All Articles
AI3 min read7 November 2026

Voters Could Not Vote On AI Safety, So They Voted On The Data Centre Next Door.

TL;DR

Voters cannot vote on AI safety, but they can vote on a data centre. This week they did, and Meta admitted why: it is the visible form of AI. Here is what I think that means.

AIData CentresPoliticsOpinionAI Policy

This week voters across America did something the AI industry had not planned for. In local votes on 3 November, several towns, reportedly in Illinois and Maine, backed moratoriums on new data centres. One exit poll reportedly found that more than half of voters called data centres a crucial or major factor in who they backed for Congress. That is one poll, reported by one outlet, so I hold it loosely. But the direction is hard to miss.

What Meta Said

Meta's president, Dina Powell McCormick, said at the WSJ Tech conference that the Democratic wins will make data centres much harder to build, and that the industry had not done a good enough job explaining the benefits. She also said data centres are the visible form of AI, the part people can oppose locally.

That last idea is the one I keep thinking about.

Visible

For years the big AI argument has been about things nobody can see. Will a model deceive us. Will it kill us all by the end of the decade, as Jacob Coxon wrote when he resigned. Ordinary people cannot vote on that. There is no ballot paper for alignment.

But a data centre is a building. It has a power line, a noise and a bill. People can vote on that, and this week they did.

I think this is what happens when a technology has no democratic door. People find the door that does exist, planning permission, the local council, the electricity bill. So the safety debate arrived at the ballot box dressed as a zoning fight.

The Markets Noticed

On Thursday the Nasdaq reportedly had its worst day in nineteen months, with companies tied to the AI buildout hit hardest. I would not say one vote did that. Reports tie it to doubts about whether the buildout can keep its pace. But I notice the timing, and that Anthropic's IPO is reportedly aimed at mid November.

Back To Orbit

In my piece on Musk's plan I wrote that on the ground AI keeps running into neighbours, councils and power limits, and that in orbit there are far fewer people to ask. This week the neighbours and the councils answered. That makes the orbital bet look less like a stunt and more like a response to something real.

It also makes me uneasy. If the answer to people saying no is to build where nobody can vote, the problem has not been solved. It has only been moved.

What I Think

The industry says people do not understand the benefits. Maybe. But I think people understand exactly what they can see, a very large building, a very large demand for power, and nobody asking them first. If AI companies want to win these votes, they will have to answer the question people are really asking. Who pays, who benefits, and who gets a say. Explaining is not the same as asking.

AI did not get a vote on safety. It got a vote on a building. I expect a lot of this argument to be settled that way from now on, one town at a time.

Have a follow-up?

Explore more from the archive

Just finished "Voters Could Not Vote On AI Safety, So They Voted On The Dat…"? Ask anything related, I'll surface other articles that might help.

Found this useful?

Share it with someone who'd enjoy it.