Voters in Trenton, Ohio, backed a charter amendment prohibiting construction of large data centers in the November 3, 2026, election, according to unofficial county returns. About 80% supported the measure, providing a documented local result amid broader reports of voter support for restrictions on computing infrastructure. (Butler County Board of Elections)

The result establishes support for a specific construction prohibition, not a nationwide rejection of artificial intelligence. The county’s ballot list identifies the amendment as a data-center construction measure; its election-night returns do not establish effects on individual projects, operating facilities or national computing capacity. (Butler County Board of Elections ballot list)

What the Trenton vote establishes

The Butler County Board of Elections’ posted returns show roughly four votes in favor for every vote against. These are election-night results, not certified returns. The county’s ballot inventory separately identifies the question as a proposed change to Trenton’s charter, rather than an advisory expression of public opinion. (Butler County Board of Elections)

Proposed amendment language filed in the state court record describes a prohibition on construction of covered data centers. It includes a power threshold above 25 megawatts. That document supports the broad description of a large-data-center construction ban, but it is not a substitute for confirming the final operative text and its implementation. (Supreme Court of Ohio public case docket)

This article has not established an effective date, enforcement arrangements or the treatment of previously approved and already-started projects. It therefore does not characterize the vote as an order to close operating facilities or halt a particular development. Those distinctions matter before translating a local election result into a forecast of lost AI capacity.

The proposed wording also addresses data centers, not generative AI alone. Its definition concerns facilities used for digital information services. On that basis, treating the entire restriction as a measure of computing capacity dedicated to AI would go beyond what the text establishes. (Supreme Court of Ohio public case docket)

Why a national total needs caution

Ars Technica reported that 38 of 44 restrictive measures tracked across 10 states were passing, attributing the tally to Redbud Consulting. That remains an attributed reporting snapshot, not an independently reconciled national result. This article does not adopt it as a count of newly enacted bans. (Ars Technica)

Redbud’s published methodology helps explain why aggregation requires care. Its inventory includes bans, power caps, zoning referendums, developer agreements, voter-approval requirements and advisory questions. It classifies each answer by whether it restricts or allows development, because a “no” vote can represent opposition to a proposed project. (Redbud Consulting, Data Centers on the Ballot)

The tracker also explicitly distinguishes its own completion labels from certification. It says none of its results is certified and warns that its counts can lag election offices. A tally combining these different questions cannot, by itself, show how many binding prohibitions took effect or how much construction they cover. (Redbud Consulting, Data Centers on the Ballot)

Campaign commitments belong in a separate category again. A promise to seek a construction pause is not evidence that a pause has been enacted. This article consequently assigns no legal effect to candidates’ campaign positions and makes no estimate of capacity affected by their election.

Industry interpretation is not a finding about voters

In a November 4 commentary, Chamber of Progress founder and chief executive Adam Kovacevich argued that the election reflected dissatisfaction with Donald Trump and the economy rather than an anti-technology mandate. He also criticized temporary data-center bans as postponing the work of developing policies that benefit communities. Those are the industry-policy representative’s interpretations, not findings adopted here about voters’ motives. (Chamber of Progress)

Local limits meet existing infrastructure pressures

The election debate follows other local government action. On June 9, Seattle’s City Council announced that it had adopted a temporary freeze on new large-data-center siting while the city studied potential impacts. That was a council action, not a November ballot outcome, and should not be added to this election’s results. (Seattle City Council)

Electricity planning provides another relevant backdrop. A June 2026 Lawrence Berkeley National Laboratory report found that growth in data centers and other large electricity users was creating bottlenecks in grid connections. Its proposed responses spanned forecasting, connections, resource planning, operations and the allocation of costs. (Lawrence Berkeley National Laboratory)

A separate August report from the laboratory examined how utilities and regulators manage large customers’ operational and financial risks through electricity rates and service agreements. It identified insufficient supply and underused investments as risks that can affect other customers—not simply the companies building facilities. (Lawrence Berkeley National Laboratory)

Together, those findings suggest that local permission and electricity availability should be assessed separately when evaluating a proposed buildout. They do not quantify any additional delay caused by this election. For AI infrastructure planning, the supported conclusion remains narrower than a nationwide slowdown: Trenton voters backed a construction prohibition, while its project-level consequences and the broader national legal picture remain unresolved. (Lawrence Berkeley National Laboratory)