Musk’s xAI ‘To Build Nvidia Supercomputer’

Elon Musk-founded start-up xAI is planning to build a supercomputer based on large numbers of Nvidia H100 graphics accelerator chips to power the next version of its AI chatbot Grok, Musk reportedly said in a presentation to investors.

Musk said he wants the supercomputer to be operational by the autumn of next year and that the company could work with Oracle to build the system, The Information reported, citing a presentation to investors earlier this month.

The system would use at least four times the number of clustered Nvidia H100 chips used in the biggest such systems today, Musk reportedly said.

He compared the system to Tesla’s “gigafactory” plants, calling it “the gigafactory of compute”, The Information said.

Image credit: xAI

Supercomputer

Musk founded xAI to compete with generative AI systems from the likes of OpenAI, which he also co-founded, and Google, after such systems became popular with users.

Earlier this year he said the Grok 2 large language model (LLM) required 20,000 H100 chips to train, and that Grok 3 would require 100,000 of the in-demand H100 chips.

xAI said over the weekend it had closed a $6 billion (£4.7bn) funding round giving the start-up a post-money valuation of $24bn.

On Monday Musk posted a recruitment call on his social media plaform X, formerly Twitter, writing, “Join xAI if you believe in our mission of understanding the universe, which requires maximally rigorous pursuit of the truth, without regard to popularity or political correctness.”

‘No pressure’

Meta Platforms chief AI scientist Yann LeCun poked fun at Musk’s post, writing on X: “Join xAI if you can stand a boss who: – Claims that what you are working on will be solved next year (no pressure). – Claims that what you are working on will kill everyone and must be stopped or paused (yay, vacation for 6 months!). – Claims to want a ‘maximally rigorous pursuit of the truth’ but spews crazy-ass conspiracy theories on his own social platform.”

LeCun, who has a history of public feuds with Musk, was referring to Musk’s claim in April that artificial general intelligence will arrive by next year and his estimate that there is a 10 to 20 percent chance AI could destroy humanity.

He also referenced a letter signed by Musk and others last year warning that advanced AI development should be paused for six months and Musk sharing posts that appeared to endorse conspiracy theories.

Matthew Broersma

Matt Broersma is a long standing tech freelance, who has worked for Ziff-Davis, ZDnet and other leading publications

Recent Posts

OpenAI Tests Search Engine Prototype Called ‘SearchGPT’

Google's dominance of online search is being challenged, after OpenAI unveiled a search prototype tool…

15 hours ago

Elon Musk To Discuss $5 Billion xAI Investment With Tesla Board

Conflict of interest? Elon Musk to talk with Tesla board about making $5 billion Tesla…

19 hours ago

Amazon Developing Cheaper AI Chips – Report

Engineers at Amazon's chip lab in Austin, Texas, are racing ahead to develop cheaper AI…

1 day ago

Apple Smartphone Sales In China Drop 6.7 Percent, Canalys Finds

China woes. Apple's China smartphone shipments decline during the second quarter, dropping it down into…

2 days ago

Meta Ordered To Clean Up AI-Generated Porn By Oversight Board

Oversight Board orders Meta to clarify rules over sexually explicit AI-generated images, after two fake…

2 days ago