In Falls Township, Pennsylvania, residents pushed back against a proposed AI data center. In Becker, Minnesota, a similar protest formed. In rural Virginia, in suburban Arizona, in communities across the country where land is available and power infrastructure exists, the same pattern is appearing: a technology company announces a large facility, local residents discover what it means for their water supply and power grid, and a public conflict begins that the companies involved did not anticipate and the communities feel they should have been consulted about before the decision was made.
The data center is the physical infrastructure of artificial intelligence. Training and running the large language models, image generators, and reasoning systems that are now embedded in consumer products, financial systems, and government services requires electricity in quantities that strain regional grids, water for cooling in volumes that affect local aquifer levels, and land in areas where the alternative use might have been agricultural, residential, or commercial. The facilities are enormous, they operate continuously, and they are not quiet. The communities hosting them bear costs that the companies operating them have not always been required to fully account for.
The political profile of the resistance is unusual. It does not align cleanly with the partisan polarization that characterizes most national debates. The concerns being raised — about water use, about power costs, about noise, about local tax incentives extended to large technology firms — are concerns that animate both environmentally-oriented progressives and fiscally-conservative community members who object to corporate subsidy structures. The AI data center resistance is one of the cleaner examples of an issue that currently has no natural party home.
The industry argument is that data centers bring jobs, tax revenue, and infrastructure investment to communities that benefit from economic activity. The argument is not false. Data centers do employ people, do generate tax revenue, and in some cases do improve regional infrastructure. The argument omits the questions about distribution: which people get the jobs, at what wages, compared to the water and power costs borne by the full community; what the tax revenue covers versus what local governments have committed to subsidize; and who made the decision to prioritize data center attraction over other economic development alternatives.
The deeper structural question is one that AI policy discussions rarely address directly: the physical footprint of the digital economy is not evenly distributed. The communities that host the infrastructure that powers AI products are not, in most cases, the communities whose residents are the primary users of those products. Silicon Valley builds the products. Rural Pennsylvania hosts the servers. The economic logic that produced that distribution was not subject to democratic review. The community meetings happening across the country are an informal and belated version of that review.
What is being contested in Falls Township and Becker and rural Virginia is not primarily a technology question. It is a governance question: who decides where the infrastructure goes, at whose cost, with whose consent, and with what accountability when the costs exceed what was promised.
American Life is independent analysis. Become a Member
