NVIDIA and AWS Supercharge Cloud AI with New Blackwell GPUs, 10x Faster Vector Search, and Elite Training Certification
Summary
NVIDIA and AWS are rolling out a powerful trio of AI advancements: new EC2 G7 instances with Blackwell GPUs delivering 4.6x faster inference, a 10x faster vector search engine now default in Amazon OpenSearch, and AWS earning elite NVIDIA Exemplar Cloud status for large-scale AI training workloads.
Key Points
- NVIDIA and AWS are launching Amazon EC2 G7 instances powered by NVIDIA RTX PRO 4500 Blackwell Server Edition GPUs, delivering up to 4.6x AI inference performance and 2.1x graphics performance compared to previous G6 instances.
- NVIDIA cuVS is now the default GPU-accelerated vector indexing engine in Amazon OpenSearch Serverless, enabling vector indexing up to 10x faster at a quarter of the cost compared to CPU-only solutions, making billion-scale vector databases buildable in under an hour.
- AWS has achieved NVIDIA Exemplar Cloud status for GB300 training workloads, signaling that AWS meets NVIDIA's rigorous performance benchmarks through deep co-engineering, giving enterprises greater confidence in cloud infrastructure for large-scale AI training.