Elon Musk has admitted that many of the Nvidia chips originally destined for Tesla's electric vehicle production were diverted to X Corp due to logistical issues.
“Tesla didn't have anywhere to send Nvidia's chips to power them up, so the chips would have just sat in a warehouse,” the EV maker's chief executive said in a post on X on Tuesday.
Musk's post was in response to a CNBC report early Tuesday, which cited a December Nvidia memo saying that 12,000 of its flagship artificial intelligence chips, the H100 graphics processing units, that were destined for Tesla, had been redirected to X. Similar orders for X that were scheduled for delivery in January and June of this year were then redirected to the electric vehicle maker, CNBC reported.
Nvidia did not immediately respond to a request for comment.
Tuesday's post may intensify criticism that Mr. Musk's push for a strategic shift that includes a focus on robotics and self-driving cars will deprioritize Tesla's traditional auto business. Mr. Musk has been scrutinizing AI plans since January, when he threatened to take his cutting-edge tech ideas elsewhere if he wasn't given more ownership of the electric-vehicle maker.
Musk owns 13% of Tesla directly, or about 21% if outstanding options are included. He has publicly sought a 25% ownership stake to have more voting power in company matters. Tesla shareholders are due to attend the company's annual meeting on June 13 to vote on Musk's compensation package, which was recently valued at several billion dollars.
In a post on Tuesday, Musk also said that an expansion of Tesla's Gigafactory in Texas is nearly complete and will house 50,000 H100 chips. In a separate post, Musk added that about half of the $10 billion in AI-related spending Tesla expects to make this year will be internal, primarily for Tesla-designed AI inference computers and sensors in all of the company's cars, as well as Tesla's supercomputer, called Dojo.
“NVIDIA hardware accounts for about two-thirds of the cost of building our AI training supercluster,” Musk said on Tuesday, adding that Tesla plans to spend $3 billion to $4 billion on hardware from Nvidia this year.
Tesla is working on building its own supercomputer as part of its efforts to develop self-driving car technology. During the company's first-quarter earnings call in April, Musk said the company plans to increase the number of H100s in operation from 35,000 to about 85,000 by the end of the year.
Write to Kimberley Kao at kimberley.kao@wsj.com.
