Sixty-four percent of Americans oppose AI data centers being built in their local community, according to a Pew Research Center poll released June 10. The number is up from 52% in March — a 12-point jump in three months that tracks the accelerating pace of data center construction across the country. The AI infrastructure sprint is colliding with local resistance, and the resistance is winning. [1]

The opposition crosses partisan lines. Seventy-eight percent of environmentally concerned respondents oppose data centers locally. Sixty-one percent of Republicans and 67% of Democrats say the costs outweigh the benefits. The opposition is not ideological — it is practical. Data centers consume enormous amounts of water, generate noise, increase traffic, and provide few permanent jobs. The promised economic benefits rarely materialize at the community level. [2]

The gap between national AI ambitions and local acceptance is the story. Tech companies need data centers to run the AI models that power their businesses. Communities don't want the facilities in their backyards. The result is a planning and permitting crisis that threatens to slow the AI infrastructure buildout at exactly the moment when demand is accelerating. [3]

X's frame treats the opposition as NIMBY hypocrisy. Users noted that the same Americans who use ChatGPT, Google AI, and Copilot daily reject the infrastructure that makes those services possible. The data center is the factory floor of AI — and nobody wants a factory in their neighborhood. The irony is that the opposition is strongest in the communities that benefit most from AI services. [4]

The practical objections are concrete. A large AI data center consumes five million gallons of water per day — equivalent to a town of 50,000 people. The typical facility uses 300,000 gallons daily. Eighty percent of that water evaporates and never returns to aquifers. In Virginia, where 13% of the world's data centers are located, sixteen projects have been stopped by local opposition. The numbers tell a story that national AI policy has not yet reckoned with. [5]

MSM coverage treated the poll as a policy challenge. The Washington Post focused on the permitting bottleneck — communities blocking projects that companies need to deploy AI at scale. The New York Times framed it as a tension between national competitiveness and local democracy. Neither outlet addressed the economic structure: data centers concentrate costs locally while spreading benefits globally. [6]

The economics are asymmetric. Data centers provide 20 to 50 permanent jobs after construction. They consume massive amounts of power and water. They increase traffic during construction and generate noise during operation. The tax revenue they generate is often offset by the infrastructure costs they impose. Virginia and Georgia gave up more than a billion dollars in tax incentives to attract data centers — money that could have funded schools, hospitals, or roads. [7]

X users pointed to the international comparison. India is building 3.5 gigawatts of data center capacity with $160 billion in investment. China is expanding even faster. The United States, where 64% of Americans oppose local data centers, risks falling behind in the infrastructure race that will determine AI's future. The gap between American NIMBYism and Chinese state-directed construction is the story of the next decade. [8]

The broader pattern is the collision between abstract benefits and concrete costs. AI is a national priority — the government, the military, and the economy all depend on it. But the infrastructure required to run AI — the data centers, the power plants, the cooling systems — exists in specific places, affects specific communities, and generates specific costs. The 64% opposition number is not a poll result. It is a structural constraint on the AI industry's growth. [9]

The paper's position is that local resistance is the biggest risk to AI infrastructure allocation. The industry needs data centers. Communities don't want them. The gap between national ambition and local acceptance will determine how fast AI scales — and who pays the cost. [10]

-- THEO KAPLAN, San Francisco