At AWS re:Invent, NVIDIA and Amazon Web Services announced new technology integrations to enhance AI capabilities, including NVIDIA NVLink Fusion platform for custom AI infrastructure. AWS will support Trainium4 chips, Graviton CPUs, and Nitro System with NVLink Fusion, simplifying deployment and systems management across platforms.
AWS is designing Trainium4 to integrate with NVLink and NVIDIA MGX, the first of a multigenerational collaboration between NVIDIA and AWS for NVLink Fusion. The integration will accelerate time to market for cloud-scale AI capabilities and increase performance.
NVIDIA and AWS are expanding accelerated computing offerings with NVIDIA Blackwell architecture, including GPUs for training and inference. They’re launching AWS AI Factories to provide customers with dedicated AI infrastructure in their data centers, ensuring control of data and compliance with regulations.
NVIDIA’s Nemotron open models are now integrated with Amazon Bedrock, allowing customers to build generative AI applications at scale. The integration makes high-performance NVIDIA models accessible via Amazon Bedrock’s serverless platform, with industry leaders like CrowdStrike and BridgeWise already using the service.
AWS is offering serverless GPU acceleration with NVIDIA cuVS on Amazon OpenSearch Service for vector index building, achieving faster indexing speeds at a lower cost. This shift to GPU-accelerated unstructured data processing reduces search latency and accelerates productivity for AI techniques like retrieval-augmented generation.
The collaboration between NVIDIA and AWS extends to software optimizations, including Strands Agents, NVIDIA NeMo Agent Toolkit, and Amazon Bedrock AgentCore for agent development and infrastructure. This support enables organizations to deploy AI applications faster and more efficiently by combining various tools and frameworks.
NVIDIA Cosmos world foundation models are now available on AWS for training and validating robot models in simulation before real-world deployment. Leading robotics companies are leveraging the NVIDIA Isaac platform with AWS for tasks ranging from data processing to training and simulation, enhancing robotics development.
NVIDIA received the AWS Global GenAI Infrastructure and Data Partner of the Year award for their continued collaboration with AWS on AI infrastructure and data solutions. The partnership between NVIDIA and AWS aims to advance AI innovation globally and provide customers with cutting-edge computing capabilities.
Read more at NVIDIA: NVIDIA and AWS Expand Full-Stack Partnership