OpenAI Pledges to Pay Full Power Costs of AI Campuses After Voter Backlash
The company's new community compact promises energy and water disclosure and fewer secrecy deals, but sets out no penalties, dispute process or independent checks.

OpenAI has pledged that people living near its future AI facilities will not end up paying for the electricity those sites consume. The commitment came on Thursday, two days after voters in several states delivered a setback to data centre expansion at the ballot box.
The company also promised to publish energy and water figures for its sites and to restrict the use of confidentiality agreements with public officials, according to an announcement reported by Decrypt.
What OpenAI has promised
The commitments sit in a document OpenAI calls its National Community Compact. It applies to future campuses, the company's preferred term for AI data centres, that OpenAI both owns and operates.
The central pledge concerns electricity. OpenAI said it would cover the full cost of powering its compute campuses and would ensure that local residents do not bear the burden of its energy needs.
Beyond the headline promise on power, the compact includes several further undertakings:
- Financial assurances before expensive grid upgrades go ahead, intended to protect other customers if a project is scaled back or abandoned
- A water plan for each AI data centre
- Annual public progress reports for every campus
- Independent studies of noise
- Limits on non-disclosure agreements with officials
OpenAI said each future campus it owns and operates will also receive its own local compact alongside the annual report.

A wider ask of lawmakers
Alongside the compact, OpenAI released a separate policy blueprint aimed at legislators and regulators. Where the compact binds only the company, the blueprint asks governments to impose comparable requirements on the industry as a whole.
On electricity, the blueprint goes further than OpenAI's own pledge. It calls on policymakers to instruct state public utility commissions to establish large-load tariffs, under which major electricity users would pay the additional costs they create for the system.
A large-load tariff is a dedicated electricity rate for very heavy consumers, such as data centres. The aim of such tariffs is generally to stop the cost of new generation, transmission or substations built for one large customer from being spread across ordinary household and business bills.
The blueprint also recommends mandatory reporting, tight restrictions on officials signing NDAs and provisions that would let governments claw back tax incentives when developers fail to meet the commitments they agreed.
Why it matters now
The timing is significant. The announcement followed what Decrypt described as a bruising election for the data centre industry, in which voters in multiple states supported restrictions on the facilities and candidates advocating closer oversight won important contests.
The message from those results, as Decrypt characterised it, is that communities want more influence over how these facilities are built and how their operations affect neighbours. Concerns about data centres typically centre on their heavy demand for electricity and water, the noise they generate and the incentives local governments offer to attract them.
Electricity costs have become a particularly sensitive issue. When a utility must expand its network to serve a single very large customer, the question of who pays for that expansion can determine whether residential bills rise. OpenAI's promise of financial assurances before grid upgrades speaks directly to the risk that infrastructure built for a data centre is left underused, with the bill passed on to other ratepayers.
The limits on NDAs address another recurring complaint. Confidentiality agreements between developers and local officials can keep residents in the dark about a project until key decisions have been taken. Publishing energy and water figures would, in principle, give communities data they currently struggle to obtain.
By pairing a voluntary compact with a call for industry-wide rules, OpenAI appears to be positioning itself as a willing participant in tighter oversight rather than an opponent of it. A blueprint that applies the same requirements to all developers would also mean rivals face the same costs.
The enforcement gap
For all its detail, the compact leaves important questions unanswered. It does not set out penalties if OpenAI fails to meet its commitments, nor does it establish a process for resolving disputes. There is no provision for independent verification of the electricity and water figures the company publishes.
That means residents would be able to read OpenAI's disclosures but, on the basis of the document, would have no clear route to force corrections or to recover losses if the promises were not kept. In effect, the compact relies on the company's own reporting and reputation rather than on an external enforcement mechanism.
The scope is another limitation. The compact covers only campuses that OpenAI itself owns and operates. Much of the infrastructure reserved for the company is built and owned by partners, including Oracle, according to Reuters. OpenAI has not said that its national compact automatically extends to those partner-owned facilities.
If a substantial share of the computing capacity OpenAI relies on sits in buildings it does not own, the practical reach of the pledge may be narrower than the headline suggests. Communities near a partner-owned site serving OpenAI could find that the commitments do not formally apply to them.
The blueprint's recommendations may partly address these gaps, since tax clawbacks and mandatory reporting would carry legal force if adopted. But those measures depend on legislators and regulators choosing to act, and the compact itself does not commit OpenAI to anything beyond its own undertakings.
What to watch next
The first test will be the local compacts OpenAI has promised for each future campus. Their content will show whether the national pledges translate into specific, measurable obligations at individual sites, and whether any include enforcement terms absent from the national document.
The annual progress reports will be another marker. Their usefulness will depend on how detailed the energy and water data are and whether any outside party is able to check them.
Observers will also be looking for clarity on partner-owned infrastructure. A statement from OpenAI on whether facilities built by Oracle and others fall under the compact would significantly affect how far its commitments extend.
Finally, the policy blueprint's fate rests with state utility commissions and legislatures. Following an election in which voters signalled discontent with data centre growth, lawmakers may be receptive to proposals such as large-load tariffs and tax-break clawbacks. Whether they adopt OpenAI's framework, go further, or pursue their own approach will shape the costs and conditions facing the entire industry.
For now, the compact represents a public statement of intent. How much weight it carries will depend less on the wording of the pledge than on whether anyone beyond OpenAI is in a position to hold the company to it.
More from peatpost

US Treasury Sanctions Eight Canadian Money-Service Firms Over Iran Crypto Links
Washington says the firms swapped rials for digital assets on behalf of sanctioned Iranian exchanges and IRGC-linked wallets, in action coordinated with Canada's FINTRAC.

OCC Issues First AI Supervisory Framework, Ranking Autonomous Payment Agents as Highest Risk
The 22-page guidance tells bank examiners to judge AI by the harm a failure could cause, not by technical sophistication, and flags outside agents connecting to accounts.

CFTC Chair Cites $6.4bn Binance Hack in Push for US Oversight of Offshore Exchanges
Michael Selig says the record breach supports the agency's proposed crypto rules, though Binance's own account of the attack is more complicated than his.
Discussion
0 commentsNo comments yet — be the first to weigh in.