Two data center projects, two very different local outcomes

How — and where — data center siting is actually being decided.

The data center build-out is usually described in big numbers: gigawatts of power, billion-dollar campuses, and the names of the world’s largest technology companies. But many of the decisions that determine whether a given project actually gets built happen far from those headlines — in individual communities weighing what a facility would mean for their local resources and infrastructure. Two recent examples, in very different parts of the country, show how varied those outcomes can be.

Rosebud, Texas

Rosebud is a small Central Texas city of roughly 1,400 residents. It has been considering whether to provide municipal water service to a proposed data center project referred to as “Stargate Data Center.” As of now, no final agreement has been reached. The city has approached the question deliberately, weighing a long-term commitment to a large industrial water customer against its own ongoing water and infrastructure needs.

A few points are worth stating plainly. The project’s apparent water requirements are modest relative to the very large “hyperscale” campuses that dominate national coverage. And despite the shared name, there is no confirmed connection between this project and any similarly named national data center program; that link should not be assumed without independent verification.

The broader takeaway is simple: data center demand is now reaching smaller communities that have never had to evaluate this kind of project before, and the resource trade-offs — especially water — are real enough that even a willing town may move cautiously.

Allen Park, Michigan

Outside Detroit, the city of Allen Park was the site of a proposed 26-megawatt data center that did not move forward. A local political campaign, Joanna Whaley for State Representative, publicly stated that the city’s planning commission denied the project and credited organized community opposition for the outcome.

That account has not been independently confirmed against the commission’s official record, and we present it as a claim rather than a settled fact. What is clear is that the project did not advance, and that — as in a growing number of communities — local sentiment and the public approval process can be decisive.

The bigger picture: more “yes, here” than “no”

Step back from any single project and a clearer pattern emerges. Across the local-government meetings we tracked in the same week, 207 separate meetings — in 28 states — took up data centers in some form. And contrary to the prevailing “backlash” narrative, more of them advanced zoning amendments, land-use definitions, and permitting frameworks to accommodate data centers than moved to restrict or ban them.

The activity is also highly concentrated. Pennsylvania alone accounted for roughly 40% of it — 84 meetings — and leaned decisively permissive, with 26 zoning or permitting actions against just 5 moratorium-related discussions. A handful of states drove most of the restrictive momentum: Ohio, Maine, Wisconsin, New York, and North Carolina. Several others — Texas, Virginia, South Carolina, Missouri, Minnesota, Louisiana, and Oregon — were quietly building the rules to welcome development. And Michigan sat squarely in the middle, splitting 25 meetings evenly between permitting and moratoriums — the country’s clearest data-center battleground.

The national story, in other words, isn’t rejection. It’s divergence — a market sorting itself, community by community, into the places that say yes and the places that say no.

Why local outcomes matter

These two cases point in opposite directions: one community still deliberating whether to enable a project, another where a project appears to have been turned away. Neither outcome was the subject of national reporting, yet both carry real information for anyone tracking where data center capacity will, and won’t, be built.

That is where Obedio focuses — surfacing these local developments early, across the country, and, just as importantly, distinguishing what has actually been decided from what has merely been claimed. In a build-out moving this fast, the difference between a confirmed outcome and an unverified one is exactly the difference that matters.

Note on sourcing: the national figures reflect 207 local-government meetings discussing data centers across 28 states for the week of June 1–3, 2026 — one week of tracked activity and a conservative lower bound, with state-level coverage that varies week to week. The account of the Allen Park outcome comes from a public statement by a local political campaign and has not been verified against official records. Where facts remain unconfirmed, we have said so.