A few months into my time at Aston, I started noticing a pattern at public meetings.
The project on the table was always specific: a site, a building, a power line. The questions at the microphone were not. Somewhere in hour two, after the water questions and the traffic questions, someone would stand up and ask something like: “Aren’t you the companies that say AI is going to take all our jobs?” Or: “The people building this say even they don’t understand it. Why would we want it here?”
The first few times, I filed those questions as digressions — someone airing a general grievance about technology at the wrong microphone. I was wrong. Those questions weren’t off-topic. Increasingly, they were the topic.
The one place you get to vote on AI
For most people, AI is an abstraction. You can’t picket a model. You can’t file an injunction against a training run. There’s no public comment period for a capability.
A data center, though, has an address. It needs a zoning variance, a water hookup, and often a tax deal voted on by people who actually live nearby. For a public carrying enormous diffuse anxiety about AI — job loss, surveillance, technocratic control, wealth concentrating in fewer hands — the data center is where the abstract becomes concrete. It is the one place ordinary people get a vote on the technology reshaping their lives.
So yes: the data center fight is about siting. Water, power, noise, taxes — those concerns are real, and I wrote about them in my first post. But it’s also become a proxy war over AI itself. And once you see that, the ferocity of the backlash stops being mysterious.
The case against AI, brought to you by AI
Here’s the strangest part of this whole debate: the most damning arguments against AI haven’t come from activists. They’ve come from the industry’s own leaders.
Dario Amodei, the CEO of Anthropic, has warned that AI could eliminate half of all entry-level white-collar jobs and push unemployment to 10–20% within five years. In 2023, more than 350 AI executives and researchers — including the CEOs of OpenAI, Google DeepMind, and Anthropic — signed a one-sentence statement declaring that “mitigating the risk of extinction from AI should be a global priority alongside other societal-scale risks such as pandemics and nuclear war.” Extinction. Their word, not mine.
More than one leader has said they’d prefer to slow down, if only their competitors would too — a remarkable admission that the race matters more than the risk. And they freely concede that no one fully understands how their own systems work.
Imagine a pharmaceutical CEO announcing that their new drug might eliminate millions of jobs, that nobody entirely understands how it works, and that it poses a risk to civilization on par with nuclear war — and then expressing surprise that the town down the road isn’t thrilled about the factory.
Layer on the industry’s on-the-ground behavior — land assembled through shell companies, local officials bound by NDAs, utility deals that shift costs onto ordinary ratepayers — and you get the trust deficit that’s now showing up at every planning commission in America.
The backlash is deserved
I want to be precise here, because “deserved” is a strong word and I mean it.
Deserved doesn’t mean every project deserves to die, or that every opponent is right on the facts. It means the burden of proof sits exactly where it belongs: on the people asking to build. When your sector’s own CEOs warn of mass unemployment and civilizational risk, you don’t get to be shocked when communities treat your infrastructure accordingly. Skepticism isn’t a messaging failure. It’s a rational response to the industry’s own words.
None of this makes me anti-AI. Some of the most important work happening anywhere is happening in these models. But “the technology is exciting” is not a public-interest argument.
Nifty isn’t worth a substation
Which brings me to the deeper problem: the benefits on offer so far are, for most people, modest. Chatbots that draft emails. Summaries of meetings you skipped. Images of anything you can describe. Customer service that’s somehow worse than the phone tree it replaced.
Useful? Sure. Convenient, handy, occasionally delightful. But “handy” does not pay for a substation. “Nifty” does not offset a higher electric bill.
The implicit deal offered to communities looks like this: you absorb the local, immediate, measurable costs — land, water, power, noise, public subsidy — in exchange for benefits that are diffuse, distant, and mostly accruing to shareholders in another time zone. People can smell a bad trade. That’s not ignorance. That’s arithmetic.
What would actually change minds
For AI to earn the infrastructure it demands, the technology has to move from convenient to transformative — transformative in ways an ordinary person can point to:
Transportation. Roughly 40,000 people die on American roads every year. Self-driving that cuts fatalities by 99% isn’t a gadget; it’s one of the great public health achievements in history — with car insurance bills falling to nearly zero as a bonus.
Medicine. AI that designs drugs, catches cancers early, and cracks diseases that have wrecked families for generations. The Nobel Prize for AlphaFold was the down payment. Cures are what make the trade obvious.
Abundance. Robotics driving down the cost of things people actually want — with the crucial qualifier that the prosperity is broadly shared, not pooled in a handful of stock tickers.
The report card underneath it all: inequality going down, not up.
Deliver that, and the politics of data centers change overnight. People don’t picket the infrastructure of things they love. Nobody storms a town hall over the server farm behind the drug that saved their kid.
What this means for the fight in front of us
Two implications, one for each side of the room.
For developers: know what you’re walking into. You’re not merely defending a building; you’re being asked to answer for an entire industry’s promises, warnings, and behavior. You cannot win by dismissing the people raising those concerns as NIMBYs, and you cannot outsource credibility to a trade group. What you can do is be the exception: show up early, tell the truth about power and water, put benefits in enforceable agreements, site where you’re wanted — and be honest about what you can’t answer. Notice that not one item on that list is a communications tactic. They’re all structural, which is why they work and why they’re rare.
For communities: it is entirely fair to ask why you should host the infrastructure of a technology whose own builders keep warning you about. Hold that question; it’s legitimate. But separate the two decisions in front of you. The first — what AI should become, and who it should serve — is a national argument, and yours to join as a citizen. The second — is this project, on these terms, good for this town — is a local decision, and yours to negotiate as a community. A planning board can’t set AI policy. It can absolutely set its price, its protections, and its terms. And it can say no.
The second conversation goes better when it’s run by someone who doesn’t profit from how it ends. That’s the whole design premise behind CommonWatts: one process rather than two advocates, facts both sides help verify, and fees that are published in advance and don’t move with the outcome. A neutral you can’t audit is just a consultant with better manners.
The way out
I started CommonWatts because I believe local fights deserve to be settled on local merits — on facts both sides had a hand in checking, and terms written down where anyone can read them. But the national argument won’t be settled by better messaging. It will be settled when AI’s benefits show up in people’s lives as visibly as its costs show up in their utility bills.
Until then, expect more packed town halls. The way out of the proxy war isn’t PR. It’s a technology worth fighting for instead of against.
Make it worth it, and the war ends itself.