Communities across the United States are pushing back against the rapid expansion of AI-powered data centers, which are consuming electricity and water at rates that strain local grids, deplete aquifers, and threaten to raise utility bills for ordinary residents. Federal data shows the problem is accelerating: data center electricity consumption tripled between 2014 and 2023, and projections suggest it could triple again by 2028. From small towns in Washington state forced to build new wastewater systems to Georgia ratepayers facing billions in long-term costs, the friction between Big Tech’s ambitions and finite local resources is intensifying.
Electricity Demand Is Outpacing the Grid
The scale of the energy surge is difficult to overstate. Data centers consumed 176 terawatt-hours of U.S. electricity in 2023, up from 58 TWh in 2014, according to a Department of Energy analysis evaluating the increase in electricity demand from these facilities. That 176 TWh figure represented approximately 4.4% of total national electricity consumption in 2023. By 2028, the DOE projects data centers will account for between 6.7% and 12% of U.S. electricity, with consumption reaching a range of 325 TWh to 580 TWh.
Those numbers carry direct consequences for households and businesses that share the same power infrastructure. When a single data center campus can draw as much electricity as a mid-sized city, utilities must build new generation and transmission capacity. Someone pays for that construction, and in regulated utility markets, the cost typically flows through to ratepayers.
The investment needs are beginning to reshape utility planning nationwide. Developers seeking to interconnect large loads are increasingly directed toward regions where grid capacity and fuel supply can scale. Federal tools such as the Department of Energy’s Genesis platform and the research holdings available through the Office of Scientific and Technical Information are being used by planners and policymakers to understand where new generation, transmission, and efficiency investments might be most effective. But even with better data, the timeline for building large power plants and high-voltage lines is measured in years, while AI demand is rising much faster.
Georgia Ratepayers Face Billions in New Costs
That dynamic is already playing out in Georgia, where the Public Service Commission approved a $16.3 billion Georgia Power expansion driven heavily by data center demand. The construction price tag alone is staggering, but critics of the plan have pointed to long-run customer cost estimates of $50 billion to $60 billion when financing, fuel, and maintenance are factored in. The central question Georgia regulators faced, and one that utilities nationwide will confront, is whether residential customers should subsidize infrastructure built primarily to serve tech companies whose facilities generate relatively few local jobs per megawatt consumed.
Most coverage of data center growth focuses on aggregate national statistics, but the real burden falls unevenly. Rural counties and smaller metro areas often lack the regulatory sophistication or political leverage to negotiate favorable terms with tech giants. Local officials may welcome the tax base and construction activity, yet find themselves locked into long-term power contracts that socialize risk. The result is a growing gap: communities absorb the environmental and financial costs of powering AI while the economic benefits concentrate in corporate headquarters and shareholder returns elsewhere.
Some regulators and advocates are pushing for new cost-allocation rules. Ideas include higher demand charges for ultra-large users, dedicated “behind-the-meter” renewable projects that limit stress on shared infrastructure, and performance conditions that tie incentives to measurable community benefits. But for now, most states are still operating under frameworks designed for traditional industrial customers, not clusters of AI servers that can ramp from zero to hundreds of megawatts in a few years.
Water Stress from Cooling Systems
Electricity is only half the resource equation. Data centers require enormous volumes of water for cooling, and the wastewater they produce can damage local water systems. In Quincy, Washington, a small agricultural city that became a major data center hub, cooling blowdown from facilities contributed high levels of total dissolved solids that created compliance concerns for groundwater quality limits. The contamination risk was serious enough to prompt regulatory pressure and force the construction of a separate industrial wastewater treatment and reuse system, effectively requiring the town to build new infrastructure to handle pollution it did not create.
Quincy’s experience illustrates a pattern that is repeating in arid regions where water scarcity is already acute. In Arizona, a proposed data center prompted the city of Tucson to pass an ordinance regulating large water users and requiring conservation measures. That ordinance includes public record requirements, city council approval for major water consumers, and conditions under which water service can be suspended. The nearby town of Marana went further, imposing a potable-water prohibition for such facilities. A University of Arizona water researcher quoted in coverage of the debate raised concerns about the long-term sustainability of allowing industrial-scale water drawdowns in a desert basin already under stress.
