Amazon Web Services (AWS) and NVIDIA have announced a partnership, with plans to deploy 2 million additional NVIDIA GPUs across AWS’ global infrastructure in 2027 and 2028 as demand for AI computing grows.
The companies said the expansion will support workloads including agentic AI, scientific discovery, enterprise automation and robotics. The partnership will also cover CPUs, networking, open models, data processing and physical AI.
“NVIDIA and AWS have built one of the great growth engines of the AI era, and demand is running ahead of every forecast,” said Jensen Huang, Founder and CEO of NVIDIA. “For 16 years, we have scaled NVIDIA computing in the cloud together. Now, we are expanding our partnership across the full stack—GPUs, CPUs, networking, open models and software.”
The planned GPU deployment will include NVIDIA Blackwell Ultra, Rubin and Rubin Ultra GPUs. AWS will also expand its Blackwell capacity, including NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs for Amazon EC2 G7 instances.
AWS said G7 instances deliver 4.6 times the AI inference performance and 2.1 times the graphics performance of the previous-generation G6 instances. AWS is the first major cloud provider to offer compute instances using the RTX PRO 4500.
The companies are also working to bring NVIDIA Vera CPU-based infrastructure to AWS. Vera is designed to support AI workloads that require high-performance CPU computing alongside accelerated infrastructure.
AWS and NVIDIA are further expanding the use of NVIDIA NVLink Fusion, which connects processors and accelerators at high speed. Amazon’s Annapurna Labs is working with NVIDIA and memory suppliers to support NVIDIA’s custom high-bandwidth memory technology with Trainium chips.
The partnership also includes plans to build AI factories for the US government, with 1,00,000 NVIDIA GPUs on secure AWS infrastructure for federal and national-security workloads. The infrastructure is intended to support workloads at Impact Level 6 (IL6) and above.
“Customers want the freedom to choose the best tools for their AI workloads, and they want confidence that everything works seamlessly together,” said Matt Garman, CEO of AWS. “This expanded collaboration gives frontier labs, enterprises and governments even more ways to build and deploy AI on AWS.”
The companies will also continue integrating NVIDIA GPU-based infrastructure and AWS Trainium instances with the AWS Nitro System and Elastic Fabric Adapter (EFA). The two technologies are designed to provide security, reliability and network performance for AI workloads at scale.
NVIDIA’s Nemotron family of open models will remain available through Amazon Bedrock and Amazon SageMaker. Bedrock provides managed access to the models, while SageMaker allows customers to deploy and fine-tune them on their own infrastructure.
AWS and NVIDIA are also working on GPU-accelerated data processing and vector indexing. On Amazon EMR, NVIDIA’s cuDF library running on EC2 G7 instances can deliver up to 3.7 times faster processing and 30% better price performance than CPU-based configurations, according to AWS.
For Amazon OpenSearch Service, GPU-accelerated vector indexing can deliver up to nine times faster indexing at a quarter of the cost, the companies said.
The partnership also extends to robotics. Amazon Robotics is working with NVIDIA on next-generation robots using NVIDIA’s Jetson platform, Omniverse libraries and Isaac robotics development platform. The work covers simulation, synthetic data generation, robot training, route optimisation, functional safety and real-to-sim validation.
AWS and NVIDIA said the expansion builds on 16 years of collaboration as customers move AI workloads from pilots into production and scale applications across enterprises, governments and other sectors.
Your reaction