Elon Musk’s xAI is making waves in Memphis with its Colossus supercomputer cluster, which is currently housed in a former Electrolux facility. The company activated Colossus in July, initially powered by 100,000 Nvidia H100 GPUs. However, in late October, Musk announced plans to upgrade Colossus to 200,000 Nvidia H100/H200 GPUs.
The Greater Memphis Chamber has revealed that xAI is set to expand Colossus’ GPU count to at least 1 million units, as reported by Insider. This move aims to strengthen xAI’s position against competitors like OpenAI and Google by enhancing the capabilities of its large language model, Grok.
Brent Mayo, an xAI engineer, highlighted the project’s significance in both local and technological advancement. Mayo stated, “In Memphis, we’re pioneering development in the heartland of America. We’re not just leading from the front; we’re accelerating progress at an unprecedented pace while ensuring the stability of the grid utilizing megapack technology.”
Elon Musk added a touch of humor to the update, joking that at least 1 billion GPUs will eventually be used in the facility.
Colossus is particularly impressive due to its rapid construction timeline. Musk mentioned that it took just 122 days to build Colossus, with the supercomputer starting training within 19 days of hardware installation. Nvidia CEO Jensen Huang praised xAI’s speed and effort, describing the project as a “superhuman” task and acknowledging Musk’s engineering prowess. Huang noted that a project like Colossus typically takes around three years to plan and one year to set up.
xAI’s market value has soared to $50 billion, doubling its valuation since earlier this year. The company has also raised a significant $11 billion in funding in 2024.
For more news tips, readers can contact Simon at simon@teslarati.com. The xAI supercomputer in Memphis is set to undergo a major upgrade to 1 million GPUs, solidifying its position in the AI industry.