AWS and NVIDIA Expand Collaboration on AI Infrastructure
AWS and NVIDIA are expanding their collaboration to meet growing demand for AI infrastructure. The two companies will deploy an additional 2 million GPUs across AWS's global infrastructure, including AI factories, in 2027-2028.
This expansion is part of a larger effort to bring NVIDIA Vera CPUs, advanced networking, Nemotron open models, and physical AI technologies to AWS. AWS and NVIDIA are working together on several fronts, including building AI factories for the US government, integrating NVIDIA's platform with the AWS Nitro System and Elastic Fabric Adapter (EFA), and supporting NVIDIA Nemotron open models on Amazon Bedrock and Amazon SageMaker.
The companies are also collaborating on heterogeneous AI infrastructure using NVIDIA NVLink Fusion with custom high-bandwidth memory (NVHBM) technology. This will give Trainium access to faster, more power-efficient memory, enhancing performance and efficiency for AI workloads.