AI Data Centers vs Local Backlash - Will Ending NDAs Stop Moratoriums

TL;DR

• Dozens of cities and counties across the U.S. have passed data center moratoriums or strict zoning pauses in 2025-2026 as AI infrastructure plans collide with concerns over power bills, water use, and secret deals.
• At the center of the backlash is secrecy: tech giants and developers routinely ask local officials to sign NDAs, fueling mistrust that projects are decided before communities get a say.
• Dropping NDAs alone won't stop moratoriums, experts argue — the AI industry will also need transparent energy and water commitments, and real community benefits, to rebuild trust.

THE MORATORIUM WAVE IS NO LONGER FRINGE

What started as a handful of zoning fights in Northern Virginia and rural Georgia has become a national movement. From Tucson, Arizona to Prince William County, Virginia, from Indiana townships to suburbs outside Columbus, Ohio and Atlanta, local governments are hitting pause on new AI data centers.

The numbers tell the story. Industry trackers now count more than 100 local moratoriums, interim bans, or de facto pauses on data center approvals nationwide, with dozens more considering new setbacks, noise rules, and special-use permits. Even in traditionally pro-development states like Texas and Arizona, councils are voting for 6- to 12-month timeouts to study grid and water impacts.

The driver isn't anti-tech sentiment. It's scale and speed. A single hyperscale AI campus can now demand 500 megawatts to over a gigawatt — enough to power hundreds of thousands of homes — with construction timelines that local utilities and planning boards say they were never built to handle.

ENERGY, WATER, AND HIGHER BILLS: WHY NEIGHBORS ARE PUSHING BACK

On Equity this week, the conversation kept returning to the same kitchen-table issues: who pays for the power, and who gets the water.

Residents packing town halls from Michigan to Mississippi cite the same fears. Data centers could strain the local grid, drive up residential electricity rates to fund new transmission, and drain millions of gallons a day for cooling in drought-prone regions. Noise from backup generators, loss of farmland, and 24/7 construction traffic only add to the anger.

Utilities are starting to blink. Dominion Energy in Virginia, Georgia Power, and utilities in Ohio and Indiana have all faced regulatory fights over whether Big Tech should pay its own way for new substations and generation. Consumer advocates warn of a scenario where families subsidize AI cloud profits through higher monthly bills, while developers secure discounted industrial rates locked in for decades.

Water has become equally explosive. While Microsoft, Google, Amazon and Meta all pledge to be water-positive by 2030 and tout closed-loop cooling, locals say project filings are vague, with actual consumption figures often redacted or reported only after approval.

THE SECRECY PROBLEM: CODE NAMES, NDAS, AND DONE DEALS

If energy is the substance of the backlash, secrecy is the spark.

It has become standard practice for data center developers to pursue sites under code names like Project Blue or Project Sail, requiring city council members, county commissioners, and economic development officials to sign non-disclosure agreements before they can even see who is behind a proposal.

Supporters say NDAs are normal for competitive site selection. Critics say they poison the process. Residents learn about a 300-megawatt campus only after tax incentives are negotiated, land is optioned, and utility interconnection requests are filed. By the time there is a public hearing, opposition feels performative.

That dynamic was central to the Equity debate: can the AI industry rebuild trust while still demanding silence from the elected officials supposed to represent the public.

DROPPING NDAS: A FIRST STEP, NOT A FIX

There are signs of a shift. Microsoft President Brad Smith recently acknowledged the industry has a communication problem and said Microsoft is rethinking its use of NDAs with local governments. Google and Amazon executives have made similar noises about earlier community engagement, open houses, and publishing water and energy impact assessments upfront.

Transparency advocates welcome the rhetoric but say ending NDAs won't stop moratoriums on its own. A moratorium is often the only leverage a small town has to force a pause and get data.

What communities actually want, they say, is plain-language disclosure before deals are signed: how much power and water will this use at full build-out, who builds the substations, will residential rates rise, what tax breaks are on the table, and what happens if the developer walks away.

Some developers are testing a new playbook — community benefit agreements, funding for grid upgrades, paying full cost of power infrastructure, committing to bring-your-own-power gas plants or nuclear co-location, and air-cooled designs that slash water use. But those promises remain voluntary and patchwork.

WILL MORATORIUMS SLOW THE AI BUILDOUT

For now, Wall Street and Silicon Valley are watching closely. Hyperscalers still plan to spend hundreds of billions on AI infrastructure in 2026, with OpenAI's Stargate project, xAI's Memphis expansion, Meta's Louisiana campus, and Amazon's nationwide cluster buildout all racing for capacity.

Moratoriums don't kill demand — they redirect it. Developers are shifting to friendlier counties, industrial megasites, and states offering fast-track permitting. But that game of whack-a-mole adds delay, cost, and political risk at the exact moment AI labs say they need compute yesterday.

The Equity takeaway was blunt: the AI industry treated data centers like invisible cloud infrastructure for a decade. They are now highly visible, highly physical neighbors. Until companies trade code names and NDAs for open books on energy, water, and bills, expect more pauses, more lawsuits, and more local backlash — even if the AI boom keeps building.

AndroGuider Team

Articles written by the AndroGuider team. We try to make them thorough and informational while being easy to read.