Voters just gave the AI industry a lesson in local government: having enough money to build a data center doesn’t mean your neighbors want one.
Across Tuesday’s midterm elections, communities backed restrictions on data center development, while candidates campaigning against the industry’s expansion scored major wins. WIRED’s election reporting described a backlash stretching from local zoning questions to governor’s races.
Two days later, OpenAI released a new pitch for how these projects should work.
Its National Community Compact sets commitments for future campuses OpenAI owns and operates. An accompanying Compute Campus Policy Blueprint asks policymakers to establish protections across the industry.
The most striking commitment is refreshingly easy to understand:
OpenAI says public officials won’t have to sign an NDA to discuss or negotiate a project.
For an industry asking communities to accommodate enormous new demands on land, electricity, and water, that’s a meaningful place to start.
Residents deserve to know what their elected officials are negotiating before the construction equipment arrives.
The no-NDA pledge deserves attention—and a close reading
The compact puts it plainly: “We will not use non-disclosure agreements (NDAs) with public officials.”
OpenAI says it won’t require those agreements as a condition of discussion or negotiation. It allows narrow protections for legitimate security information, trade secrets, and proprietary or competitively sensitive information.
Those exceptions matter. This isn’t a promise to publish every technical specification or commercial detail. The practical test will be whether confidentiality stays confined to information that needs protecting, while residents can understand the deal being made on their behalf.
The policy blueprint would extend that approach beyond OpenAI, asking lawmakers to prohibit developers from requiring public officials to sign NDAs, subject to similar exceptions.
There’s also useful history tucked into the document: OpenAI says it voided NDAs in Effingham County, Georgia, in July.
So this didn’t all materialize between election night and Thursday morning. OpenAI’s announcement describes the package as an extension of community commitments it began rolling out in January.
The election gives that work considerably more urgency.
My read: OpenAI is recognizing that secrecy makes an already difficult sales pitch waaaay harder. A community can’t meaningfully evaluate a project when basic questions about its future disappear behind confidentiality agreements.
The electricity promise includes what happens if a project falls apart
OpenAI also commits to paying for the electricity its campuses consume and the incremental infrastructure required to serve them.
The more consequential detail may be the promise to provide financial assurances before major spending begins, protecting other utility customers if a project is delayed, reduced, or canceled.
Consider a utility that expands its system for a huge proposed campus. If the campus never reaches its expected size, somebody still has to pay for that infrastructure.
OpenAI is committing to protect other customers from being left with those costs.
That addresses a harder question than whether the company pays its monthly electric bill. It addresses who carries the financial risk of building around its forecasts.
The documents also promote flexible electricity use. OpenAI says its agreement with Georgia Power allows the utility to reduce power delivered to its planned Effingham campus by up to 1,000 megawatts during high-demand periods.
That’s an agreement permitting future reductions, rather than evidence that the campus has already delivered that relief.
As we’ve covered in The Neuron’s reporting on flexible data center power demand, the useful question is how these arrangements perform when the grid actually needs help.
Communities would get more numbers they can check
The compact promises public information about individual campuses, including water and electricity use, construction impacts, economic contributions, and independent noise studies before operations begin and annually afterward.
The blueprint adds some particularly useful reporting distinctions.
Water disclosures should separate initial cooling-system fills from ongoing use, and water withdrawn from water consumed. Economic reports should distinguish projections from measured results, and direct benefits from modeled spillover effects.
That may sound like accounting homework. It’s how residents can tell whether an impressive number describes something that happened or something a consultant hopes will happen.
The same principle applies to jobs. Construction work and permanent operating positions serve different purposes, with different timelines. Reporting them separately gives communities a clearer picture of what remains after the campus opens.
The benefits need to reach people outside the fence
OpenAI points to existing commitments of $80 million for community priorities in Effingham County and $40 million in Pike County, Ohio.
Examples include a $4.8 million grant for free school meals for all 14,000 Effingham students and a joint $6 million childcare commitment with SB Energy in Pike County.
Those are tangible benefits people can understand without attending an AI conference.
They also reflect the larger problem explored in The Neuron’s coverage of who gets the upside from AI: communities want a credible connection between the disruption they experience and the prosperity they’re promised.
A school-meal grant can help families. Residents still need answers about water, noise, electricity costs, and the terms of any tax incentives. Each commitment has to stand up on its own.
OpenAI still wants to build and wants government to help
The blueprint also advances OpenAI’s business interests.
It asks for faster, more predictable permitting, changes to grid connection processes, expanded transmission authority, and restored and expanded clean electricity tax incentives.
That’s the bargain running through the package: stronger community protections alongside policies that make construction easier.
There’s a reasonable case for that combination. Clear requirements can help residents assess a project and help developers understand what approval will require. But the details determine whether faster review gives agencies the resources to make sound decisions or simply increases pressure to approve.
The documents are also different kinds of promises. The compact establishes company commitments for future OpenAI-owned and operated campuses. The blueprint proposes laws and policies. Publishing it doesn’t enact those protections or automatically extend the compact to every partner-operated facility supplying OpenAI with computing capacity.
The next meaningful evidence will come from local agreements, utility arrangements, public disclosures, and what happens when a commitment is missed.
After this week’s election, our read is that community trust has become a constraint on AI expansion that money alone can’t resolve.
OpenAI’s no-NDA pledge gives residents a concrete standard to hold it to. When the next campus comes up for negotiation, their elected officials should be able to explain the deal.
And residents should hear about it while there’s still time to change it.
Related reading
- Data Centers Have an Electricity Problem. This Alliance Could Help. — How flexible power demand could help address grid pressure and community concerns.
- The AI Backlash Is a Fight Over Who Owns the Upside — The broader debate over who shapes AI development and shares its economic benefits.
