Elon Musk announces xAI Colossus, says world’s most powerful AI training system is now online
Elon Musk has announced that the xAI Colossus 100k H100 training cluster went online last weekend, and will be doubled in size to 200k (50k H200s) in a few months.
Listen to the story
The growing aspiration to build the most powerful artificial intelligence systems is evident in the tech world. It seems that Elon Musk has made a big leap in this race. On Monday, Musk announced that the xAI Colossus is now online. The x boss says it is the world’s most powerful AI training system. xAI is Elon Musk’s AI research and development company. The xAI Colossus training cluster is really huge – it uses 100,000 H100 Nvidia GPUs, and Musk claims that “it will double in size to 200k (50k H200s) in a few months”.
“This weekend, the @xAI team brought our Colossus 100k H100 training cluster online. From start to finish, it was completed in 122 days. Colossus is the world’s most powerful AI training system. Plus, it will double in size to 200k (50k H200s) in a few months. Excellent work by the team, Nvidia, and our many partners/suppliers,” Elon Musk wrote in a post on X.
Elon Musk also revealed that his team successfully brought the Colossus AI training cluster online within 122 days. An unprecedented development in artificial intelligence, Colossus was developed in collaboration with leading semiconductor chip maker Nvidia. Nvidia’s H200 chips used in Colossus are in high demand.
For reference, Nvidia’s H200 has impressive specifications, including 141 gigabytes of HBM3e memory and 4.8 terabytes per second of bandwidth.
Nvidia has also responded to the Colossus launch. The company congratulated Musk and the xAI team, stressing that the system will not only be the most powerful AI training setup in the world, but will also offer “exceptional gains” in energy efficiency.
The Nvidia data center’s X handle wrote on the platform, “It’s exciting to see Colossus, the world’s largest GPU supercomputer, come online in record time. Colossus is powered by @nvidia’s #AcceleratedComputing platform, delivering spectacular performance with exceptional gains in #energyefficiency. Congratulations to the entire team!”
The scale of this achievement was highlighted by an Xe user, who noted that Colossus surpasses the capabilities of other leading AI models, including OpenAI’s most powerful model, which uses 80,000 GPUs. According to the data shared, Google AI uses 90,000, Meta AI 70,000, and Microsoft AI 60,000 GPUs.
Musk’s xAI, founded last year, is a strategic move to establish itself among the top players in the AI industry, challenging giants such as Microsoft-backed OpenAI and Google, which is a subsidiary of Alphabet. Musk’s involvement in this competitive field is particularly notable as he was one of the co-founders of OpenAI. His decision to launch xAI reflects his ambition to further push the boundaries of AI development, leveraging cutting-edge technology such as Nvidia’s powerful GPUs to achieve this goal.
