AWS and Nvidia are stepping up their partnership: the two companies have agreed to roll out an additional two million GPUs in AWS's cloud infrastructure during 2027 and 2028. ForkLog reports this move is driven by the relentless surge in demand for compute power to fuel both agent and physical AI.

Scale and Timeline

The new plan calls for more than 2 million GPUs to be delivered to AWS's global infrastructure over 2027 and 2028. This stacks on top of earlier commitments: at GTC 2026, AWS announced plans to install over 1 million Nvidia GPUs starting in 2026. All told, we’re talking about a multi-million fleet of accelerators coming online.

Focus: Agent and Physical AI

Both companies are scaling up as interest in agent and physical AI keeps climbing—these fields need serious horsepower for training and running models. The extra GPUs are aimed at making high-performance compute even more accessible in the AWS cloud for these demanding workloads.

This expanded partnership highlights AWS and Nvidia’s strategic bet on scaling AI infrastructure and getting ready for even bigger global demand for compute.