Groundbreaking Collaboration at AWS re:Invent
Amazon Web Services (AWS) and NVIDIA have taken a giant leap in their strategic collaboration, announced at AWS re:Invent. This partnership is set to revolutionize the field of generative AI by combining AWS’s advanced cloud infrastructure with NVIDIA’s cutting-edge GPU technology.
Uniting NVIDIA’s Supercomputing with AWS’s Cloud Prowess
The collaboration is a synergy of NVIDIA’s latest multi-node systems, which include next-generation GPUs, CPUs, and AI software, with AWS technologies like the Nitro System for advanced virtualization, Elastic Fabric Adapter (EFA) interconnect, and UltraCluster scalability. This union is poised to offer unparalleled resources for generative AI innovations.
Key Highlights of the Expanded Collaboration
NVIDIA GH200 Grace Hopper Superchips on AWS
- AWS becomes the first cloud provider to offer NVIDIA GH200 Grace Hopper Superchips.
- The NVIDIA GH200 NVL32 multi-node platform allows scaling to thousands of GH200 Superchips, delivering supercomputer-class performance.
Hosting NVIDIA DGX Cloud on AWS
- The collaboration will see NVIDIA DGX Cloud, an AI-training-as-a-service, hosted on AWS. It features GH200 NVL32 for rapid training of generative AI and large language models.
Project Ceiba Supercomputer
- The two companies are joining forces on Project Ceiba to create the world’s fastest GPU-powered AI supercomputer, boasting 16,384 NVIDIA GH200 Superchips and a processing capability of 65 exaflops.
Introduction of New Amazon EC2 Instances
- AWS introduces three new Amazon EC2 instances, including P5e instances powered by NVIDIA H200 Tensor Core GPUs, catering to large-scale generative AI and HPC workloads.
Software Innovations
- NVIDIA introduces new software on AWS, such as the NeMo Retriever microservice for chatbots and summarisation tools, and BioNeMo for accelerating drug discovery in the pharmaceutical industry.
Impact on Industries and Internal Use
This partnership marks a significant commitment to advancing generative AI. Internally, Amazon’s robotics and fulfilment teams are utilizing NVIDIA’s Omniverse platform for optimizing warehouses in virtual environments. The integration of NVIDIA and AWS technologies is set to expedite the development, training, and inference of large language models and generative AI applications across various sectors.