Amazon says it has dropped NDAs as data center backlash grows

Amazon Web Services says it has stopped using nondisclosure agreements with government agencies as it seeks approval for new data centers, a notable shift as public resistance to the industry intensifies. The move comes amid a growing backlash over transparency, water use, electricity prices, pollution and local economic tradeoffs tied to AI infrastructure.

In a blog post published on October 3, AWS chief executive Matt Garman argued that data centers are widely misunderstood and can bring meaningful benefits to surrounding communities. He also directly addressed one of the most sensitive issues in the debate: the use of NDAs during siting and permitting discussions. According to Garman, Amazon no longer uses those agreements with the public agencies it works with on projects.

The timing matters because local opposition to large data centers has turned into a broader political and regulatory fight across the United States. New York has already imposed a one-year moratorium on permits for large data centers, and Garman said more than 100 similar moratoriums are under consideration nationwide. That suggests the industry’s expansion could face real delays just as demand for AI computing keeps rising.

Amazon’s response is the latest attempt by a major cloud and AI infrastructure player to reframe the conversation around data centers. But critics remain skeptical. Community advocates and environmental groups have increasingly argued that residents are left in the dark until projects are too far along to influence, creating distrust that no corporate messaging campaign can easily reverse.

Why Amazon is talking about NDAs now

Amazon’s acknowledgment that it no longer uses NDAs with government agencies is significant because secrecy has become one of the main flashpoints in the data center debate. Opponents say local officials are often asked to sign confidentiality agreements before residents even know a project is being considered, limiting public scrutiny of environmental and economic impacts.

That concern has helped turn data centers from a niche zoning issue into a national political problem. The facilities are essential for cloud computing and AI services, but they can also require substantial power, water and land resources. As more communities confront proposed projects, the balance between economic development and transparency has become harder to manage.

Amazon says it no longer uses nondisclosure agreements with the government agencies it works with on data center projects, a change it presented as part of a broader effort to address public concerns.

Garman’s statement appears designed to signal that Amazon wants to be seen as more open than some of its peers. Still, the disclosure came inside a much longer defense of the data center industry, suggesting the company is not only trying to reassure regulators but also trying to shape the broader public narrative.

How big is the backlash against data centers?

The backlash is large enough that it is now affecting permitting and legislation. In New York, the state enacted a yearlong pause on permits for large data centers, and Garman warned that more than 100 moratoriums are being considered elsewhere in the U.S. If even a portion of those proposals advances, developers could encounter slower approvals, additional hearings and tighter environmental review.

The controversy is not limited to one state or one complaint. Instead, it has spread across issues that matter to local residents: whether a campus will strain the power grid, whether it will raise household electric bills, whether it will increase pollution and whether nearby water supplies will be affected. Those concerns have made opposition unusually durable.

Environmental activist Erin Brockovich has said transparency is the top complaint she hears from communities dealing with data center proposals. She described a pattern in which projects are announced after permits are secured, developers do not return calls and local officials sign NDAs before residents know a project is under discussion. Her comments capture the sense of exclusion many opponents say they feel.

What community critics say they want

Critics are not only asking whether a data center should be built. They are also asking how the project will be governed. In practice, that means demanding:

• earlier disclosure of proposed sites and timelines
• clearer environmental reporting
• public details on power and water use
• independent analysis of local grid impacts
• more realistic accounting of jobs and community benefits

Those demands reflect a broader shift in public expectations. In many places, residents no longer accept the idea that a technology project should be approved first and explained later. That tension is at the center of the current backlash.

What myths does Amazon say it is trying to correct?

Amazon says it is trying to correct four common criticisms: that data centers use too much water, that they drive up electricity costs, that they produce excessive pollution and that they provide little value to the communities hosting them. Garman framed these as misconceptions that distort the public conversation.

The company’s argument is that, taken in context, the industry’s footprint is smaller than many people assume and that local economic benefits are often overlooked. But each of those claims is complicated, and in some cases the available data is incomplete because reporting standards are uneven.

Water use: what Amazon claims

Garman pointed to an Amazon report to argue that direct water use by data centers is modest relative to industrial water demand overall. He said direct data center water consumption accounts for 0.5% of U.S. industrial water use, which he described as far lower than sectors such as golf courses or almond farming.

That framing is intended to minimize the impression that data centers are one of the country’s biggest water users. Yet critics argue the picture is incomplete. Some of the water burden is indirect, tied to electricity generation and semiconductor manufacturing, and researchers have said there is still too little standardized reporting to compare facilities accurately.

The issue has become more acute as AI infrastructure scales up. Even when cooling systems use less water than expected on-site, the wider supply chain can still impose costs that are difficult for communities to see in advance.

Electricity costs: what Amazon says

Amazon argues that data centers are not automatically responsible for higher power prices. Garman said electricity rates have risen in some states with heavy data center concentration, but have declined or risen more slowly in others. In his view, where costs are climbing, the real problem is an aging grid that was not expanded quickly enough before new demand arrived.

