Elon Musk has told investors that he wants to deploy a 100,000 H100 GPU data center for his xAI startup.
Musk said that he wanted to get the supercomputer running by the fall of 2025, The Information reports. The Tesla and SpaceX CEO has a history of overpromising on tight deadlines that are then widely missed.
The supercomputer, referred to as a “gigafactory of compute," would be used to train the latest version of the Grok chatbot.
It is not yet known if Musk will deploy the hardware through a wholesale data center, or if it will be part of the company's existing partnership with Oracle Cloud. XAI is believed to currently rent around 16,000 Nvidia H100 GPUs from the cloud service.
The news comes as the year-old startup closed a $6 billion fundraising round that values the business at $24bn. Backers include venture capital investors Andreessen Horowitz, Sequoia Capital, and Fidelity Management & Research Company, along with Kingdom Holding, a Saudi investor run by Prince Alwaleed bin Talal, a member of the Saudi royal family.
Many of the investors have also backed Tesla, which has deployed some 35,000 Nvidia H100s for training self-driving vehicles, as well as developing supercomputers using its custom Dojo chips.
