Elon Musk, a prominent U.S. businessman, shared with investors that his artificial intelligence startup xAI is working on developing a supercomputer to power the next iteration of its AI chatbot Grok. According to a report by The Information, Musk aims to have this supercomputer up and running by the fall of 2025 and is exploring a potential partnership with Oracle to bring the project to fruition.
The proposed supercomputer would consist of interconnected Nvidia H100 GPUs, which are renowned for their performance in AI applications. Musk mentioned in a presentation to investors in May that these GPU clusters would be at least four times larger than the largest existing clusters. However, obtaining Nvidia’s H100 GPUs can be challenging due to their high demand in the data center chip market.
xAI, Musk’s venture, was established last year to compete with industry giants like Microsoft-backed OpenAI and Google, which is under Alphabet. Musk was also a co-founder of OpenAI. Earlier this year, Musk revealed that training the Grok 2 model required approximately 20,000 Nvidia H100 GPUs, with future models like Grok 3 expected to need up to 100,000 of these high-performance chips. Information regarding xAI’s response to inquiries or Oracle’s comment on the matter was not immediately available.