Here’s a number that should terrify every AI executive in Silicon Valley: 71% of Americans don’t want a data center anywhere near where they live. Not a slim majority. Not a vocal minority. Seven in ten adults, cutting across party lines, age groups, and income brackets, looked at the AI industry’s $200 billion infrastructure buildout and said: not here, not in my backyard, absolutely not.
That’s according to a comprehensive Gallup poll released this week — and the findings are far more damning than the headline number suggests. Nearly half of respondents (48%) didn’t just oppose local data centers; they strongly opposed them. For context, that’s more opposition than Americans express toward having a nuclear power plant built in their community (53% opposed). The AI industry has somehow managed to become less popular than nuclear energy in the public imagination.
The Water and Power Problem Nobody Wants to Talk About
The reasons aren’t abstract or ideological — they’re viscerally local. Respondents cited excessive energy consumption, water use, pollution, higher utility bills, increased traffic, and degraded quality of life. These aren’t hypothetical concerns. In places like Mesa, Arizona and Goose Creek, South Carolina, residents have already watched their water tables drop and their electricity rates climb as data centers moved in next door.
A single large AI data center can consume as much electricity as 80,000 homes. It can drink millions of gallons of water daily for cooling — in regions already facing drought. And when your local utility has to build new transmission infrastructure to feed a Meta or Microsoft campus, guess who pays for it? The rate base. Which means you.
Nearly half of all survey respondents said they worry “a great deal” about the environmental footprint of AI data centers. This isn’t climate activism from the coasts — this is kitchen-table economics from everywhere.
Big Tech’s $200 Billion Problem Just Got a Zoning Permit
Here’s where it gets existential for companies like Amazon, Microsoft, Meta, and Google. All four have committed hundreds of billions in combined capital expenditure on data center buildouts through 2028. Microsoft alone pledged $80 billion for fiscal year 2025. Meta is planning a $65 billion spend. Amazon’s capex run rate is north of $100 billion annually.
All of that money assumes you can actually build these facilities somewhere. And “somewhere” is increasingly running out of options.
The Gallup numbers translate directly into political pressure at the local level — where zoning boards, county commissioners, and state legislatures make the actual permitting decisions. A 71% opposition rate means that any politician who approves a new data center is actively defying their constituents. That’s not a recipe for fast-tracking permits.
We’re already seeing this play out. Communities in Virginia, Georgia, Indiana, and Texas have blocked or delayed data center projects in recent months. Some municipalities have enacted moratoriums on new construction. Others are imposing impact fees that dramatically change the economics of these builds.
The Bipartisan Revolt That AI Lobbyists Can’t Buy Their Way Out Of
What makes this particularly dangerous for the industry is that opposition isn’t partisan. Majorities of Republicans, Democrats, and independents all said no. You can’t lobby your way around that with campaign contributions to one party. You can’t frame this as a red-state vs. blue-state issue. The NIMBY instinct is the one thing that unites the American electorate in 2026.
The Trump administration has been broadly pro-AI and pro-infrastructure, but even executive orders can’t override local zoning authority. The president can declare data centers critical infrastructure all he wants — it doesn’t make a county supervisor approve a building permit when 71% of their voters are against it.
And the industry’s standard playbook — jobs, tax revenue, economic development — is failing. Data centers employ shockingly few people relative to their physical footprint. A $2 billion facility might employ 50 full-time workers. Compare that to a manufacturing plant of similar investment that might employ 2,000. The value proposition simply doesn’t resonate when residents can do basic math.
Follow the Money: Who Actually Gets Hurt
The immediate losers are obvious: Amazon Web Services, Microsoft Azure, and Google Cloud are all racing to expand capacity to meet AI inference demand. Every month of permitting delay costs them market share and slows their ability to serve enterprise customers.
But the second-order effects are more interesting. If domestic construction slows, AI companies will look overseas — which creates national security concerns around data sovereignty. If power constraints bite, AI model training costs go up — which favors incumbents with existing capacity over startups. If water becomes the binding constraint, we’ll see a geographic concentration of AI infrastructure in cold-climate regions with abundant hydro power, which creates its own geopolitical dynamics.
The real winner here might be nuclear energy, ironically. Small modular reactors that can be co-located with data centers, generating power on-site without tapping the local grid, suddenly look like the only viable path to scaling AI infrastructure without triggering community revolts. Microsoft, Google, and Amazon have all signed nuclear power agreements in the past year — and this poll explains exactly why.
The Verdict: AI’s Biggest Threat Isn’t Regulation — It’s Geography
Wall Street is pricing AI stocks as if the infrastructure buildout is a foregone conclusion. It isn’t. A 71% opposition rate — higher than nuclear, higher than virtually any industrial project polled in recent memory — means the physical layer of the AI revolution faces a constraint that no amount of capital can solve quickly: public consent.
The AI industry spent 2023-2025 convincing investors that demand was infinite. Now it needs to convince 71% of Americans that the noise, the water, the power draw, and the empty parking lots are worth it. That’s a much harder sell than any earnings call.
Big Tech has fought off regulators, antitrust suits, congressional hearings, and EU fines. But you can’t fight a zoning board with a lobbying budget. And you definitely can’t fight seven out of ten voters who just don’t want you in their neighborhood.
