Commentary. The next phase of the artificial-intelligence debate will not be decided only in laboratories or in Congress. It will be decided in places where residents receive electric bills, attend public hearings and watch industrial-scale data centers rise near their homes.
That is why our newsroom believes the central question is not whether AI should develop. It is whether its infrastructure can expand without shifting its costs onto communities that did not consent to them.
The political signal is already clear. Harvard’s Ash Center reports that 40 of 41 local data-center measures that passed across 10 states in the November 2026 elections favored limiting development, with the restriction side winning more than 70% of the vote in 30 of those contests. The center’s analysis also identifies electricity prices, employment and local oversight as increasingly tangible technology issues.
Those votes should not be dismissed as anti-technology populism. Data centers consume large amounts of electricity and, depending on their cooling systems and location, can place pressure on water supplies and transmission networks. Residents are justified in asking whether promised jobs and tax revenue will compensate for new substations, road construction, environmental impacts or higher utility costs.
Nor is the concern confined to local activists. A draft discussed at the COP31 climate talks flagged worries that rapid data-center construction could increase emissions while intensifying pressure on land, water and minerals. Reuters reported that negotiators remain divided over energy and climate-finance goals, while only nine of nearly 200 countries at the talks had signed an AI-related pledge.
The case for building
The counterargument deserves more than a caricature. Data centers support cloud computing, medical research, financial systems and AI tools that businesses and public agencies increasingly use. Delaying construction could push investment to regions with dirtier power, weaker labor standards or less transparent regulation. It could also leave the United States dependent on foreign computing capacity at a moment when advanced technology is tied to economic and national-security competition.
Companies and their supporters also argue that new facilities can bring construction work, permanent technical jobs and a broader tax base. A blanket ban would sacrifice those benefits without guaranteeing lower bills or cleaner energy. Restricting supply may simply make computing more expensive while doing little to slow demand.
That case is strongest when developers pay the full marginal cost of expansion. It is weakest when public authorities offer tax breaks and expedited approvals while existing customers absorb grid upgrades. A national AI strategy that celebrates investment but treats local infrastructure as someone else’s problem is not industrial policy; it is cost transfer.
A fairer bargain
The answer should be neither unconditional acceleration nor an indefinite moratorium. Congress and state regulators should require transparent, location-specific accounting before approval: projected electricity and water use, transmission upgrades, emissions, tax concessions, backup-generation plans and the number and quality of jobs created.
Large facilities should finance the additional infrastructure they require, rather than relying on residential customers to subsidize growth. A Senate proposal discussed after the midterm elections would require data centers to pay for added electricity infrastructure, according to AZFamily’s report. That principle should be widened to include water systems, emergency services and meaningful community-benefit agreements.
Regulation should also be coherent. The federal government’s push for a lighter-touch approach has collided with state laws and local initiatives, creating what CIO describes as a more complicated compliance landscape rather than genuine deregulation. A national baseline could prevent a race to the bottom, but it should not erase stronger protections where local conditions demand them.
Our view is straightforward: innovation is valuable, but it is not an exemption from democratic consent or basic accounting. Communities should not have to choose between affordable power and participation in the digital economy. If AI companies want public legitimacy, they must show that prosperity will be shared and that the bill will not quietly arrive in someone else’s mailbox.
