Anger over data centers just showed up at the ballot box. TechCrunch AI reports that the November 3 midterms gave Democrats full control of Congress, while candidates who opposed the rapid AI buildout won seats in Texas, Maine, Michigan, Nevada, Georgia, and South Carolina. An NBC exit poll found that almost 80% of respondents oppose data centers in their neighborhoods. Almost 60% called them a crucial or major factor in their vote.
That 60% is the number to watch. What used to be a zoning fight now moves votes, even after pro-AI super PACs spent millions trying to sway the outcome. “It is clear that money does not buy outcomes,” said tech and policy researcher Anshi Bhatt.
Ohio shows how fast this can turn
Ohio is the clearest warning. OpenAI has eyed a $500 billion data center there, backed by more than $100 billion in Nvidia financing. But the county where Meta planned a 700,000-square-foot facility just passed a data center ban, and local communities across the state overwhelmingly voted against buildout measures.
Wall Street has noticed:
- Morgan Stanley analysts see Democratic governor-elect Amy Acton as a major risk to Ohio’s buildout.
- Goldman Sachs predicts construction delays in states like Georgia and Texas because of Democratic gains.
Both calls came right after the vote, so treat them as early reads. Still, political risk is now part of the investment conversation.
Congress gets the headlines, but states may move first
The biggest federal swing is the bill from Sen. Bernie Sanders and Rep. Alexandria Ocasio-Cortez, which would pause new data centers nationwide until Congress adds guardrails. It faces a possible veto, and Democrats aren’t united behind it. “Even among Democrats, a blanket freeze is contested,” said Nadine Smith, president and CEO of Color of Change. Some worry a moratorium would slow AI progress and hurt national security.
Sen. Martin Heinrich’s GRID Savings Act is narrower. It would stop grid infrastructure costs from landing on communities during large AI buildouts. My bet is that this kind of “pay your own way” rule has a better shot than a freeze, because it’s a much easier promise to defend to voters staring at a power bill.
The faster action will likely come from states and towns. “Regardless of federal action, I could imagine that we will see additional movement at the state and local level,” said Maia Woluchem of Data & Society. She pointed to Texas governor-elect Gina Hinojosa, who supports a statewide moratorium. Smith also expects AI lobbyists to start calling lawmakers, and possibly more executive orders to get around whatever Congress blocks.
The buildings may not be the real target
One private equity executive told TechCrunch that people don’t necessarily hate data centers. They hate AI, and data centers have become a stand-in for those fears. The poll backs that up: more than half of respondents said AI will do more harm than good.
It makes sense when you think about it. You can’t picket a chatbot, but you can show up at a county meeting about a building down the road.
The industry’s own tactics haven’t helped. Amazon built a data center in Gilroy, California, without many residents knowing, by satisfying a 45-year-old zoning law. Some local officials have had to sign NDAs barring them from discussing projects. Pollution adds another flash point, and Smith tied it directly to the national security argument: “National security matters. And so do the families in South Memphis breathing exhaust from the gas turbines powering an AI supercomputer.”
What the next 1-3 years could look like
This is my read, not a sure thing:
- Expect a patchwork. More county bans and state moratoriums will make the map harder to plan around than any single federal law.
- Paying for grid upgrades could become the entry fee for big projects, whether by law or through local deals.
- Secrecy gets expensive. NDAs and old zoning loopholes may win a project once, then fuel the next ban.
- Timelines slip. If Goldman’s right about delays, compute capacity could arrive later than current plans assume.
What to do now
- If you build or finance data centers, score political risk next to power and land. Go to the community early, and budget for your own grid costs.
- If you run AI workloads, ask your cloud vendors where new capacity is coming from and how exposed it is to local opposition. Don’t build a roadmap around one region.
- If you lead an AI company, stop treating this as a permitting issue. Voters are asking what they get out of AI, and “beat China” won’t answer that on its own.
Smith’s warning to her own side shows where this goes next: “Democrats ran on making Big Tech pay. Our job is to make sure they remember it once the cameras leave.” The first test comes when Congress returns and the lobbyists start calling. You can find more detail in the original TechCrunch AI report.