Amazon Drops NDAs as Data Center Backlash Grows

Cartoon illustration of a data center and community protest about NDAs

Amazon’s recent move to stop using nondisclosure agreements (NDAs) with government agencies marks a tactical shift in a widening dispute over the role of large data centers in local communities. Announced by AWS CEO Matt Garman in a blog post, the change comes as cities and states across the U.S. consider moratoriums and tougher scrutiny of new facilities. The announcement is short but strategic: Amazon is trying to blunt a narrative about secrecy and to position data centers as economic assets rather than hidden liabilities. Whether that will be enough to rebuild trust is an open question.

What sparked the backlash

Community pushback against data centers has intensified in recent months. Critics—from environmental activists like Erin Brockovich to local organizers—point to a pattern of surprise announcements, permitting before public notice, and agreements that limit what local officials can share. Those dynamics have contributed to high-profile policy responses, including New York’s one-year moratorium on large data center permits and more than 100 proposed moratoriums or similar measures under consideration nationwide. At their core, these objections are less about server racks than about transparency, local control, and long-term environmental impact.

Amazon’s response: NDAs and the broader defense

In his post, Matt Garman directly addressed the secrecy complaints, stating that Amazon no longer uses NDAs with government agencies on its projects. That pledge is paired with a broader defense of data centers’ community contributions—Amazon highlights more than $1 billion in community investments over three years—and an effort to rebut common claims about water use, energy costs, pollution and local benefit. The NDA change is a symbolic concession: it doesn’t resolve every concern, but it acknowledges that earlier practices helped fuel distrust.

Separating claims from context: four contested points

Amazon frames its rebuttal around four frequently raised concerns. It’s worth looking at each with care.

Water usage

Amazon notes that “direct data center water consumption” represents a small fraction of U.S. industrial water use—about 0.5%—and argues other local industries (golf courses, almond farming) use far more. Critics counter that water impacts should be understood holistically: water used for electricity generation and chip manufacturing ties into data center demand, and independent study is needed because reporting standards are inconsistent across states and companies.

Electricity rates

Garman said electricity prices have risen in some states with many data centers but have fallen or increased more slowly in others; where rates moved higher, he blamed aging grids and lack of prior investment. Independent analyses, however, point to concentrated demand spikes on major grids that can correlate with data center growth and may drive rate pressure for other consumers if not properly planned and mitigated through grid upgrades.

Pollution and generator emissions

Amazon emphasizes that on-site backup generators are idle the vast majority of the time—“99.9% of the time,” per Garman—and therefore permitted emission caps don’t reflect routine operations. But permit limits do set potential maximums, and some planned centers have unusually high permitted emissions on paper, raising alarm among residents and environmental groups about worst-case scenarios and cumulative impacts.

Community benefits

Amazon highlights its monetary contributions to area communities, positioning data centers as drivers of jobs and investment. Yet community groups frequently say the gains are uneven: data centers can bring construction jobs and tax revenue but fewer long-term high-wage positions, and perceived or real opacity around siting decisions can undercut claimed benefits.

The deeper problem: a crisis of trust

A core theme beyond technical tradeoffs is trust. Observers of the AI and tech landscape—like Anthropic’s Dario Amodei—have framed current public reactions as a “crisis of trust,” where communities assume institutions are acting against their interests. Even factual rebuttals can fail if they don’t address how decisions are made, who is included in the conversation, and whether communities can independently verify claims. For many opponents, transparency and meaningful local engagement matter more than aggregated statistics.

What policymakers and companies can do next

If the goal is to move from confrontation to constructive planning, a few pragmatic steps follow from the debate:

Commit to transparent permitting and public timelines so communities aren’t blindsided.
Standardize reporting on water and energy impacts so independent researchers can assess cumulative effects.
Tie community benefits to measurable outcomes—workforce development, long-term tax-sharing, environmental mitigation—and make those conditions of approval where appropriate.
Coordinate with utilities and regulators early so grid and water investments precede major demand increases.

Conclusion

Amazon’s decision to stop using NDAs with government agencies is a notable concession, but it is only one piece of a larger puzzle. Technical claims about water, energy, and emissions matter, and so does the process by which projects are planned and communicated. For communities wary of rapid technological expansion, assurances without verifiable transparency will likely fall short. The coming months—where moratoriums, local hearings, and company-community negotiations play out—will show whether policy, practice, and public engagement can close the gap between corporate claims and community confidence.