Amazon AWS drops NDAs amid backlash over AI data centers

The cloud buildout behind frontier AI has become politically contentious, and AWS is adjusting its playbook. Dropping NDAs for government data center deals is less about altruism and more about salvaging the social license to build the compute that underpins large models and AI agents. As communities from British Columbia to New York push back on water use, power draw and secrecy, hyperscalers have to trade some negotiating leverage for transparency.

In terms of AGI timelines, this move is unlikely to slow core research. The big labs already have multi year GPU contracts and internal data center plans. But if local opposition hardens into broad moratoriums, siting new high density facilities could become slower, more expensive and more geographically constrained. AWS’s posture looks like an attempt to head that off early by signaling that it is open to scrutiny about where and how AI infrastructure is built.

Strategically, AWS is also protecting its role as the default place where enterprises rent frontier scale compute. If data centers become politically toxic, governments may favor their own sovereign facilities or stricter procurement rules that advantage more transparent providers. By getting slightly ahead of the curve on transparency, AWS is trying to keep hyperscale cloud central to the AGI hardware stack instead of ceding ground to state run or consortium operated supercomputers.