Virginia's answer to the AI power crunch: keep the clean-energy mandate, and bill the data centers

Virginia Gov. Abigail Spanberger released a state energy plan on Oct. 1, 2026 that tries to thread a needle most states have avoided: keep the 2020 Virginia Clean Economy Act (VCEA), let the data-center boom run, and put the bill for new power plants on the data centers rather than on households. The blueprint is the first U.S. governor's energy plan to use advanced planning software to model multiple supply scenarios, and it is also the first to publish a "no clean goals" counterfactual that critics can cite. Both halves of that ledger are now in play.

The plan's headline number is 85%: the projected growth in Virginia's electricity demand by 2050 under a "moderate" scenario that assumes most, but not all, proposed data centers come to fruition. Across the four modeled pathways, system costs run as high as $422 billion through 2050, of which $265 billion is projected to come from cost allocation to data-center customers. Every pathway requires 1.2 to 1.8 gigawatts of new solar per year. The plan states the trade-off in plain language: "These costs appear high because the scale of growth Virginia faces is enormous, but the modeling shows that the problem is solvable. Further, these results suggest that with proper cost allocation, Virginia can serve new data center growth with clean energy solutions."

Three of the four pathways layer a demand-flexibility regime on top of the supply buildout: data centers agree to cut draw at peak times, leaning on batteries, on-site clean energy, or behind-the-meter generation, so the grid is not built to the size of their worst hour. Spanberger's chief energy officer, Josephus Allmond, framed the modeling as a rebuttal to the idea that VCEA and data-center growth are mathematically incompatible.

The contrast with other states is the point. Texas and New York governors have moved toward moratoria on new data-center hookups while they sort out rate impacts. Virginia's blueprint offers a third path: instead of pausing the buildout, change the tariff and contract terms so the new load pays for the new wires and the new gas, and let data centers act as flexible load that drops off when the grid is tight.

The same document, however, contains the numbers the plan's opponents want. The fifth modeled scenario strips out the clean-energy mandate and saves at least $90 billion in electricity system costs through 2050, at a price of $145 billion in modeled health impacts. Republican legislators in Richmond have already cited the $90 billion figure to push for VCEA repeal and a withdrawal from the Regional Greenhouse Gas Initiative (RGGI), a multistate cap-and-trade compact. The plan thus lands as ammunition for both sides, and Spanberger's team has chosen to publish the counterfactual rather than bury it.

Two caveats frame the exercise. The 85% demand growth is a model assumption, not a contracted outcome, and PJM's 2026 long-term load forecast has already marked down Dominion-zone summer peak growth to 5.4% per year from 6.3% a year earlier. PJM data show the Dominion zone hit a summer peak of 23,905 megawatts in 2025, 23% above 2019, and a winter peak of 25,413 megawatts in the 2025-26 season, 45% above 2019-20, with EIA reporting that Virginia commercial electricity sales rose nearly 30 million megawatt-hours from 2019 to 2025, second only to Texas. Second, the plan is not law. It informs the State Corporation Commission and the General Assembly as they weigh Dominion's pending large-load filing (Case No. PUR-2026-00011) and a suite of data-center tariff reforms. The who-pays machinery still has to be built in dockets, not declared in a press release.

What to watch next: how Dominion's SCC filing treats the cost-allocation assumption, whether any data-center operator signs a demand-flexibility contract at scale, and whether the General Assembly uses the $90 billion counterfactual to reopen VCEA before the model assumptions are stress-tested against actual filings.