Amazon and others are done keeping data center deals secret. Is it enough to build trust?

Amazon just announced it will stop using non-disclosure agreements when negotiating data center deals with local governments, according to reporting across tech outlets. Microsoft made the same pledge earlier this year. The shift comes as communities from New York to San Francisco have proposed or enacted hundreds of moratoriums on new data center projects, often citing secrecy around the deals as a key concern.

The timing is not coincidental. AI training and inference require massive compute infrastructure, and that means data centers are popping up faster than ever. But these facilities draw enormous amounts of power and water, and when negotiations happen behind closed doors, locals feel blindsided when they learn a new energy-hungry complex is coming to their town.

Dropping NDAs is a step toward transparency, but it may not be enough to resolve the underlying tension. Communities are not just upset about secrecy, they are worried about strain on electrical grids, water usage during droughts, and whether they will see real economic benefits beyond construction jobs. Opening up the negotiation process gives residents a chance to voice concerns earlier, but it also exposes tech companies to organized opposition before deals are signed.

This is part of a broader reckoning in AI infrastructure. The industry has spent years moving fast and breaking things, and now it is running into the physical limits of power, cooling, and public patience. Companies that can show they are genuinely listening and adapting to local needs will have an easier time expanding. Those that treat transparency as a PR move without changing their approach will keep hitting resistance.

Meanwhile, the underlying demand is not slowing down. Startups are building AI agents that need always-on access to cloud compute, and enterprises are spinning up private AI deployments. Every new model, every agentic workflow, every voice assistant, it all runs somewhere, and that somewhere needs power and cooling.

What this means for you: if you are building AI products or workflows, factor in infrastructure constraints as a real business risk. Cloud costs are rising, and availability is not guaranteed in every region. When planning AI deployments, ask your provider about their data center pipeline and regional capacity. For teams experimenting with agents, try this prompt with your AI assistant: "Map out the compute and API call volume for [your planned agent workflow], estimate monthly cloud costs, and suggest three ways to optimize for lower infrastructure usage without sacrificing core functionality." Planning for efficiency now beats scrambling when your bill or your quota spikes later.