The AI revolution has a landlord problem. It needs staggering amounts of power, water, and real estate — and the communities being asked to supply it are increasingly slamming the door.

300 Bills and Counting

State lawmakers have proposed more than 300 bills in 2026 alone targeting data center construction and energy consumption. That’s not a slow regulatory awakening. That’s a stampede.

The proposals span the political spectrum: construction moratoriums in New York, Maine, Vermont, Oklahoma, and Georgia. Virginia is weighing whether to scrap $1.6 billion in annual tax breaks. Democratic legislators in blue states and Republican legislators in red ones are pushing essentially the same restrictions — a genuinely rare phenomenon in 2026 American politics.

“There just are not very many issues these days that you can’t predict what a state is going to do based on their partisan makeup,” said Morgan Scarboro, VP and economist at MultiState. “It’s been really interesting to see this play out in a way that doesn’t make clear sense on a partisan basis.”

When Bernie Sanders and Ron DeSantis agree on the direction of a problem, something fundamental has shifted.

The Ground-Level Reality

This isn’t an abstract policy debate happening in think tanks. It’s visceral and personal.

Archbald, Pennsylvania — population 7,500 — is staring down proposals for 51 data center buildings with 13.4 million square feet of floor space. That’s nine times the size of the U.S. Capitol, in a small town in the Lackawanna River valley. Residents packed council meetings waving “Boycott AI” signs.

“You are destroying our lives if you allow these data centers to come in,” resident Carolyn Mizanty told council members.

In Memphis, Elon Musk’s xAI has been running methane gas turbines — sometimes without permits — accused of pumping pollutants into predominantly Black neighborhoods already burdened by industrial waste. Mississippi’s environmental regulator approved the generators despite fierce local resistance.

In Adams County, Ohio, voters are circulating petitions for a statewide constitutional amendment to ban data centers entirely. Not regulate. Ban.

Mayors Are Breaking Ranks

The twist that separates this from typical NIMBY fights: elected officials who previously championed tech investment are publicly turning against the industry.

Phoenix Mayor Kate Gallego told a SXSW audience that her city is “growing tired” of multiplying data centers. Arizona’s largest utility has warned that if every proposed data center were approved, electricity demand would hit 19,000 megawatts — more than double the grid’s record peak.

Gallego dropped a damning detail: she often discovers tech companies have arrived in her city only by checking the utility’s biggest-customer list, thanks to NDAs that keep communities in the dark.

Even Sunnyvale’s mayor — in the heart of Silicon Valley — admitted being in “constant battle” with the local utility provider over data center strain.

The Political Powder Keg

Here’s where it gets consequential. 2026 is a midterm year with 36 governor’s races. Data centers are becoming a campaign issue.

The numbers are brutal for Big Tech:

  • 57% of registered voters say AI’s risks outweigh its benefits (NBC News)
  • 36% of Americans say data centers do more harm than good
  • More Americans believe they’ll personally lose from data center construction than gain (30% to 25%)

Georgia AG Chris Carr, running in a Republican gubernatorial primary, says he gets asked about data centers “everywhere I go.” Political strategists call it a perfect storm — cost of living anxiety colliding with AI insecurity, transforming a once-local issue into a national debate.

The issue reached the White House, where President Trump assembled Big Tech CEOs to demand they bear the cost of powering their facilities rather than passing it to ratepayers.

Big Tech’s Self-Inflicted Wound

The industry is caught in a bind of its own making. Amazon alone plans $200 billion in 2026 capex. These companies need the data centers. The AI models they’re building require exponentially more compute with each generation.

But speed outpaced everything — community engagement, environmental review, grid planning. The standard playbook of generous tax incentives, utility NDAs, and moving fast has backfired spectacularly.

Communities sold on jobs and tax revenue discovered the reality: data centers employ relatively few people, consume enormous resources, and can transform neighborhoods overnight. Maryland Governor Wes Moore put it plainly: “No one wants to see what we’ve seen in Northern Virginia, which is, essentially, where they just built them out in a very unregulated way, without any community input.”

What Comes Next

This isn’t going away. As models get larger and demand more compute, infrastructure buildout intensifies — and so does resistance.

Expect a patchwork of state regulations making some states friendlier than others, a gradual end to the most generous tax incentives, and increasing requirements for community engagement and environmental review. Some states will keep rolling out the welcome mat. But the broader trend is unmistakable.

The AI industry spent three years optimizing for speed and scale. Now it’s learning that social license matters too. You can build the most powerful AI in the world, but you still need somewhere to plug it in — and the people who live near the outlet get a vote.