AWS and NVIDIA Expand Partnership to Boost AI Production Capabilities

“The real opportunity with AI lies in deploying it at scale to drive meaningful business outcomes,” said an AWS spokesperson during NVIDIA GTC 2026. To meet this demand, AWS and NVIDIA announced a broadened collaboration featuring new technology integrations designed to enhance AI compute capacity and enable production-ready AI solutions.

Starting in 2026, AWS will deploy over 1 million NVIDIA GPUs, including the Blackwell and Rubin architectures, across its global cloud regions. This expansion positions AWS as the cloud provider with the widest range of NVIDIA GPU-based instances, supporting diverse AI and machine learning workloads. The partnership also advances interconnect technologies, such as the NVIDIA Inference Xfer Library (NIXL) on AWS Elastic Fabric Adapter (EFA), to accelerate disaggregated Large Language Model (LLM) inference by reducing communication latency and improving GPU utilization.

New Amazon EC2 instances powered by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs will be available soon, marking AWS as the first major cloud provider to support this GPU. These instances are optimized for workloads including data analytics, conversational AI, content generation, recommender systems, and video processing. Built on the AWS Nitro System, these instances offer enhanced resource efficiency, security, and stability by isolating sensitive workloads and enabling seamless firmware updates without downtime.

AWS and NVIDIA also reported a threefold increase in Apache Spark performance using Amazon EMR on Amazon Elastic Kubernetes Service (EKS) with Amazon EC2 G7e instances powered by NVIDIA RTX PRO 6000 Blackwell Server Edition GPUs. This improvement accelerates data processing pipelines critical for AI/ML feature engineering and real-time analytics, reducing time-to-insight for data scientists and engineers.

Additionally, NVIDIA Nemotron models will soon support fine-tuning on Amazon Bedrock via Reinforcement Fine-Tuning (RFT), enabling developers to customize models directly within the cloud environment. This expanded collaboration reflects over 15 years of joint innovation aimed at scaling AI infrastructure and advancing autonomous AI systems capable of complex reasoning and planning.

Read more: aws.amazon.com