Amazon and Other Tech Giants Are No Longer Hiding Data Center Deals—But Will It Earn Trust?

Meta description: Amazon ends NDAs for data center talks with local governments, following Microsoft, as AI infrastructure secrecy drives community backlash and moratoriums.

Tags: Amazon, data centers, AI infrastructure, NDAs, personal AI agents

Featured image: Abstract technology style rendering of interconnected server nodes, network lines, and civic buildings, no people, logos, or text.

Amazon says it will stop using non-disclosure agreements, or NDAs, when negotiating data center deals with local governments. The change follows a similar move from Microsoft earlier this year. It arrives as secrecy has fueled community backlash against AI infrastructure. The same reporting frame extends to a second frontier: a wave of startups is betting that consumers will hand AI agents access to their inboxes, files, and credit cards, if they can get websites to let them in.

The NDA shift follows a backlash over hidden infrastructure

NDAs are contracts that restrict what parties may disclose. In local government negotiations, they can keep terms of a data center deal private while approvals and site impacts are debated. Amazon’s stated move away from that practice responds directly to an environment in which secrecy has fueled community backlash against AI infrastructure. Opposition has led to hundreds of proposed and enacted moratoriums from New York to San Francisco, showing that local scrutiny has become a central risk for AI buildouts.

Microsoft’s earlier decision makes Amazon’s move part of a pattern rather than an isolated exception. The question is not only whether companies disclose more, but whether disclosure changes bargaining power. If negotiations are public, local governments may face pressure from residents, while large AI firms may face pressure to explain terms that affect communities. On TechCrunch’s Equity podcast, Anthony Ha, Sean O’Kane, and Rebecca Bellan dig into whether ditching NDAs will change how data center deals get done, why trust may be the real product for personal AI startups, plus a couple of deals of the week.

Personal AI startups need permission, not just models

The startup thread concerns consumer AI agents, software that can carry out tasks on a user’s behalf. A wave of startups is betting that consumers will hand AI agents access to their inboxes, files, and credit cards, if they can get websites to let them in. That condition is the key: the agent’s usefulness depends less on its ability to generate text than on its ability to take action in existing digital systems. In this framing, trust may be the real product for personal AI startups. Users are not only asking whether an agent can summarize email or complete a checkout; they are asking whether it can be allowed to see private data and spend money.

The podcast ties these topics together as a broader trust problem. Data center deals require public trust in AI infrastructure. Personal AI agents require private trust in AI software. Both depend on mechanisms that make access, cost, and control legible.

What comes next is proof from local reviews and agent permissions

The current stage is a shift in posture, not a settled result. Amazon says it will stop using NDAs in local government data center negotiations, but the impact will show up in future deals: what terms are disclosed, which communities approve, and whether moratoriums slow or reshape projects. For personal AI startups, the test is whether websites allow agents to act and whether consumers accept that access. The next signal will be whether more large companies adopt public negotiation terms and whether agent permissions become a standard interface between users, businesses, and AI systems.