Amazon and Nvidia expand AI partnership with 2 million GPUs

Amazon Web Services (AWS) and Nvidia have announced a major expansion of their strategic partnership to meet soaring global demand for artificial intelligence infrastructure. Under the expanded agreement, AWS plans to deploy an additional 2 million Nvidia graphics processing units across its worldwide cloud infrastructure between 2027 and 2028. The deployment will feature next-generation architecture including Nvidia's Blackwell Ultra, Rubin, and Rubin Ultra GPUs, alongside RTX PRO 4500 Blackwell Server Edition GPUs for Amazon EC2 G7 instances. Beyond commercial cloud scaling, the partnership encompasses specialized infrastructure for the public sector and advanced robotics. AWS and Nvidia intend to build dedicated AI factories for the U.S. government, providing 100,000 GPUs on AWS secure infrastructure designed for Impact Level 6 and higher federal workloads. Furthermore, the collaboration integrates Nvidia's Vera CPU-based infrastructure, the Nemotron model family into Amazon Bedrock and SageMaker, and links Amazon Robotics with Nvidia's Jetson, Omniverse, and Isaac platforms for physical AI development.