Games Contact us
Select Page
⚡ TL;DR
Opposition to data centres has become a bipartisan political issue in the United States. Community groups fighting projects more than doubled to 833 in the first months of 2026, and the fight is now reaching Congress and the midterm campaigns. Technology leaders who depend on AI capacity should treat community consent, power and water as core project risks, not afterthoughts.

For two years the artificial intelligence conversation has been dominated by chips, models and capital spending. This autumn a different constraint is moving to the front: whether communities will let the buildings that house all that computing be built at all. A BBC report this week on how the backlash against data centres could shape the US midterm elections captures the shift. A longtime Democratic senator who was barely thinking about data centres a year ago is, according to one of the senator’s advisers, now meeting staff regularly to track their impact and drafting legislation to curb their spread.

That adviser put it plainly: “This happens sometimes, something starts slow and then it goes fast.” For technology executives, infrastructure investors and anyone buying cloud or AI capacity, that sentence deserves attention. This article sets out what is driving the backlash, why it cuts across party lines, and what technology teams can do about it.

From invisible infrastructure to local flashpoint

Data centres used to be anonymous. They were windowless warehouses near fibre routes, and few residents could say what happened inside. AI changed that. Packed with rows of chips, sometimes seven feet high, they have become what the BBC calls the physical embodiment of AI’s explosive growth, and of the varied fears people associate with it.

The scale of individual projects is part of the story. In Indiana, resident Angelita Soriano told the BBC that a new Amazon data centre is being built directly across the street from her home. When complete, the 26-building complex will cover more than a square mile and maintain 782 diesel-powered generators. “I don’t want to be the guinea pig for this,” she said. In Ohio, a joint effort between OpenAI, Nvidia, SB Energy and the US Department of Energy is set to build a complex stretching roughly six square miles, about the size of a typical US township.

There are roughly 3,000 data centres operating in the US today, and a Pew Research Center report cited by the BBC suggests that number could double in the next few years. That means the number of communities with a proposal on the table is about to grow sharply.

The numbers behind the backlash

The organising data is striking. Research from Data Center Watch, which tracks the issue for AI industry stakeholders and is part of an AI startup, 10a Labs, found that community groups organising against data centres more than doubled to 833 during the first few months of this year. They now appear in almost every US state. The BBC spoke with residents in Indiana, Arizona, Michigan, Pennsylvania and Oregon, and heard a consistent list of concerns:

  • Size and construction impact: large footprints, heavy trucking and long build periods.
  • Operational noise: cooling systems and generators that run around the clock.
  • Air, water and energy: diesel backup generation, water used for cooling, and demand on local grids.
  • Process speed: a sense that local authorities approve projects too quickly, with too little public input.

Stephanie Maldonado, political director of the Arizona organising group LUCHA, called the pushback “a David and Goliath story” that began several years ago. “People have been ready to put a stop to it,” she said. “Now, people are angry.” New national polling cited by the BBC puts AI regulation and data centres among the top issues for voters in this election cycle.

Why it crosses party lines

Chloe Autio, an AI policy adviser, told the BBC that this is not just a left-wing cause: “there is huge pushback among conservatives to data centres.” Her advice to politicians is that whoever articulates a clear position and sticks to it, in either party, stands to benefit. The adviser working on the Democratic senator’s bill agreed, describing a “clear policy need” and a nationwide backlash from constituents who are “seeing the impacts of data centres and they do not like them.”

That puts many Republicans in direct tension with President Donald Trump, who has championed unrestrained AI growth. Trump has called the backlash a “sick conspiracy” and has said that people and towns working to keep data centres out are “backwards and poor.” Democrats looking to win over unhappy or undecided voters are using that language as a campaign asset.

There is a structural reason the issue has travelled so fast. Data centres are approved through local zoning and permitting, which means federal lawmakers have little direct say. Yet voters expect their senators and representatives to act, so national politicians are being pushed into positions on decisions made by county commissions and planning boards. That mismatch generates both political heat and a flood of proposals for federal guardrails.

