Elon Musk announced the commencement of operations for the Memphis Supercluster of xAI, a supercomputer equipped with 100,000 NVIDIA H100 GPUs liquid-cooled in tandem. This Supercluster stands as the most potent computing behemoth globally for AI training purposes.
This supercomputer is the same machine that was previously reported to have Elon loaning Tesla’s H100 GPUs to xAI for temporary usage. The companies responsible for producing the servers are Dell and Super Micro.
Kudos to the teams at xAI, X, Nvidia, and supporting companies for initiating training on the Memphis Supercluster at around 4:20 am local time. With a staggering 100k liquid-cooled H100 units connected via a single RDMA fabric, it undeniably ranks as the paramount AI training cluster worldwide!
TLDR: Elon Musk reveals the cutting-edge Memphis Supercluster for xAI powered by 100k liquid-cooled NVIDIA H100 GPUs, making it the ultimate AI training hub globally.
Leave a Comment