Scholars propose ways to manage data centers’ demands on electricity, water, and local communities.
Growing consumer and business demand for artificial intelligence (AI) services is driving a rapid expansion of data center infrastructure across the United States. Data centers represented approximately 4.4 percent of U.S. electricity usage in 2023, according to the U.S. Department of Energy, which estimates that data centers could make up between 6.7 percent and 12 percent of U.S. electricity demand by 2028. The Energy Department has identified AI applications as an important driver of this projected growth, and noted that individual large-scale data centers can require hundreds of megawatts of electricity.
The anticipated growth has created regulatory tension between accelerating data center construction and managing its demands on public infrastructure. In July 2025, President Donald J. Trump issued Executive Order 14318, directing federal agencies to accelerate permitting for data centers and related energy infrastructure. Federal agencies must now expedite environmental reviews, open federal lands to development, and commit other federal resources to these projects.
At the same time, federal energy regulators have begun addressing how rapidly growing data center loads should connect to the electric grid. In June 2026, the Federal Energy Regulatory Commission (FERC) directed all six regional grid operators within its jurisdiction to justify or reform their rules governing data centers and other large electricity users. Through these orders, FERC hopes to address the need to integrate large loads more quickly while maintaining grid reliability and protecting existing electricity customers from costs associated with serving new loads. FERC had previously begun examining the issue in 2025, when it launched a proceeding to consider whether data centers could share power plants with electricity generators without compromising grid reliability or raising costs for consumers.
Many regulatory questions surrounding data centers, however, extend beyond federal electricity regulation. State and local governments make decisions concerning land use, water resources, utility regulation, environmental review, and economic development. Recent scholarship has examined how data center development can strain local infrastructure and natural resources and has questioned whether existing regulatory institutions are equipped to evaluate projects that can impose substantial demands on local resources.
The resulting policy questions extend beyond whether data centers should be built. Regulators must determine how the costs of new electricity and other infrastructure should be allocated, what information developers should publicly disclose about resource use, and how governments should balance the benefits of additional computing capacity against effects on infrastructure and surrounding communities.
In this week’s Saturday Seminar, scholars propose ways to regulate the growing demands of AI data centers on electricity, water, and local communities.
- In an article in the George Washington Law Review, Alexandra B. Klass, of the University of Michigan Law School, and Dave Owen, of UC Law San Francisco, argue that utilities should not have to guarantee uninterrupted electricity service to new large data centers. Drawing on natural gas regulation and water law, they propose data centers be allowed to connect to the grid immediately if they accept curtailment during peak demand. They note Minnesota, Texas, and FERC’s directives to a major transmission organization already point toward such policies. Klass and Owen contend this approach would reduce overbuilding, shift supply-demand risk onto sophisticated operators rather than ratepayers, and spur innovation, though legal complexity could limit its effectiveness.
- In an article in AGU Advances, Ana Pinheiro Privette, Ana Barros, and Ximing Cai, of the University of Illinois’s Center for Secure Water, argue that there is too little transparency about data center water use, undermining regulation and planning. They note that data centers rarely disclose facility-level water data, few operators track basic efficiency metrics, and local governments often sign nondisclosure agreements tied to incentive packages. Privette, Barros and Cai cite Google’s Council Bluffs, Iowa facility, which made up about 21 percent of the city’s 2023 water use, as evidence of the severity of local impacts. They recommend standardizing water reporting, incentivizing low-water cooling, integrating water availability into site selection, and requiring disclosure of water commitments in incentive agreements.
- In a recent article in the Journal of the American Planning Association, Justin Kollar, a Ph.D. candidate at the Massachusetts Institute of Technology, explains that recent pieces of state legislation have limited local governments’ authority over land use and zoning. Although these legislative developments have emerged in response to public controversies regarding AI data center construction, Kollar finds that the legislation preempting local land authorities can come in multiple forms, including direct statutory overrides, federal intervention, and procedural changes. He warns that this deliberate power-shifting may fulfill state industrial goals while ignoring community interests. He asks public planners to consider local concerns when approving or denying projects.
- In a report for Harvard Law School’s Electricity Law Initiative, Ari Peskoe and Eliza Martin, of Harvard Law School, argue that utility ratepayers are being forced to subsidize data center power costs. They find that utilities offer secret special contracts to data centers, connect data centers directly to existing power plants, and exploit gaps in federal and state regulation in order to shift data center energy costs onto ordinary consumers’ electricity bills. Martin and Peskoe warn that the secrecy of utility rate-setting makes it nearly impossible to detect these cost shifts. They recommend stronger oversight of utility contracts with data centers and new rules to protect ratepayers.
- In a working paper, Benjamin Jaros, of Stanford University’s Hoover Institution, examines the impacts of data center tax breaks on local economies. Studying 135 tax break agreements across 44 counties in 20 states, Jaros finds that the most significant impact is in local government spending, which rose by about $200 per person in the year after a data center opened. He concludes that this extra spending did not come from higher tax collections. Instead, he finds that counties borrowed more money—about 5.5 percent more in the first year—suggesting they went into debt to cover the costs of roads, utilities, and other services the new data centers required.
- In an article in Oxford Open Energy, Corey Young, of Washington & Jefferson College, argues that data centers should be treated like power plants or water systems—major infrastructures that need real government oversight. Young argues that case-by-case review can obscure the cumulative effects of data center development on regional infrastructure and resources. He shows that data centers put heavy strain on local power grids, water supplies, land, and city budgets. He warns that data center companies often share little information about their operations, leaving local officials without the knowledge or tools to manage these impacts. Young calls for governments to plan ahead, require transparency from operators, and coordinate regulation across city and state levels.