That explanation clashes with some independent analysis. A watchdog recently said data centers were the primary driver of a 76% year-over-year price increase on the country’s largest electrical grid. The gap between those interpretations shows why local communities remain unconvinced by broad industry assurances.

For residents, the key question is not whether rates can vary from state to state, but whether the arrival of a new power-intensive facility will make their bills more volatile or force upgrades they ultimately help pay for.

Pollution: what Amazon says

On pollution, Garman said critics often focus on the maximum emissions allowed under permits rather than the amount data center generators actually produce. In one example cited in the debate, a planned Amazon facility in Texas is permitted to emit up to 33 million tons of carbon dioxide annually, more than any other power plant in the United States.

Amazon’s response is that emergency generators and backup systems rarely run for long. Garman said they are idle 99.9% of the time and operate for roughly 10 hours a year, mainly for maintenance tests. That defense may be technically accurate in many cases, but it does not answer the broader concern about the scale of the permitted emissions envelope.

Garman said the company believes critics are overstating the pollution impact because backup generators almost never run, even though permit limits can look extremely large on paper.

The discrepancy between permitted maximums and everyday operations is one reason the pollution debate is so difficult. Regulators, environmental groups and companies are not always arguing about the same thing.

Are data centers really good for communities?

Amazon says yes, but the answer depends on what a community values and how much trust it places in the developer. Garman argued that data centers can generate local economic activity and provide direct community support, while also serving as the backbone of cloud services and AI applications used across the economy.

In his post, he said Amazon has contributed more than $1 billion to communities across the U.S. over the past three years in places where it has a significant data center footprint. The company presents that figure as evidence that the facilities do more than consume resources; they also bring investment, donations and infrastructure-related spending.

Yet critics say community benefits often arrive after the major decisions are already made. If the public feels shut out of the process, financial contributions may not be enough to overcome the suspicion that the project was designed for corporate gain first and local benefit second.

What counts as a benefit?

When companies talk about community benefits, they may mean any combination of:

• direct charitable contributions
• local construction and maintenance jobs
• property tax revenue
• infrastructure upgrades
• economic spillover from supplier spending

But those benefits are not always evenly distributed, and they can be difficult for residents to evaluate against the potential downsides. A neighborhood near a data center may see little of the long-term upside while still bearing the environmental and infrastructure costs.

That is why the debate over benefits is increasingly tied to process. Communities that trust the planning process are more likely to accept a project’s tradeoffs. Communities that feel excluded are far more likely to resist them, regardless of the size of the company’s commitments.

Why trust has become the central issue

Trust is now the defining problem in the data center fight. Even when companies present technical data or economic arguments, many residents assume the figures are selected to make the project look better than it is. That skepticism is not limited to Amazon; it reflects a broader reaction to how large technology projects have often been introduced.

Anthropic chief executive Dario Amodei recently described the backlash around AI as fundamentally a trust crisis. In his view, many people believe governments and tech firms are always looking for a new way to take advantage of them. That sentiment helps explain why even detailed rebuttals from industry leaders often fail to persuade local opponents.

According to Anthropic’s Dario Amodei, the opposition to AI infrastructure is rooted in a “crisis of trust,” with many people assuming companies and governments are acting against their interests.

Writer Jasmine Sun has made a similar observation, saying that when opponents hear the tech industry’s explanations, the usual reaction is simple disbelief. That reaction may be the hardest obstacle for Amazon and its peers to overcome. A company can revise its policies and publish reports, but trust is built slowly and lost quickly.

What happens next for Amazon and the industry?

The immediate question is whether Amazon’s pledge on NDAs and its broader defense of data centers will change any minds. It might help with some regulators, especially if the company can demonstrate a more open permitting process. But in many communities, the debate is already about more than disclosure.

Local leaders are now weighing whether the economic promise of AI infrastructure justifies the potential stress on power systems, water resources and air quality. At the same time, state and local governments are facing pressure from residents to slow down or pause approvals until they can better understand the impacts.

That means the industry may be entering a more difficult phase. As AI demand continues to rise, companies need more data centers, more transmission capacity and more political goodwill. Yet the very scale of that need is making communities more alert, not less, to what is being proposed in their backyards.

For Amazon, the message in Garman’s blog post is clear: the company wants to be seen as a responsible partner, not a secretive outsider. Whether the public believes that distinction will likely depend less on one policy change than on what comes next in permitting fights across the country.

Key facts at a glance

Bottom line

Amazon’s decision to say it has dropped NDAs in government data center talks is both a defensive move and a political signal. The company is trying to show that it has heard the criticism, but the larger fight over data centers is now about trust, transparency and whether the public believes the benefits outweigh the costs.

As the AI boom accelerates, that fight is likely to intensify. Companies want faster approvals, communities want more information and regulators are caught between economic development and public concern. Amazon’s latest response may soften some criticism, but it is unlikely to end the backlash on its own.