The outfit said it would spend $50 billion to build what it calls high-performance computing infrastructure, specifically for federal agencies, and clearly hopes this will tempt more departments to adopt its AI services.

The plan adds 1.3 gigawatts of compute and widens access to bits such as Amazon SageMaker AI, model customisation tools, Amazon Bedrock, model deployment services and Anthropic’s Claude chatbot, according to AWS.

The company expects the data centre build to start in 2026.

AWS chief executive Matt Garman said: “Our investment in purpose-built government AI and cloud infrastructure will fundamentally transform how federal agencies leverage supercomputing.”

He added: “We’re giving agencies expanded access to advanced AI capabilities that will enable them to accelerate critical missions from cybersecurity to drug discovery. This investment removes the technology barriers that have held government back and further positions America to lead in the AI era,” he said.

AWS has been flogging its cloudy wares to the US government since 2011.

It rolled out AWS Top Secret East three years later as the first air-gapped commercial cloud tuned for classified work, then launched AWS Secret Region in 2017 with accreditation for all security classifications.

Big tech has been trying to cosy up to the US government’s AI budgets for more than a year.

OpenAI pushed out a government-only flavour of ChatGPT in January, then in August offered federal agencies its enterprise tier for $1 a year. That same month, Anthropic said it would hand out enterprise access to Claude for $1, while Google launched its Google for Government scheme for 47 cents for the first year.