Local opposition stalled about $68 billion of US data center projects in one quarter

"It's just so far beyond any bill that I've ever had," John Steinbach told Consumer Reports ^[3]. His January 2026 bill at the Manassas, Virginia, home where he has lived for nearly 40 years was $281, up from roughly $100 the month before ^[2]. That is one household in one month. A month-over-month jump on a single bill cannot separate a rate increase from higher usage.

The firmest market data comes from PJM, the grid operator for 13 states and Washington, D.C., whose capacity auctions pay power plants to be available when demand peaks ^[11]. Its clearing price rose roughly ninefold in a single auction ^[21]. The attribution to data centers comes from the market's independent monitor, and it is the closest thing in this evidence to a measured effect ^[24]. Working back from the monitor's share, the full 2025/26 increase was about $14.8 billion ^[25].

Retail evidence is softer. EIA data show Virginia residential prices up about 13% over 12 months ^[13]. In a January 2026 survey by Global Strategy Group and the Chesapeake Climate Action Network Action Fund, an environmental advocacy group, nearly three-quarters of Virginia voters put the blame on data centers ^[14]. A survey like that measures what voters believe caused the rise.

Each side cites a study, and the two measure different quantities. The Electric Power Research Institute looked at 2015 to 2024 and found average retail prices 3.5% lower for every doubling of data center capacity ^[1]. Its explanation is that large, steady loads spread a utility's fixed costs across more sales ^[1]. A peer-reviewed model by researchers at NC State, Carnegie Mellon and two other universities puts national wholesale prices in 2030 at 6% to 29% above where they would be without data center growth ^[15]. In the hardest-hit regions the projection reaches 57% ^[15]. One result is a correlation in past retail prices. The other is a forecast of future wholesale prices.

I think both can hold. AI clusters draw power around the clock during training runs ^[16]. A load like that fits EPRI's explanation, and it also adds to the peak demand that capacity auctions pay plants to cover ^[11].

The political delay sits on top of a physical one. A Lawrence Berkeley National Laboratory report prepared for the Department of Energy puts US data center consumption in 2023 at about 176 terawatt-hours, or roughly 4.4% of all the electricity the country used ^[20]. Bloom Energy's January 2026 report projects data center IT load rising from about 80 gigawatts in 2025 to about 150 gigawatts in 2028 ^[17], an increase of about 88% ^[23]. Bloom sells on-site power systems, so the forecast comes from a company with a stake in the answer ^[17]. In the same report, developers expect about a third of data centers to rely solely on power generated on site by 2030 ^[19]. On-site generation shortens the wait for utility power. It does not help in a jurisdiction that has barred data centers outright ^[7].

Data Center Watch counted about 120 projects disrupted by local opposition in the first half of 2026 ^[5]. The stalled second-quarter projects average roughly $1.5 billion each ^[22]. The figures as reported do not include how many projects were proposed over the same months, so the share of the pipeline at risk cannot be worked out from them. Indianapolis voted 23-1 in August to bar new construction through 2027 ^[8]. Charlotte paused approvals for 150 days after a city survey found 78% opposition ^[9], and Brookings reports that candidates in both parties now run ads against the facilities ^[10].