AWS commits $50B to build AI infrastructure for U.S. government

Amazon Web Services is making a major push into government-focused artificial intelligence, announcing a $50 billion investment to build high-performance computing infrastructure designed specifically for U.S. federal agencies . The company said the new buildout will greatly expand access to its advanced AI and cloud tools, including Amazon SageMaker, Amazon Bedrock, model customization services, and Anthropic’s Claude chatbot. AWS plans to add 1.3 gigawatts of compute as part of the project and expects construction on the data centers to begin in 2026.

AWS CEO Matt Garman said the investment will help federal agencies use supercomputing in new ways, removing long-standing technology barriers. He added that improving access to AI systems will support essential government work, from cybersecurity to drug development.

The company has a long history of supporting U.S. government operations. AWS first began building specialized cloud infrastructure for federal clients in 2011, later launching AWS Top Secret-East in 2014 as the first commercial cloud environment designed for classified workloads. In 2017, it rolled out the AWS Secret Region, which offers accredited access across all security levels.

The move comes as major tech companies compete to provide AI tools to government agencies. OpenAI introduced a government-only version of ChatGPT in January and later agreed to offer enterprise ChatGPT access to federal organizations for $1 per year . Anthropic followed in August with a similar $1 deal for its Claude enterprise tier. Google launched its “Google for Government” program soon after, offering its first-year service for just 47 cents.

AWS’s latest investment signals how rapidly the AI landscape is evolving as federal agencies seek powerful and secure systems capable of handling sensitive workloads. The new infrastructure aims to give government workers faster and broader access to the AI technologies shaping modern public-sector operations.