What the wider news and search signals show

The data centre debate is not happening in isolation. This week’s Google Trends data shows US interest in an “AI bubble” narrative, with commentary from outlets including the New York Post, the Financial Times and valuation professor Aswath Damodaran examining whether AI spending is outrunning earnings. BBC business coverage reports chip stocks falling after Broadcom chief executive Hock Tan said some investors have “no clue” and have set “extreme expectations.” Meanwhile, TechCrunch has reported that India has become the latest country to see rogue AI agent activity, and that Anthropic says its workforce has become less junior and spends more time in meetings.

Read together, these stories describe an industry under scrutiny on three fronts at once: the financial return on its spending, the safety of its systems, and the physical footprint of its infrastructure. Data centre politics is the front where ordinary voters, rather than investors or researchers, have the loudest voice.

What this means for technology leaders

Treat social licence as a critical path item

Site selection checklists usually cover power, fibre, land and tax incentives. Add community acceptance as a scored criterion with an owner. Projects that surprise a neighbourhood with a 26-building campus and hundreds of diesel generators invite exactly the organising the BBC describes. Early, specific disclosure of building heights, noise levels, water use and backup-power plans is cheaper than fighting a rezoning vote.

Cut the diesel and publish the numbers

The 782 generators figure is memorable precisely because it is concrete. Operators who can commit to cleaner backup power, closed-loop cooling or reclaimed water, and who publish measured performance after opening, will have a stronger case than those who offer general assurances. Independent monitoring of noise and air quality builds trust that promises alone do not.

Plan for slower, more conditional approvals

If your AI roadmap assumes new capacity coming online on historic permitting timelines, stress-test it. Moratoria, additional hearings and conditions on water and power use all stretch schedules. Build buffer into delivery dates, and keep a pipeline of alternative sites in jurisdictions with clear, stable rules.

Diversify capacity

Companies that buy rather than build should ask cloud providers where their capacity sits and how exposed it is to local opposition. Multi-region and multi-provider architectures protect against outages, and they also protect against a single project being delayed by a vote. Efficiency work matters too. Smaller models, better utilisation and workload scheduling reduce the amount of new construction your plans require.

Engage in the policy process

With legislation reportedly being drafted in the Senate, industry has a choice: shape workable standards or react to restrictive ones. Transparent reporting on energy and water, community benefit agreements and grid-friendly operation are all areas where practical proposals can be made before rules harden.

💡 Pro Tip: Run a “local newspaper test” on every planned site. If the facts about size, generators, water and traffic would alarm you on a front page, assume residents will find them, and address them in your first public presentation.

Risks and open questions

Several uncertainties remain. The BBC piece reports that the backlash is strong, but local outcomes will differ widely: some communities welcome the jobs and tax revenue, and the 833-group figure comes from an organisation that serves AI industry stakeholders, so it should be read as a count of organised opposition, not of every resident’s view. It is also unclear whether any federal bill can pass, or whether the issue will fade after the midterms. And the political dynamic may change if electricity prices, rather than land use, become the dominant complaint.

What does seem durable is the underlying tension. AI is being treated as a strategic priority at the national level, while its costs are felt at the level of a street, a school district or a water table. Policy that ignores either side will be unstable.

Bottom line

The data centre backlash has moved from scattered neighbourhood disputes to a bipartisan electoral issue within a year. With roughly 3,000 US facilities operating and the number potentially doubling, the number of communities facing a decision is rising fast. Technology leaders who bake transparency, lower-impact design and realistic schedules into their plans will be better positioned than those who assume permits are a formality.

Sources: BBC News, “How backlash against data centres could shape the US midterm elections” (October 2026), citing Data Center Watch and Pew Research Center; BBC Business and TechCrunch headlines; Google Trends US (18 October 2026).

Discover more from Kurums | Business Intelligence

Subscribe now to keep reading and get access to the full archive.

Continue reading