xAI Memphis Supercluster

The xAI Memphis Supercluster, located in Memphis, Tennessee, is a groundbreaking AI infrastructure project spearheaded by Elon Musk's xAI. This facility is designed to be one of the most powerful AI training clusters in the world, with plans to scale up to 1 million GPUs by 2026. Currently, the supercluster operates with 100,000 liquid-cooled Nvidia H100 GPUs connected via a single RDMA fabric, offering unparalleled computational power. The Memphis Supercluster is a key component of xAI's mission to advance AI capabilities and support the development of cutting-edge technologies like the Grok chatbot.

At the heart of the Memphis Supercluster is the Colossus Supercomputer, a marvel of engineering and efficiency. Colossus is equipped with over 200,000 GPUs and boasts a total memory bandwidth of 194 petabytes per second. Its network bandwidth per server reaches an impressive 3.6 terabytes per second, and it offers more than 1 exabyte of storage capacity. Built in just 122 days, Colossus exemplifies xAI's commitment to rapid innovation and scalability. This supercomputer is not only a technological powerhouse but also a symbol of xAI's ambition to lead the AI revolution.