Amazon Ends Secret Data Center NDAs as AI Buildout Faces Trust Test

Amazon says it will stop using nondisclosure agreements when discussing data center projects with local governments, joining Microsoft in a retreat from secrecy as backlash grows over the rapid expansion of AI infrastructure. The change matters because resistance to opaque dealmaking has already helped trigger hundreds of proposed and enacted moratoriums on data centers and related projects across the U.S.

The shift lands at a moment when the AI boom is colliding with local concerns about electricity demand, water use, tax incentives and zoning. As major cloud and AI companies race to secure land and power for new facilities, they are also facing a harder question: whether the public will accept more data centers if communities are kept in the dark about what is being built next door.

Amazon changes course as data center opposition mounts

Amazon’s decision to drop NDAs from talks with local officials follows Microsoft’s similar move earlier this year and reflects a growing recognition inside Big Tech that secrecy can damage the politics of infrastructure expansion. In practice, NDAs have often prevented elected leaders from discussing proposed facilities before deals are finalized, leaving residents to learn about projects only after site selection, incentives or power commitments are already in motion.

That approach has become increasingly difficult to defend. Across the country, critics have argued that communities deserve more visibility into how data centers may affect utility bills, water supplies, environmental targets and land use. The result has been a wave of backlash that has already translated into local restrictions, delayed approvals and formal pauses on new builds.

On the latest Equity podcast, TechCrunch hosts Anthony Ha, Sean O’Kane and Rebecca Bellan used Amazon’s announcement as a window into the broader trust problem facing AI infrastructure. Their discussion centered on whether ending confidentiality agreements will materially change how deals are negotiated, or whether it is only a first step toward rebuilding confidence after months of public skepticism.

Why do data center deals trigger so much distrust?

Data center deals trigger distrust because communities often believe the economic promises are clearer than the costs. Local governments are frequently pitched job creation, tax revenue and long-term digital investment, while residents worry about immediate strains on infrastructure, increased power demand, industrial-scale land use and limited transparency around subsidies.

That tension has intensified with the AI era. Traditional cloud facilities were already large and resource-intensive, but the next generation of data centers is being designed to support far more compute-heavy workloads. That means bigger energy footprints, more cooling requirements and more pressure on local grids, all at a time when many municipalities are trying to balance growth with climate goals.

The secrecy surrounding negotiations has become a political liability because it makes it easier for opponents to frame deals as backroom arrangements. When residents only learn about a project after a location is effectively chosen, they tend to ask why public officials were not more open from the start.

Industry leaders are now acknowledging that hidden negotiations can undermine the social license needed to build at the scale AI requires.

What changed in the Amazon-Microsoft playbook?

What changed is the industry’s willingness to concede that confidentiality is no longer worth the reputational cost. Microsoft’s earlier decision to stop using NDAs in local government negotiations set an example that Amazon is now following, suggesting that the largest cloud companies see transparency as a competitive necessity rather than just a public-relations gesture.

Neither company is abandoning negotiation strategy entirely. They still need to secure land, permits, power access and local support for enormous new campuses. But by giving officials more freedom to talk, the companies are trying to reduce the suspicion that often surrounds major site-selection processes.

The move also reflects a practical reality: communities are increasingly organized, informed and willing to push back. Open records requests, local journalism and public meetings have made it harder for companies to keep large-scale infrastructure plans entirely under wraps. In that environment, secrecy can slow projects rather than protect them.

How big is the backlash against AI infrastructure?

The backlash is large enough to alter the policy landscape. Opposition to data centers and related AI infrastructure has helped produce hundreds of proposed or enacted moratoriums from major metros to suburban and regional jurisdictions, including places such as New York and San Francisco.

These moratoriums are not all identical, but they reflect a common pattern: local leaders asking for time to study the impact of data centers before more projects move forward. In some cases, the debate is about water and energy. In others, it is about noise, traffic, land consumption or the prospect of utility rates rising to accommodate enormous new power needs.

For companies, the challenge is that AI expansion depends on physical infrastructure far outside the spotlight of software launches and model releases. The public may be impressed by AI capabilities, but it still has to live with the warehouses, transformers, generators, substations and transmission lines required to deliver them.

