Voters Back Limits on AI Data Centers as OpenAI Promises to Pay Its Power Bills

American voters have sent a clear message about AI data centres. In the November 3, 2026 elections, voters favoured limiting data centre development in 40 of 41 local data center ballot measures. Two days later, OpenAI responded with a new promise to pay the full cost of the electricity its data centres use.

Together, these events mark a turning point. For the first time, local communities are pushing back hard against the massive buildings that power artificial intelligence. Here is what voters decided, why they are worried and how the AI industry is reacting.

What the Data Center Ballot Measures Decided

According to Ballotpedia, voters took the side of limiting data centres in 40 of the 41 local contests on November 3. The only exception came in Harwood, North Dakota, where voters rejected a review requirement for certain developments.

Of these, 27 measures directly limited data centres or expressed opposition. They included:

• Full bans (5): one in California, one in Florida and three in Ohio
• Partial bans (12): three in Kansas, one in Maine and eight in Ohio
• Six-month moratoriums (3): all in Maine
• Zoning rules (2): both in Adams County, Ohio
• Non-binding opposition (2): both in Illinois
• Voter approval requirements (2): Grove City, Ohio, and Janesville, Wisconsin
• Financial guarantees (1): Perry, Ohio

In addition, voters defeated 13 measures that would have supported data centres. These included four rezoning measures in Wilmington, Ohio, and a measure in Boulder City, Nevada, that would have allowed data centres in a specific area.

The trend did not start on election day. Voters had already decided six data centre measures earlier in the year, and all six favoured limits. More votes are coming, too, with two measures in Clarksville, Arkansas, and one in Mandan, North Dakota, scheduled for December.

Ohio Leads the Backlash

Ohio stands out as the centre of the fight. The state saw 21 data centre measures on the ballot, more than any other. Voters passed three full bans and eight partial bans across the state.

Some of the rules are very specific. For example, Grove City’s Issue 28 now requires a public vote before approving any development larger than 50 acres, using more than 20 megawatts of power or consuming more than 500,000 gallons of water a day. Meanwhile, Perry’s Issue 16 requires financial guarantees from large hyperscale developers that seek certain tax breaks.

Ohio has become a popular location for data centres thanks to its land, power grid and central location. However, the election results show that many residents feel the costs now outweigh the benefits.

Why Communities Are Pushing Back

Several concerns are driving the opposition. The biggest is electricity costs. AI data centres use huge amounts of power, and many residents worry that utilities will pass grid upgrade costs on to ordinary households. We explained this problem in detail in our report on how AI data centres are driving up power bills.

Water is another major issue. Many data centres use large amounts of water for cooling. According to Android Headlines, heavy water use drove much of the public opposition over the summer.

Finally, many residents felt shut out of decisions. In some cases, local officials signed non-disclosure agreements (NDAs) before reviewing proposals. As a result, communities often learned about huge projects only after the deals were nearly done.

OpenAI’s National Community Compact

OpenAI moved quickly after the vote. On November 5, the company released its National Community Compact and a Compute Campus Policy Blueprint. According to Yahoo Finance, OpenAI promised: “We’ll pay the full cost of the electricity we use and the infrastructure needed to serve us.”

The compact includes several key commitments:

• Pay its own way: OpenAI will cover the cost of power and grid infrastructure for its campuses.
• Financial guarantees: It will offer financial assurances before major spending begins. This protects other customers if a project is delayed, scaled back or cancelled.
• No more NDAs for officials: Mayors and city councils will no longer have to sign strict non-disclosure agreements before reviewing proposals.
• More transparency: OpenAI plans to publish expected water demand, site plans and noise studies for local residents.
• Fair cost rules: It urged state utility commissions to make large power users cover the extra costs they place on the grid.

OpenAI also pointed to local deals as examples. In Effingham County, Georgia, an agreement lets Georgia Power reduce electricity deliveries to its campus by up to 1,000 megawatts during periods of high demand. The company also committed funding for free school lunches for 14,000 students there. In Pike County, Ohio, it pledged $6 million toward local childcare with SB Energy.

“Families should be able to count on basic protections whoever builds a campus in their community,” OpenAI said.

Will Other Tech Giants Follow?

OpenAI called voluntary corporate commitments “an important starting point.” However, the bigger question is whether Microsoft, Google, Amazon and Meta will make similar promises. These companies are spending hundreds of billions of dollars on data centres, and they all need local approval to build.

If more communities pass bans and moratoriums, the AI industry could face delays and higher costs. On the other hand, clear rules on power, water and transparency could make it easier to win public support. Our report on the AI energy crisis explains why the industry’s power needs keep growing.

What It Means Beyond the US

The same debate is spreading worldwide. Countries in Europe, Asia and Africa are competing to attract data centre investment. At the same time, many face pressure on power grids and water supplies.

For countries like Nigeria, where electricity supply is already strained, the US experience offers a useful lesson. New data centres can bring jobs and investment. However, governments need clear rules so that local communities do not end up paying the price.

The Bottom Line

The November 3 data center ballot measures show that public patience with AI’s growing footprint is wearing thin. Voters backed limits in 40 of 41 contests, and OpenAI responded within days with promises on power costs, transparency and NDAs. Ultimately, the AI boom now depends not just on chips and capital, but also on winning the trust of the communities where data centres are built.