(Reuters) – U.S. businessman Elon Musk recently told investors that his artificial intelligence startup xAI plans to build a supercomputer to run the next version of its artificial intelligence chatbot Grok, The Information reported on Saturday, citing an investor presentation.
Musk said he wants to launch the proposed supercomputer by fall 2025, according to the report, adding that xAI could partner with Oracle (NYSE:) to develop the massive computer.
xAI could not be reached for comment. Oracle did not respond to Reuters’ request for comment.
Once completed, the associated groups of chips—Nvidia’s flagship H100 graphics processing units (GPUs)—will be at least four times larger than the largest GPU clusters in existence today, The Information reported, citing Musk from a presentation made to investors in May.
Nvidia’s (NASDAQ:) H100 family of powerful GPUs dominates the market for data center chips for artificial intelligence, but they are difficult to acquire due to high demand.
Musk founded xAI last year as a competitor to Alphabet’s (NASDAQ:) Microsoft-backed OpenAI and Google. Musk also co-founded OpenAI.
Earlier this year, Musk said that training the Grok 2 model required about 20,000 Nvidia H100 GPUs, adding that Grok 3 and later would require 100,000 Nvidia H100 chips.