OpenAI formalises future U.S. data-centre safeguards; pledges union engagement, environmental disclosures

OpenAI has announced a set of commitments for communities hosting its future AI data-centre campuses in the United States, pledging to bear the full cost of electricity and grid infrastructure, disclose environmental impacts and not use non-disclosure agreements (NDAs) with public officials, subject to limited exceptions.

In an announcement on November 5, 2026, the company published its National Community Compact, setting out commitments for future computing campuses it owns and operates in the U.S., and a separate Compute Campus Policy Blueprint proposing measures for governments to make data-centre developers more accountable to local communities.

“We’ll pay the full cost of the electricity we use and the infrastructure needed to serve us,” OpenAI said. It added that it would provide financial assurances to protect other electricity consumers if a project were delayed, reduced or cancelled.

The electricity commitment builds on assurances the company had made for individual projects, including its Stargate initiative. The latest announcement seeks to establish a common set of commitments for future campuses it owns and operates in the U.S.

OpenAI also committed to developing a water plan for each campus, prioritising closed-loop or low-water cooling systems, and publishing site-level water and electricity consumption. It said it would publish construction plans and independent noise studies before operations begin and annually thereafter.

The company said it would not use NDAs with public officials, with narrow exceptions for legitimate security information, trade secrets, and proprietary or competitively sensitive information.

On employment, OpenAI said it would publish expected job creation, work with unions and training providers to prepare local residents for employment opportunities, and report annually on hiring and apprenticeships. It also pledged to make sustained investments in priorities identified by local communities.

Alongside these commitments, OpenAI released a Compute Campus Policy Blueprint proposing measures for federal, state and local governments to make data-centre developers more accountable. The blueprint calls for large electricity consumers to pay for the additional grid infrastructure and power costs they create, and proposes measures to protect consumers if projects are abandoned.

It also calls for greater transparency on tax incentives offered to data-centre developers, including measurable commitments and mechanisms to recover benefits if companies fail to meet them.

Citing its planned campus in Effingham County, Georgia, OpenAI said its agreement with Georgia Power allows the utility to reduce electricity supplied to the facility by up to 1,000 megawatts during periods of high demand.

The announcement comes against a backdrop of growing political and public opposition to data-centre expansion in the U.S., an issue that featured in the November 3, 2026 midterm elections. However, OpenAI’s chief global affairs officer Chris Lehane told Reuters that the policy measures had been in development since August, before the elections. The company had also begun outlining community commitments under its Stargate initiative in January. The issue is also becoming relevant in India, which is attracting large investments in AI infrastructure.

Here, OpenAI has partnered with Tata Group and TCS to develop AI infrastructure in India through TCS’s HyperVault business, with an initial capacity of 100 MW and the potential to scale to 1 GW. OpenAI will be the first customer. Meta has agreed to lease capacity at a 168 MW AI-enabled data centre that Reliance Industries is developing in Jamnagar, Gujarat. Google is also developing a gigawatt-scale AI hub in Visakhapatnam in Andhra Pradesh, as part of a planned $15-billion investment in India.

These projects have also begun facing local opposition. According to The Hindu reports, civil society groups in Visakhapatnam are demanding a moratorium on new hyperscale data-centre approvals, citing concerns over their environmental and social impact. An environmental NGO Green Visakha also has demanded for the cancellation of permissions granted to data-centre projects in Visakhapatnam and Anakapalli districts.

Residents have also opposed the proposed AI data centre in the Balkum Pada area of Thane. On September 8, 2026, Shiv Sena (UBT) leader Aaditya Thackeray warned that the project could affect residents’ quality of life, citing concerns over noise, heat emissions, electricity and water consumption.