Every rate case with a data center in it eventually turns into the same argument. Are rising bills about demand, or about who’s serving it? The evidence says it’s the second one, and utilities that keep answering the first question are going to keep losing the hearing room. Virginia just gave the industry two new ways to answer the second one, in the same year.
Enverus Intelligence® Research (EIR) models Lower 48 load rising roughly 14% by 2035. That number gets read as “demand is the problem.” But almost none of that growth is organic. Organic demand is climbing about 0.5% a year. The other 13.5 points are data centers, EVs, and electrification, concentrated in a handful of markets and mostly loaded past 2030. Demand is rising fast, in the places least prepared to absorb it.
Mid-Atlantic and Northeast residential prices rose by double digits over the past year, led by a 23% jump in Washington, DC. PJM wholesale prices climbed roughly 76% over the same stretch, with the sharpest increases landing in the markets where large loads are concentrating. Virginia, the dominant data center market in PJM, is set to absorb much of the region’s projected 17 GW of new capacity through 2030.
But if demand itself were the problem, the answer is more supply, and the bill lands on everyone equally. If the problem is how a narrow slice of demand gets served and paid for, the answer is a cost-allocation tool, and Virginia just ran two of them at once.
Virginia’s Two Cost-Allocation Tools
Effective July 1, Virginia became the first state to tax data center electricity consumption directly: $0.011 per kilowatt-hour, capped at $600 million a year, with any overage refunded to operators. On the state’s existing 6 GW of capacity, that pencils out to roughly $290 million a year. Virginia kept its sales tax exemption on capital equipment intact, worth an estimated $1.9 billion annually, so the new tax targets operating cost, not the incentive to build. Amazon, which EIR identifies as accounting for all of the state’s high-confidence grid additions, carries most of that exposure.
Separately, Dominion’s GS-5 rate class, effective Jan. 2027, applies to any data center drawing more than 25 MW. Customers pay for 85% of contracted transmission and distribution capacity and 60% of contracted generation, whether they use it or not. One taxes what gets used. The other taxes what gets reserved. They are not the same instrument, and they don’t fail the same way.
| Consumption tax Bills what gets used | GS-5 rate class Bills what gets reserved | |
|---|---|---|
| Basis | Metered consumption | Contracted capacity |
| Rate | $0.011 / kWh | 85% of contracted T&D + 60% of contracted generation |
| Effective date | July 1, 2026 | Jan. 2027 |
| Applies to | All data center load in Virginia | Data centers over 25 MW (Dominion) |
| Revenue / cost impact | ~$290M/yr on existing base; capped at $600M/yr, overages refunded | Bills reserved capacity whether used or not |
Which Tool Actually Holds Up
We told a version of this story in our e-book, Planning You Can Defend where AEP Ohio built a data center tariff that filtered speculative load out of its interconnection queue, and connection requests dropped roughly 50% within months. That’s a queue-filtering tool built inside a rate case. Virginia’s approach pairs a consumption tax with a separate capacity charge, aimed at cost recovery and reserved-capacity discipline rather than queue control. Same pressure but different mechanism, different state, different failure mode.
Get the mechanism wrong, and the load doesn’t just sit in the queue, it leaves. EIR projects roughly 40% of new U.S. data center capacity through 2030 will be built behind the meter on dedicated gas generation, partly to avoid congested queues and rate classes like GS-5.
For any vertically-integrated IOU watching a data-center cost-allocation fight head toward its own rate case, the question isn’t whether to act. It’s which mechanism actually holds up. A consumption tax is simple to explain to ratepayers but does nothing about speculative capacity sitting idle in the queue. A capacity-based rate class filters the queue but is harder to defend if a developer argues they’re being billed for power they never drew. Pick the wrong one and you’ve built your next intervenor’s opening argument for them.
Is Your State Next?
Virginia won’t be the last state to try a cost-allocation fix, and it’s already running two at once. If you’re heading into a rate case where data centers are the flashpoint, our analyst team can benchmark your state’s approach against what’s actually holding up in Virginia, Ohio, and the other states EIR is tracking, before you’re the one defending it in front
of a PUC.
About Enverus Intelligence® | Research, Inc. (EIR)
Enverus Intelligence® | Research, Inc. (EIR) is a subsidiary of Enverus that publishes energy-sector research focused on the oil, natural gas, power and renewable industries. EIR publishes reports including asset and company valuations, resource assessments, technical evaluations, and macroeconomic forecasts and helps make intelligent connections for energy industry participants, service companies, and capital providers worldwide. See additional disclosures here.
