US midterm elections: Backlash to AI data centres reaches ballot boxes

Data-centre policy became an election issue because voters were weighing the benefits of AI investment against its local costs. Democratic candidates who promised to halt development or impose safeguards claimed victories on Tuesday night. The result showed that support for rapid construction could carry political risks.

Texas, Georgia, and Ohio had become magnets for companies building facilities to power artificial intelligence. Yet voters in those states chose Democratic candidates who vowed to crack down on the technology. In Texas, Republican governor Greg Abbott had already paused permits for new data centres pending an audit.

The election outcome puts more pressure on governments to slow or regulate construction. It also challenges President Donald Trump's strong support for expansion. Elon Musk's reaction—“data centres in space looking pretty appealing right now”—captured the tension between ambitious AI plans and growing public resistance.

A data centre is a purpose-built facility containing many computer servers, storage systems, networking equipment, power supplies, and cooling systems. It processes and stores digital information. AI companies rely on these facilities because their models must handle huge amounts of data and perform vast numbers of calculations.

During training, powerful computers repeatedly adjust a model using examples. Afterward, the same infrastructure helps answer user requests, generate content, or analyse information. Running these systems continuously requires specialised equipment, reliable electricity, cooling, and fast connections. Larger models and more users increase demand.

The article describes companies such as Axiom, Paradox, Microsoft, Google, Meta, and Amazon expanding in states with suitable resources and incentives. Their growth explains why local disputes are intensifying. Residents see facilities that support AI, but also consume energy, occupy land, make noise, and create pollution.

The expansion is large enough to trigger organised resistance and major corporate commitments. At least 45 data-centre projects were disrupted during the second half of 2026 because of local opposition, according to Data Center Watch, a research project run by 10a Labs. The figure measures disrupted projects, not total construction.

Ohio shows the scale of planned spending. Meta planned to invest $1 billion in an Ohio data centre, while Amazon Web Services planned to spend another $2 billion in the state. Together, those announced projects represented $3 billion in planned investment. Ohio municipalities nevertheless rejected or restricted facilities.

The figures show a clash between enormous private investment and local control. Fifteen Ohio municipalities passed bans or introduced other restrictions. Texas also paused permits pending an audit. These actions suggest that future expansion may depend on legal safeguards, community approval, and access to energy and land.

Local opposition focuses on visible and immediate costs. The article says data centres use massive amounts of energy, make a lot of noise, cause pollution, and require a lot of land. These effects can make residents question whether promised jobs and investment justify changes to their communities.

Ohio provides the clearest example. Voters fiercely rejected data centres, while 15 municipalities passed bans or implemented other restrictions. In Texas, Democratic candidate Gina Hinojosa pledged to temporarily halt all new construction until legal safeguards existed. Republican governor Greg Abbott later paused permits pending an audit.

The opposition is now shaping policy as well as elections. More than half of Americans were fearful rather than excited about AI in a 2025 Pew Research Center poll. If concerns continue, companies may face slower approvals, tighter rules, or demands for stronger protections before projects proceed.

Tax breaks reduce the cost of building and operating a data centre. Promises of land make it easier to secure large sites, while access to energy supports facilities that need dependable, massive power supplies. Together, these policies can make one state more attractive than competing locations.

Texas illustrates the mechanism. The state offered tax breaks and promises of land and energy to technology giants including Axiom, Paradox, Oracle, Microsoft, Google, Meta, and Amazon. Republican governor Greg Abbott tried to make Texas an AI “epicentre” in September, encouraging a concentrated build-out.

The strategy brought investment but also political risk. Abbott later shifted course and paused permits for new data centres pending an audit. That U-turn followed Gina Hinojosa's pledge to halt construction temporarily until legal safeguards existed. Incentives can attract companies quickly, but local opposition can force governments to reconsider.

Large data centres concentrate many energy-hungry computer systems in one region. That can increase demand on local electricity networks and require more generation or transmission capacity. The facilities also need substantial land, while their equipment and cooling systems can create persistent noise and pollution.

The article does not quantify the exact effect on any one electricity system. It does state that data centres use massive amounts of energy, make a lot of noise, cause pollution, and require a lot of land. These combined pressures help explain local resistance in Texas and Ohio.

The political consequences are already visible. Fifteen Ohio municipalities passed bans or other restrictions, and Texas paused permits pending an audit. If regional impacts remain controversial, future projects may face tighter rules, slower approvals, or legal safeguards before companies can build at the planned pace.

Training an AI model means processing many examples and repeatedly adjusting the model's internal settings. Running it later means performing calculations for each request. These workloads can involve enormous datasets and many simultaneous operations, so powerful computers are needed to finish tasks within useful times.

Specialised data centres combine large groups of processors with storage, high-speed networking, dependable electricity, and cooling. The article focuses on their energy demand and describes Elon Musk's “Colossus” in Memphis as a massive cluster of AI-native data centres and supercomputers. Such clusters concentrate the hardware AI companies need.

The result is rapid infrastructure growth. Companies including Axiom, Paradox, Microsoft, Google, Meta, and Amazon have pursued facilities in states offering land, energy, and tax breaks. But the same concentration creates local pressure, prompting opposition, permit pauses, restrictions, and debate about how AI expansion should proceed.