NVIDIA and AWS Collaborate to Bring AI to Production at Scale

Building AI systems at scale presents significant challenges, including the need for low-latency inference, rapid vector search capabilities, strong GPU price-performance, and scalable infrastructure that avoids operational complexity. NVIDIA's latest collaboration with Amazon Web Services (AWS) directly addresses these constraints, providing enterprises with more practical pathways to deploy AI at production scale across Amazon OpenSearch and Amazon EC2. A cornerstone of this partnership is the integration of GPU-accelerated vector indexing, powered by NVIDIA cuVS, as the default compute option for all vector collections within Amazon OpenSearch Serverless. This strategic move transforms GPU-powered vector search from a specialized optimization project into a standard, readily available AWS capability. The direct impact for customers is substantial: vector indexing can now be achieved up to 10 times faster and at a quarter of the cost when compared to traditional CPU-only builds. This efficiency gain makes it feasible to construct billion-scale vector databases in less than an hour, a significant advancement for applications like retrieval-augmented generation (RAG), semantic search, and recommendation systems. Furthermore, the collaboration extends to the compute layer with the introduction of Amazon EC2 G7 instances. These instances are powered by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, expanding the capabilities for AI, graphics, video, and data analytics workloads. Data teams can leverage the improved GPU memory, local storage, and networking enhancements for analytics pipelines and vector database operations. These G7 instances are accessible through various AWS services, including Deep Learning AMIs, Amazon Deep Learning Containers, Amazon EMR, Amazon EKS, Amazon ECS, and graphics AMIs, with upcoming support for Amazon SageMaker AI. The partnership also highlights AWS's achievement of NVIDIA Exemplar Cloud status on NVIDIA GB300 for training workloads, signifying that AWS meets NVIDIA's stringent performance thresholds for AI workloads. This deep co-engineering effort between AWS and NVIDIA aims to provide a consistent, high-performance cloud infrastructure for large-scale AI training, simplifying cloud provider evaluation and accelerating the transition of AI projects from planning to efficient production. This comprehensive approach reinforces every layer of the AI infrastructure stack on AWS, enabling businesses to scale AI workloads efficiently and accelerate their time-to-value.