Commentary. The most consequential AI story in the United States may not be a new model, a spectacular product launch or a warning from a laboratory executive. It may be the growing refusal of communities to host the infrastructure behind the industry.
In the November 2026 elections, voters in at least 10 states considered measures aimed at data-center development, including moratoriums and zoning restrictions. The results reflected a broad unease: more than 70 percent of Americans reportedly oppose data-center construction in their area, while more than 80 percent believe technology is advancing too quickly. The Atlantic reported that several local contests became referendums on whether elected officials had put technology companies ahead of residents.
Our view is straightforward: communities should not be expected to subsidize AI’s expansion while companies collect most of the gains. But local anger should produce stronger rules, not reflexive opposition to every facility.
The case against business as usual
Large data centers consume enormous amounts of electricity and can strain local grids. Residents also face noise, construction disruption, water-use concerns and the possibility of higher utility bills. Those costs are immediate and visible; promised benefits, such as jobs and tax revenue, are often less certain or unevenly distributed.
That imbalance explains why opposition has crossed party lines. The issue is not simply hostility to artificial intelligence. Reporting on recent campaigns found that much of the backlash focused on practical consequences—electricity prices, noise and industrial land use—rather than on AI as a technology. Tom’s Hardware
Companies have too often treated public consent as a communications exercise. A glossy promise of innovation cannot substitute for transparent estimates of power demand, water consumption, tax incentives and permanent employment. Nor should residents have to discover after approval that a project will require grid upgrades financed by ordinary ratepayers.
Why blanket bans would be a mistake
Still, a moratorium is an instrument, not a strategy. A permanent or poorly designed ban could push investment to jurisdictions with weaker oversight, forfeit useful economic opportunities and make it harder to build the computing capacity needed for scientific research, medical applications and public services.
There is also a danger in treating every data center as identical. A facility powered largely by new renewable generation, located away from stressed residential grids and required to provide waste-heat recovery is not equivalent to one that relies on scarce local capacity while receiving generous public subsidies. Regulation should distinguish between them.
The strongest argument from the technology industry is therefore not that communities should remain silent. It is that predictable, national standards are preferable to a patchwork of local vetoes. Companies need certainty before committing billions of dollars, and fragmented rules can slow infrastructure without guaranteeing better outcomes.
A better bargain
The answer is a public-interest compact with enforceable conditions. Developers should pay the full marginal cost of grid connections and upgrades, disclose energy and water use, meet strict noise standards and demonstrate that projects will not raise household utility bills. Tax incentives should be conditional on verified jobs, local procurement and community benefits—not merely on construction spending.
Regulators should also require meaningful consultation before permits are granted. That means publishing contracts and environmental assessments in accessible language, allowing independent review and giving affected communities a genuine opportunity to challenge assumptions. Voluntary promises are useful, but they are not enough.
The White House has acknowledged public concern and sought commitments from hyperscalers, utilities and governors through a “ratepayer protection pledge.” Reporting on the pledge suggests that the administration understands the political risk. Yet pledges cannot replace legislation, especially when the companies benefiting from expansion are powerful enough to shape the rules.
AI will continue to develop, whether communities welcome it or not. The democratic question is who pays for that development and who gets to decide where its physical footprint lands. Residents are right to reject a model in which private firms claim the upside while the public absorbs the disruption.
But the goal should not be to freeze progress. It should be to make progress earn its place. If the industry wants public support, it must accept public obligations—and prove, in dollars, megawatts and measurable local benefits, that its future is not being built on someone else’s bill.