Water-intensive cooling also collides with climate resilience planning. As hotter summers drive up residential air-conditioning loads, the same heat waves can push data centers to consume more water to maintain safe operating temperatures. Municipalities that once treated water and power planning as separate issues are now being forced to consider how simultaneous spikes in both will affect public health, agricultural users, and rate stability.
Grid Buildouts Trigger Local Hearings
Northern Virginia’s data center corridor, the densest concentration of such facilities in the world, offers a case study in how grid expansion creates its own backlash. The Virginia State Corporation Commission scheduled a local hearing in Loudoun County on transmission projects tied to data center growth, covering the Aspen-Golden and Apollo-Twin Creeks lines under case numbers PUR-2024-00032 and PUR-2024-00044. These hearings give residents a formal channel to contest route selections and demand accountability for the visual, environmental, and property-value impacts of high-voltage transmission infrastructure cutting through their neighborhoods.
The permitting friction in Virginia is not an isolated case. A commercial real estate analysis showed that data center construction slowed late last year amid permitting issues nationally. That slowdown signals something important: community resistance is not merely symbolic. It is producing measurable delays in the buildout pipeline, which in turn raises costs for developers and forces tech companies to reconsider site selection strategies.
Federal efforts to modernize and expand transmission are beginning to intersect with this local pushback. Programs highlighted on the Department of Energy’s infrastructure exchange portal are channeling funding toward grid upgrades, storage, and resilience. Yet even with federal dollars on the table, state commissions and county boards retain substantial control over siting, ensuring that community concerns about landscapes, noise, and property rights will continue to shape where AI-era infrastructure can be built.
Municipal Water Agreements Under Pressure
Beyond electricity and wastewater, the legal frameworks governing water supply are being tested. In Wisconsin, the City of Racine’s water agreements with the neighboring village of Mount Pleasant, including a 2018 contract, define how water capacity is shared and paid for across municipal lines. Those arrangements were crafted with large industrial users in mind, but the sheer, continuous demand of modern data centers raises new questions about how much capacity a single customer should be allowed to reserve, and who bears the cost if long-term demand projections prove overly optimistic.
As more communities court or confront data center proposals, lawyers and planners are revisiting clauses on minimum purchase requirements, rate structures, and termination penalties. A municipality that overcommits water or sewer capacity to a single project risks crowding out future residential growth or smaller businesses. Conversely, if local governments insist on strict caps and flexible terms, they may find themselves passed over in favor of jurisdictions willing to shoulder more risk.
Some local leaders argue that the only sustainable path forward is to condition approvals on robust conservation and reuse commitments. That can include on-site recycled water systems, seasonal operating limits, or requirements that data centers fund upgrades to municipal infrastructure well beyond the boundaries of their campuses. Others are exploring regional compacts to prevent a race to the bottom in which neighboring towns underbid one another on water and power rates to lure the same set of corporate tenants.
Who Pays for the AI Boom?
The rapid expansion of AI computing has turned data centers into one of the most resource-intensive land uses in the modern economy. The benefits of that computing, from more efficient logistics to new medical tools, are diffuse and global. The burdens, by contrast, are sharply local (higher utility bills, contested transmission corridors, stressed aquifers, and complex legal disputes over who controls and pays for shared infrastructure).
Communities are beginning to demand a different bargain. Instead of accepting that growth in digital services must automatically translate into unchecked growth in physical resource use, residents, regulators, and some utilities are pressing for stricter conditions and clearer accountability. Whether through rate design that better reflects the true cost of serving massive loads, water agreements that prioritize long-term resilience, or public hearings that give neighbors a say in how and where lines are built, the politics of AI infrastructure are moving from distant server racks into local council chambers.
How those debates are resolved will determine not only where the next wave of data centers is built, but also who ultimately pays the bill for the AI revolution.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
