Reuters and other news outlets have reported that Dell Technologies and Supermicro will provide racks of servers for Elon Musks’s mammoth AI supercomputer for his xAI startup.
Dell CEO Michael Dell disclosed the development on Musk’s X (formerly Twitter) social media platform yesterday. Musk confirmed the news and said Dell is building about 50 percent of the xAI racks (photo at right posted by Michael Dell on X).
Later, Supermicro confirmed it is part of the group building the xAI supercomputer.
All of this is good news for Nvidia, which will supply processors used in xAI racks.
Reuters also reported that Dell said “in a separate post on X that the company was building an ‘AI factory’ with artificial-intelligence heavyweight Nvidia that would power the next version of xAI’s chatbot Grok.”
An AI factory, a term rapidly gaining currency in HPC-AI circles, is a centralized computing capability that uses AI to develop AI solutions. In a recent article on LinkedIn by Dilek Celik, AI factories are “frameworks that centralize AI development and deployment across an organization, improving efficiency, scalability, and business process integration. These advanced setups transform traditional data centers into dynamic and efficient hubs that use AI to power applications, particularly in AI training and inference, thereby enhancing factory workflows.”
Musk has said that training the Grok 2 model took about 20,000 Nvidia H100 GPUs, while the Grok 3 model will require 100,000 H100s. He said xAI is targeting fall 2025 for the system to be operational, according to The Information.
According to a story in Data Centre Dynamics, “the plans for what Musk has dubbed both ‘the world’s largest and most powerful supercomputer’ and a ‘gigafactory of compute’ were first unveiled earlier this month. Subject to local approval, the giant data center will be located in Memphis, Tennessee, and represent the “’argest multi-billion dollar investment in the city of Memphis’s history,’ according to … the Greater Memphis Chamber.”
