AWS and NVIDIA to Deliver 2 Million Additional GPUs and Next-Generation Infrastructure for Agentic and Physical AI

The companies said they are deepening integration across the AI stack, with AWS set to receive NVIDIA Vera CPUs, advanced networking, Nemotron open models and physical AI technologies alongside 2 million additional GPUs. The announcement was published by GlobeNewswire on August 26, 2026, and framed the expansion as a response to accelerating customer demand.

The named connection is primarily NVIDIA’s hardware and software stack with Amazon Web Services’ cloud infrastructure. For NVIDIA, the potential mechanism is additional accelerator volume plus broader adoption of its CPUs, networking products and software; for AWS, the arrangement expands the infrastructure available for agentic and physical AI workloads.

The announcement does not establish when the GPUs will be delivered, how the purchases are priced or how much revenue is already reflected in NVIDIA’s outlook. The next useful disclosures are the companies’ timing and order details, followed by NVIDIA’s next earnings report and any AWS commentary on capacity utilization and customer demand.