Local opposition to data centers just crossed a milestone. More than 500 bans, moratoriums, or restrictions on new data center projects are now on the books across the US, and two of the country’s biggest states, New York and Texas, have joined the pushback. That’s according to The Information, which tracked the surge as communities react to the physical footprint of the AI boom.
This matters because the fight over where AI actually lives just went mainstream. For two years the story has been about models, chips, and funding rounds. The 500 number is a reminder that all of it runs on buildings that need land, power, and water in somebody’s backyard.
What’s happening
- The count of local bans and restrictions on data centers has topped 500 nationwide, per The Information.
- New York and Texas are now part of the resistance, alongside the smaller towns and counties that started it.
- The opposition is bipartisan and geographic, not confined to one region or one political camp.
Why Texas and New York change the math
Texas is the tell. The state markets itself as the friendliest place in America to build, with cheap land and a grid it controls on its own. When Texas communities start saying no, the industry loses its easy fallback. New York brings the other flavor of resistance: dense, expensive, and protective of an already strained power supply.
The status quo before this was simple. Hyperscalers picked a rural county, promised jobs and tax revenue, and got a fast yes. That playbook is breaking. Residents have watched electric bills climb, wells drop, and construction traffic roll in, and they’ve learned to organize before the concrete gets poured.
The real fight is power and water
Strip away the politics and the objection is physical. A large AI data center can pull as much electricity as a small city, and the training clusters behind today’s frontier models are hungrier than the search-and-storage facilities that came before them. Two pressure points keep coming up:
- Electricity. New load competes with homes and existing businesses, and it can push up rates for everyone on the same grid.
- Water. Many designs use large volumes for cooling, which lands hard in drought-prone areas.
Communities aren’t debating whether AI is useful. They’re asking who pays for the infrastructure and who absorbs the strain.
Why it matters for the AI industry
Compute is the constraint on AI progress, and compute has to be sited somewhere. If 500 jurisdictions and counting are closing the door, the buildout gets slower and more expensive. A few things follow from that:
- Siting becomes a competitive edge. Companies that lock in power agreements and community goodwill early will move faster than rivals scrambling for locations.
- Costs drift up. More legal fights, more concessions, and longer approval timelines all land in the capex line.
- The map shifts. Expect more interest in places with spare power, including deals tied to nuclear, and in regions willing to trade restrictions for revenue.
What stands out here is the speed. Five hundred restrictions is not a fringe protest anymore. It’s a structural headwind that sits between the AI labs and the capacity they’ve promised investors.
What to watch next
Keep an eye on how the big builders respond. The likely moves are direct power deals that bypass strained public grids, closed-loop cooling that cuts water use, and bigger community payouts to buy back local support. Watch the states too. If Texas and New York start writing formal rules rather than letting towns decide one by one, that sets a template others copy.
The headline takeaway is straightforward. The AI arms race has a zoning problem, and it’s growing faster than most forecasts assumed. The companies that treat siting as a core strategy, not an afterthought, will be the ones with compute to spare when the next model generation lands. More detail is available in the original report from The Information.