By Priya Raman

Voters in the 2026 midterms rejected data-center development in 40 of 41 local ballot measures nationwide. Harwood, North Dakota, was the only community to choose differently, rejecting added construction review for a proposed industrial compound. The measures formed a direct check on the tech industry's plans to build data centers for AI (artificial intelligence) across the country.

Why did Harwood break with the national pattern?

Harwood, a small town with 845 residents, had a specific economic prospect before it. Residents hoped to attract development on a 900-acre compound just outside town and draw it away from nearby Fargo. The measure would have required public hearings before large industrial projects, including data centers, could proceed.

The Washington Post reported that around 80% of residents rejected construction-review requirements for data centers. Mayor Blake Hankey told the Washington Post that the proposed facility “would provide $1.65 million in property taxes once complete.” The proposal offered a quantified public return, not merely a general promise of industrial growth.

That distinction mattered. The national result captured residents' objections to what data centers might bring: potential environmental effects, higher electricity costs and noise. Harwood had a named project, a stated location and a tax figure attached to it. Local voters could weigh one concrete proposal rather than a general industry trend.

What did the broader election show?

Exit polls found that 76% of voters did not want data centers in their communities. That sentiment formed one part of an election shaped by industrial growth and its costs. Democrats won control of both chambers of Congress and several governorships; many candidates ran on pro-regulation platforms often centered on the breakneck growth of AI and related affordability concerns.

The political backdrop did not cause every local vote. It gave data-center development a broader place in the election. At the local level, residents used ballot measures and construction rules to place direct control over industrial projects in their own hands. The 40-of-41 result shows how consistently that control favored restraint on development.

Why had the backlash built momentum?

The opposition had been gathering strength before election day. Town hall meetings across the country often became centers of dissent. Residents raised concerns about data centers' potential environmental effects, rising electricity prices and occasional accounts of deafening noise in nearby neighborhoods.

These complaints focused on costs residents could experience directly, rather than on the technology sector in abstract terms. That distinction explains why a fast-growing industry met organized resistance at the ballot box.

It also explains the relevance of Harwood's pro-development case: the town's argument centered on a specific fiscal benefit for local residents, alongside its desire to steer investment away from Fargo.

What did the local decisions require?

The range of measures varied by state:

• Ohio: Voters in 18 communities chose to slow or outright ban data-center development, according to the Ohio Capital Journal.
• California: Voters in various cities backed bans on development, extending local control beyond a single measure.
• Maine: Many communities approved a six-month moratorium, or temporary pause, on all new construction. The restriction reached beyond data centers to construction generally.

The range matters because voters did not endorse one national policy. They chose different degrees of control: public hearings, slower development, bans and temporary pauses. The common thread was resistance to projects arriving before communities had accepted their effects.

What does Harwood's exception mean?

Harwood's vote should not be read as proof that residents broadly favor data centers. It shows the power of a local economic case when opposition elsewhere focused on electricity, environmental impact and noise. A specific tax commitment gave voters a benefit they could compare with the industrial demands of a 900-acre compound.

For developers, the strategic test is therefore straightforward: a proposal needs more than the momentum of the AI industry. It needs a credible answer on power costs, environmental effects and noise, and a benefit that residents can see.

Harwood supplied the second element in the $1.65 million property-tax figure. The other reported cases centered on restrictions rather than a named, tax-backed project.

The national result was clear, but its exception was equally clear: local economics can outweigh generalized resistance when residents see a concrete gain. The exception did not overturn the pattern; it exposed the terms on which local support might emerge.