A new kind of contract for AI infrastructure projects and communities

Rather than killing data center construction projects, the American public is conditioning them on social license to operate. But community acceptance is multifaceted, more qualitative than quantitative, and can fluctuate across a site lifecycle. A crucial question for AI infrastructure sponsors, investors, developers, and operators is: how do we measure the quality of the project’s relationship to local stakeholders? And can we benchmark that against peer companies at other sites?

It matters to assess relationship quality rigorously because social license is not, by its nature, quantitative. Far from simply a number to be optimized, social license is not earned through a calculated transfer of costs and benefits. Study of public response to data center construction projects supports this.

There are problems with procedure: local residents feel “railroaded” by timelines and decision-making processes that leave them with little or no influence. It follows that a shower of gifts from the project to residents in the form of grants, schools, workforce programs, and the like will address only a narrow part of the problem.

In public meetings I’ve observed in San José and Gilroy -- both cities in California -- residents have said they “want a say in what happens in our hometown.”* They are “against … obfuscation of public process.”** And like communities in other states, they want to be consulted, not simply told. So while it’s true that media reports and polls are conveying concerns that Americans have about measurable changes to environment and economy, a negotiation over these metrics alone will fail to address the wider social risk. See more of my thesis on the culture conflict for data centers here

Investment implications

Americans’ drive for self-determination in these controversies has critical business implications for data center sponsors, investors, developers, and operators:

• The business costs of maintaining social license are more than a simple economic transfer to communities.

• One of the most overlooked costs of conflict can be corporate staff time – particularly senior management time – devoted to managing social risk.***

• Investors invite a charade of cherry-picked data if they ask companies to disclose numbers that represent community relationship quality or trust.

• “Buying” community support will likely never be effective for preventing or mitigating conflict over the long term.

A fresh approach to legitimacy

In the face of these challenges, there is a promising way forward: a negotiated accord, often called a Good Neighbor Agreement, to customize new institutions for sharing decision-making between companies and communities, each suited to local context. It records mutual commitments between a project developer or operator and a local community about how the project lifecycle will be conducted and how the community will engage with it. In other words, it institutionalizes dialogue.

There’s no need to start from scratch – borrow from sectors with a track record of elevated social license risk. Significant land-use projects in energy and mining provide examples of Good Neighbor Agreements and other community benefits frameworks. The latter is an umbrella term for both binding and non-binding agreements that direct benefits from a project to local communities. Good Neighbor Agreements is the name of one type, although practices vary and there are no universal distinctions between types.

A widely cited reference point is the Good Neighbor Agreement executed by Sibanye-Stillwater Mining Company with three grassroots organizations in 2000 to facilitate new mining projects in Sweet Grass County, Montana. As corporate and community needs changed, it has been amended six times and remains enforceable. Since its first signing, there have been no arbitration actions or environmental litigation, a meaningful achievement in any mining context.

• Parties: Sibanye-Stillwater Mining Company, Northern Plains Resource Council, Stillwater Protective Association, Cottonwood Resource Council

• What does this Good Neighbor Agreement do?

  • Directs the parties to establish oversight committees, which are funded by the mining company, with voting power split between the mining company and the local councils.

  • Establishes procedures for committee appointments, removals, decision-making, meeting cadence, dispute resolution, scope of work (e.g., oversee the terms of the agreement; monitor the company’s compliance; oversee all audits, plans, studies, and monitoring required by the agreement).

  • Establishes right of the local councils to inspect facilities and their related obligations.

  • Creates additional covenants related to traffic plans; company-funded environmental audit program; reclamation plan; Comprehensive Surface Water, Ground Water, and Aquatic Resources Protection Plan (including citizen sampling); and other topics.

The earliest published example of a data center-specific accord was a Community Benefits Agreement signed by the City of Lancaster, Pennsylvania and Chirisa Technology Parks (a developer) in 2025. While it may be instructive for future negotiations for other sites and parties, the procedure by which it was negotiated has received criticism from local groups about inadequate community input. An even-handed critique of this agreement was published by Columbia Law School’s Sabin Center for Climate Change Law in 2026.

Analyses from the Sabin Center and other organizations such as the World Resources Institute enable firms and communities to study which aspects of Good Neighbor Agreements are markers of relationship quality between the parties.In turn, strategic advisors can help develop assessment questions to uncover the strength of relationship quality for project-specific and comparative evaluations. In the advisory space, one interlocutor who influences my work is Thomas Gaultier, creator of the Social Accord Architecture.

Assessment questions for project sponsors, investors, developers, and operators

For AI infrastructure projects, negotiated agreements with communities are literal, tangible, auditable, and benchmark-ready. They stand for relationship quality by means of questions that can be answered, assured, and compared across companies. Examples are:

Pre-construction

• Is a process being performed to negotiate a legal agreement with the community?

• Have the community’s parties to the legal agreement been identified?

• Is the project developer or sponsor providing funds for community groups to cover the expenses of negotiation with independently chosen legal representation and third-party experts?

Construction to Post-Closure

• Does an executed agreement exist?

• Was it co-designed with local stakeholders?

• Is it being performed?

• Does the agreement contain enforcement mechanisms?

• Does the agreement include provisions for amendment and renewal, including in cases of asset ownership transfer?

Glyptique Consulting LLC helps clients assess social license to operate in the data center sector by evaluating performance against emerging industry trends; designing communication strategies for developers, operators, and investors; and advising on how firms can become leaders in community relations practice.

Schedule a conversation to discuss how social license to operate applies to your portfolio and which strategic moves would strengthen community relations positioning at your firm or investees.

August 26, 2026

Endnotes

* Gilroy, California City Council Meeting (August 17, 2026)

** San José, California Uniform Standard for Data Centers and Large Energy Projects, Community Listening Session (August 19, 2026)

*** The extractive sector is illustrative. An oft-cited paper published by the Harvard Kennedy School encompassing 50 case studies and 45 interviews found that the costs of local conflict most often overlooked by extractive sector companies were indirect costs resulting from staff time being diverted to managing conflict, including in some cases that of the CEO. For one company, the working assumption was that 5% of an asset manager’s time should be spent managing social risk, yet it was in some cases as high as 35-50%. For another, the CEO’s stated reasons for pulling out of problematic joint venture was that the project was responsible for only 10% of the company’s income yet 90% of senior management time. Rachel Davis and Daniel Franks, “Costs of Company-Community Conflict in the Extractive Sector,” Corporate Social Responsibility Initiative Report No. 66, Cambridge, MA: Harvard Kennedy School, 2014.

AWS data center under construction in Gilroy, California (August 3, 2026). Photo: Jordan Famularo