases at a pace not seen in four decades, a combination that has moved data center siting from a technical planning question into an active issue in state-level elections.
