Amazon Web Services is making a major push into government-focused artificial intelligence and supercomputing, announcing plans to build the first AI- and HPC-dedicated cloud infrastructure designed specifically for U.S. federal agencies, according to Amazon's PR.

The company said it will invest up to $50 billion beginning in 2026 to expand capacity across AWS Top Secret, AWS Secret, and AWS GovCloud (US), adding nearly 1.3 gigawatts of advanced compute power.

The expansion will give agencies access to a broader suite of AI tools—such as Amazon SageMaker, Amazon Bedrock, Amazon Nova, Anthropic Claude, leading open-weights models, AWS Trainium chips, and NVIDIA AI systems—while providing classified and unclassified environments for model training, simulation, and mission-critical workloads.

Amazon wrote that the investment will help agencies apply AI to large-scale modeling and simulation, enabling “autonomous experimental steering and real-time feedback loops” and shrinking processes that once took weeks down to hours.

The company highlights applications ranging from analyzing global security data to automating threat detection across satellite imagery and sensor streams. The expanded infrastructure is also positioned to accelerate research in fields like energy, healthcare, cybersecurity, and autonomous systems.

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

AWS frames the move as aligned with the Administration’s AI Action Plan and as part of a broader shift from traditional supercomputing to AI-accelerated workflows, where natural-language interfaces and AI agents help researchers explore complex problems.

The announcement also builds on more than a decade of government-focused cloud development, including milestones such as the launch of AWS GovCloud in 2011, the first air-gapped commercial cloud for classified workloads in 2014, and full accreditation across all U.S. classification levels by 2017.

With more than 11,000 government customers already on AWS, the company says the new investment is meant to provide secure, scalable infrastructure for a new era of computational discovery—one in which AI and HPC converge to support national security, scientific innovation, and the broader U.S. industrial base.