The rise of artificial-intelligence applications has triggered a surge in large-scale data-center construction across the United States, but a striking feature of this boom is how much of it is hidden behind non-disclosure agreements (NDAs). These contracts, increasingly used by technology companies and local governments, are keeping important details about the development of AI data centers out of the public view.
The Opaque Role of NDAs in Data-Center Development
In many localities, officials negotiating with tech firms for large AI-focused data-center campuses enter into broad confidentiality agreements that bar disclosure of critical information such as the operator’s identity, power and water usage, or infrastructure demands. For example, a review found that in Virginia 25 out of 31 localities with data-center proposals had NDAs with firms, often prohibiting release of nearly any non-public information.
Residents in Kentucky, Oklahoma and Minnesota similarly report being left in the dark while multi-acre data-center projects advance under code names or hidden developers.
Writings in publications such as The Progressive report that tech firms are treating water and electricity consumption data of AI data centers as trade secrets, bolstering their case with NDAs.
Implications for Transparency, Democracy and Infrastructure Cost
The widespread deployment of NDAs in this sphere raises serious questions for democratic participation at the local level. When local officials cannot share who is developing what, or what the environmental and infrastructural impacts are, residents and stakeholders lack the information needed for meaningful debate. As Bonds and Newby argue, the secrecy “curtails meaningful debate and impairs local democracy”.
At the same time, the hidden nature of these deals has consequences for utility planning and financial burdens. Some AI data centers consume energy at levels analogous to entire cities, but because usage data are withheld via confidentiality agreements, local utilities and regulators struggle to assess whether grid upgrades or tax breaks are justified. Additionally, communities may be exposed to water-use or environmental risks without ever knowing the magnitude. For example, in Wisconsin a firm sought trade-secret status for its projected use of eight million gallons of Lake Michigan water annually for an AI-data center.
Broader Strategic and Economic Stakes
Beyond local governance, there is a national strategy dimension. Rapid deployment of hyperscale AI data centers is part of the U.S. race to build computing infrastructure, and secrecy can help firms move swiftly ahead of competitive peers. However, when the physical location, capacity, and resource draw of such centers are obscured, public policy has less visibility into how the AI economy will reshape energy, water, land and labour markets. Scholars studying “sovereign compute” note that the national-scale geography of these facilities matters for digital sovereignty, yet data often remain self-reported and opaque.
The hidden nature of these agreements also has implications for equity: neighboring residents may experience increased noise, land-value shifts or utility cost hikes without ever receiving full disclosure of what they consented to. Studies report farmer communities and homeowners raising concerns in places like Missouri and Oklahoma.
A Final Note: A call Call Recalibration of Transparency
The expanding footprint of AI-driven data centers is reshaping the built environment and public-resource usage in Americans’ communities. But as the evidence shows, NDAs and broad confidentiality regimes are being used to shield vital information from the public, with implications for democratic oversight, infrastructure cost-sharing and environmental justice. If AI infrastructure is to proceed in the public interest, there is a strong case for rebalancing secrecy with standardized disclosure regimes and more open negotiation processes.
