Data Centers Are Coming to a City Near You and the Infrastructure Isn't Ready

A story in the Miami Herald this week asked a question most city planners have been quietly dreading: can your state's infrastructure actually handle an AI data center?[¹] It's a practical question with an uncomfortable answer. The electricity load from a single large-scale facility can rival a small city. The water consumption — used to cool the servers — runs into millions of gallons per day. And the facilities themselves are proliferating fast, chasing cheap land, permissive zoning, and power access in places that weren't designed for what they're being asked to absorb.

The AI and environment conversation has been building toward this moment for months. A piece in The Nation framed it plainly: will the AI boom lead to water and electricity shortages?[²] That framing — not a technical question about efficiency, but a resource scarcity question about towns and aquifers — is the shift worth noting. The conversation has moved off the tech pages and into local government meetings, utility commissions, and community boards. Advocacy researchers at Stand.earth put a number to Microsoft's trajectory this week, citing research showing a 160% increase in its data center carbon footprint and arguing the company has been pushed off any credible path to its stated climate goals by a $7 billion partnership with Big Oil to power AI infrastructure.[³] That finding didn't circulate primarily among climate specialists — it moved through general tech communities and news feeds, which suggests the audience for this argument is widening.

What's driving the energy appetite argument into new territory is the local dimension. Abstract emissions figures have a way of disappearing into policy debates. But a data center appearing in a specific county — drawing on a specific aquifer, straining a specific substation — makes the tradeoff concrete for people who hadn't been part of the AI conversation at all. The backlash documented in tech and local news isn't ideological opposition to AI; it's the predictable response of communities discovering they're being asked to subsidize the physical costs of a buildout whose economic benefits flow almost entirely somewhere else. Zoning fights, utility rate disputes, and water access conflicts are the kinds of local political battles that tend to escalate slowly and then all at once.

The timing matters too. This conversation is surging at a moment when Congress has shown little appetite for federal infrastructure standards around AI facilities, leaving states and municipalities to negotiate individually with companies that have far more resources and far more experience at that table. The communities most likely to host these facilities — those with cheap land and available power — are often the least equipped to model long-term grid stress or water table impacts before signing agreements. By the time the consequences are legible, the concrete has been poured. That asymmetry is what the pushback is really about, and it isn't going away as the buildout accelerates.