Nvidia and Amazon Web Services (AWS) are deepening their longstanding strategic partnership with a series of significant announcements at the AWS re:Invent conference. Among these announcements is the introduction of a groundbreaking DGX Cloud offering by Nvidia, marking the inaugural integration of the Grace Hopper GH200 superchip into AWS infrastructure.
This collaboration takes a giant leap forward with the unveiling of Project Ceiba, an ambitious initiative that aspires to establish what could potentially become the world's largest public cloud supercomputing platform. Project Ceiba, powered by Nvidia technology on AWS, is poised to deliver an astounding 64 exaflops of AI power, showcasing the unparalleled capabilities of this collaborative effort.
Adding to the advancements, AWS is set to enrich its EC2 service with the inclusion of four new types of GPU-powered cloud instances. This expansion reflects the commitment to providing users with enhanced computing capabilities and performance. The collaboration extends its focus beyond infrastructure improvements with Nvidia's introduction of NeMo Retriever technology at AWS re:Invent. NeMo Retriever adopts a Retrieval Augmented Generation (RAG) approach to bridge enterprise data with generative AI, facilitating the creation of more robust large language models (LLMs).
Nvidia and AWS have shared a collaborative journey spanning over 13 years, with Nvidia GPUs making their debut in AWS cloud computing instances in 2010. Ian Buck, VP of Hyperscale and HPC at Nvidia, highlighted the ongoing efforts to enhance innovation and operational efficiency at AWS and benefit mutual customers, including prominent entities like Anthropic, Cohere, and Stability AI. The continued synergy between Nvidia and AWS at the forefront of cloud computing and AI underscores their dedication to pushing the boundaries of technology and fostering advancements for a wide range of industries.