Virginia’s tax landscape for hyperscale data centers is evolving. After months of budget negotiations, lawmakers preserved the Commonwealth’s sales and use tax exemption for certain data center equipment and simultaneously imposed a first-in-the-nation excise on data-center electricity consumption. This legislative compromise leaves in place the up-front incentives to build data centers in the state, while also asking the industry to contribute directly to nonlinear energy costs.

What’s notable about this compromise? It’s essentially what industry lobbyists asked for. Below the fold: the state-level legislative deal, plus another layer of local action. What’s behind data center growth in the United States? It’s federalism all the way down.

Virginia adopted data center tax incentives almost twenty years ago, and, over that period, the state became the world’s largest hub for those facilities. A nationwide pushback against burgeoning AI infrastructure, however, threatened Virginia’s longstanding tax incentives—even as (or perhaps because) the region saw $80 billion of data center investment over the last two years. After the Virginia Senate voted to terminate the tax expenditure early, the incentive’s future looked uncertain.

Then, in late June, a special legislative session yielded a compromise. Virginia lawmakers approved a budget that maintained data centers’ sales and use tax exemption through 2035 and also created a temporary $0.011 per kilowatt-hour industry-specific surtax on electricity consumed at data centers, capped at $600 million annually for two years—widely described as the first such tax in the nation. This agreement emerged from industry lobbying, legislative bargaining, and increasingly organized public opposition to data center expansion. In a highly negative environment for data center incentives and after a prolonged budget stalemate, this type of political deal seems to work.

More broadly, the sales and use tax exemption matters because AI data centers turn over equipment quickly, and the electricity surtax imposes a Pigouvian instrument that reflects the third-party (and regionally concentrated) harms that AI data centers may bring. The exemption loses roughly three times the maximum revenue that can be raised by the surtax, which makes clear that the provisions are not pure offsets. Moreover, the surtax applies to self-generated electricity as well as purchased power. Essentially, Virginia decided to exempt certain business inputs from the tax base while taxing a reasonable proxy for asserted externalities from the operation of these inputs.

Until now, much of the state-level data center debate has asked whether established tax incentives should be curtailed, and by how much. The compromise in Virginia suggests another model: retain ex ante investment incentives (including for replacement property), and impose ex post operational surtaxes to recoup additional costs. Although the Virginia electricity surtax was targeted, one could imagine broader instruments aimed at incremental water and environmental costs. Whatever the quantitative merits of Virginia’s specific deal, the overall package strikes a potentially meaningful balance between capturing the benefits of AI infrastructure and allocating the costs of AI development.

If other states begin pairing locational subsidies with operational surtaxes, local politics may matter more in siting hyperscale data center projects. In Virginia, a Blackstone portfolio company recently terminated its portion of the Prince William Digital Gateway project—a proposed 2,100-acre data center campus near the Manassas National Battlefield—over organized local resistance and a procedural defect involving rezonings. Barriers to entry arrive in many forms and at different levels. This dynamic reveals a second emerging reality: even industry-friendly tax incentive regimes may not overcome dedicated local opposition to large-scale AI infrastructure.

Taken as a whole, these developments show greater nuance in how public and private interests are approaching hyperscale data center development. Although generative AI’s implementations tend toward abstract interfaces on screens, the AI revolution’s reality is grounded in tangible assets and the nuts and bolts (or racks and circuit boards) of real estate law, state tax law, and local regulations. The questions are bigger than just the level of incentives for data center development. The more important debate involves choice of base, timing, and the relative pricing of investment and resource consumption—and perhaps, in the future, the specific use of funds raised through AI taxes.

Related TaxProf Blog coverage:

• ⁠Checking in on Data Taxes and Related Reforms (May 4, 2026)
• States Seek to End Data Center Tax Breaks (Jan. 19, 2026)
• ⁠States Continue to Decouple From OBBBA (Jan. 12, 2026)
• Data Center Development to Impact Income Tax Bases (Nov. 24, 2025)