Data Centers Aren’t the Problem. Scarcity Politics Is.

Data Centers Face Growing Scrutiny Over Electricity and Water Demands

Data centers consumed 4 percent of U.S. electricity in 2023, with projections indicating their share could reach between 6.7 percent and 12 percent by 2028 according to multiple industry analyses. These facilities support cloud computing, artificial intelligence training, financial transactions, and other digital services. Some individual sites draw as much as 5 million gallons of water per day for cooling systems.

Local governments in several states have responded with temporary moratoriums on new construction, revised zoning rules, and extended permitting reviews. Officials in Virginia, Georgia, and parts of the Midwest have cited risks of strain on existing power infrastructure and water supplies. Data center operators note that demand growth stems from broader adoption of digital services rather than any single sector.

Electricity systems require continuous balance between generation and consumption. When large loads connect without corresponding increases in supply, wholesale prices can rise sharply during peak periods. Retail rates in many regions remain fixed, which limits price signals that might otherwise encourage shifts in usage timing or efficiency investments. Some utilities have introduced interruptible-service contracts and time-of-use rates for large customers, allowing operators to reduce draw during constrained hours in exchange for lower overall charges.

Proposals to address these dynamics include real-time wholesale pricing for qualifying users, dedicated large-load tariffs, and requirements that new facilities contribute to transmission or substation upgrades. Several states already apply similar cost-recovery mechanisms for industrial projects. Removing regulatory barriers to new generation and transmission lines has also been discussed as a parallel step in utility commission proceedings.

Water consumption varies by cooling technology. Facilities using evaporative systems withdraw more water than those relying on air cooling or closed-loop recycling. Industry reports document a shift toward the latter methods in newer projects, driven in part by operating costs and local permitting conditions. Policies that set performance standards at the property boundary rather than prescribing specific equipment allow operators to select methods that meet those standards.

Land-use decisions involve trade-offs between existing zoning and new infrastructure needs. Property owners who meet published standards for noise, traffic, setbacks, and utility connections have historically received predictable approvals for other industrial uses. Some jurisdictions have applied retroactive changes or discretionary reviews to data center applications after public opposition emerged. Consistent application of published rules, including shot-clock deadlines for permit decisions, has been used in other infrastructure sectors to provide certainty while retaining oversight.

Developers in multiple regions have funded on-site substations or transmission extensions under agreements that allocate costs to the project rather than spreading them across all ratepayers. This approach appears in utility tariffs in Texas and Arizona and has been proposed elsewhere. It leaves residential and smaller commercial customers unaffected by incremental grid investments tied to a single large user.

Historical infrastructure expansions, including highways, power plants, and fiber networks, encountered similar local concerns over resource use and land conversion. Outcomes depended on the combination of technical standards, cost allocation rules, and permitting timelines in place at the time. Current discussions center on whether existing frameworks for electricity and water allocation can accommodate the documented growth in data center demand or require adjustments to pricing, interconnection, and siting processes.

Jen Springman is the Coalitions Manager at the Goldwater Institute.