The AI infrastructure boom is becoming a transparency battle as communities question who pays its real costs.
Seattle
Amazon Web Services says it will stop requiring confidentiality agreements from local governments negotiating future data center projects in the United States. The policy change follows growing criticism that some municipalities were prevented from publicly discussing projects whose electricity, water and infrastructure demands can significantly affect surrounding communities.
AWS chief executive Matt Garman said the company will no longer use nondisclosure agreements with government agencies involved in new projects. The decision applies prospectively and will not automatically reopen confidentiality arrangements already in force. Questions also remain about agreements signed through contractors or intermediaries working on Amazon projects.
The controversy intensified after cases in Indiana and Texas. In South Bend, Indiana, Amazon has invested more than $13 billion in a major data center complex, while elected officials said some local government employees were unable to answer public questions because confidentiality agreements restricted what they could disclose. In Wharton County, Texas, a local official signed a similar agreement with an outside consulting firm involved in an Amazon linked project.
The change arrives as political scrutiny grows in Washington. Lawmakers have requested information from major technology companies about the electricity and water consumed by data centers and about whether confidentiality provisions were used to deny public information requests. The United States hosts roughly 45 percent of the world’s data centers, with more than 1,000 additional projects under development across more than 40 states.
Amazon is simultaneously trying to address community resistance through a new five year, $1 billion initiative called Built Together. The program will fund technical training, tuition free college courses and energy efficiency projects in schools and homes located near data center developments.
The pressure reflects the enormous scale of the AI infrastructure race. Amazon expects to spend close to $220 billion this year, largely on data centers, chips and related computing capacity. Yet opposition is rising. More than 100 local moratorium proposals have reportedly emerged across the United States, while dozens of projects worth approximately $130 billion were blocked or delayed during the first quarter of the year.
The strategic issue is no longer whether AI requires massive physical infrastructure. It does. The emerging question is whether communities hosting that infrastructure will receive enough information to evaluate its costs before decisions become irreversible.
Transparency is becoming part of the price of computing power.
Behind every data point, the intention.