Trust is becoming the product for personal AI startups

Amazon’s announcement matters beyond the data center business because it sits alongside another fast-growing trust debate in AI: whether consumers will allow startups to connect personal agents to their private data and financial lives. A wave of companies is betting that users will eventually grant AI systems access to inboxes, files, calendars and even credit cards if the tools deliver enough convenience.

That bet is ambitious. It assumes people will hand over sensitive permissions to software agents that can read emails, book travel, summarize documents and make purchases on their behalf. But the same trust problem that affects local governments and data center buildouts also shapes consumer AI. The question is not only whether the technology works, but whether people believe it will be used safely and responsibly.

One important difference is control. In infrastructure deals, communities are often reacting after the fact. In consumer AI, users can technically grant or revoke access themselves. But that does not eliminate the need for trust; it raises the stakes, because the systems may be operating inside a person’s most private digital spaces.

Why do AI agents need so much access?

AI agents need broad access because their value depends on acting across multiple services, not just answering questions. To schedule meetings, draft emails, process documents or complete transactions, they must see context from accounts, files and web services. Without those permissions, agents are far less useful and far less autonomous.

That creates a difficult trade-off. The more access a product has, the more helpful it can be — but also the more risky it becomes if the system is hacked, misconfigured or manipulated. Startups building in this space are therefore trying to sell not just capability, but confidence in safeguards, identity checks and permission boundaries.

It also means trust is becoming a core part of product design. Clear consent flows, audit logs, limited permissions and easy revocation are no longer nice-to-have features. They are part of the sales pitch.

What the data center fight reveals about AI’s next phase

What the data center fight reveals is that AI’s next phase is as much about institutions as it is about models. The companies that dominate the current wave of AI will need electricity, land, cooling and public approval on a scale that makes the politics of infrastructure inseparable from the economics of the industry.

That reality is forcing a broader rethink. For years, tech firms often treated opacity as a standard part of negotiating complex real estate and infrastructure arrangements. Now, the biggest players appear to be concluding that transparency may be the only way to secure the social permission needed for continued growth.

At the same time, the personal AI market is learning a related lesson. If companies want consumers to trust systems with calendars, mailboxes and payment credentials, they will need to prove that those systems are useful without becoming invasive. In both cases, the core issue is not simply scale. It is legitimacy.

What comes next for Amazon and other hyperscalers?

What comes next is likely a slower but more public negotiation process. Amazon, Microsoft and other hyperscalers will still pursue new facilities aggressively, but they may have to do so with more disclosure, more community engagement and more scrutiny from local officials.

That may not end opposition. Some communities will remain wary regardless of how open the process becomes, especially if projects threaten scarce resources or collide with local climate goals. But ending NDAs could reduce the perception that companies are hiding critical information until it is too late to challenge.

For the industry, that distinction is important. Trust will not be built by a single policy change, but by repeated behavior that shows companies are willing to explain what they want, what they are offering and what communities stand to bear in return.

Deals of the week, and why they matter

While the podcast discussion focused on infrastructure and trust, it also touched on a couple of deals of the week, underscoring how crowded and fast-moving the AI market remains. New funding rounds, product launches and enterprise bets continue to pour into the sector even as public scrutiny grows.

That combination creates a paradox. Investment remains strong because the market sees enormous upside. Public resistance remains strong because the costs are concrete and local. The companies that succeed will be the ones that can manage both realities at once.

• Big Tech wants to keep building AI infrastructure at scale.
• Local communities want more information before approving large projects.
• Consumer AI startups need users to trust them with sensitive data.
• Transparency is becoming a strategic requirement, not just a moral one.

Bottom line

Amazon’s decision to stop using NDAs in local data center negotiations is a notable sign that the AI infrastructure boom has entered a more politically fragile phase. The move may ease some suspicion, but the deeper challenge remains the same: convincing governments and consumers that the benefits of AI will outweigh the costs of expanding the physical and digital systems behind it.

In that sense, the story is bigger than one company, one podcast or one policy change. It is about whether the next era of AI can win trust at the same speed it is winning capital.