- Elon Musk redirected $500 million in AI processors from Tesla to X, CNBC reported.
- The move could delay Tesla's development of autonomous vehicles and humanoid robots.
Elon Musk diverted $500 million worth of AI processors meant for the carmaker, according to correspondence from Nvidia employees obtained by CNBC.
Instead, he sent them to his social media platform X, the report, which was published Tuesday, said.
"Elon prioritizing X H100 GPU cluster deployment at X versus Tesla by redirecting 12k of shipped H100 GPUs originally slated for Tesla to X instead," an Nvidia memo from December said, according to the report. "In exchange, original X orders of 12k H100 slated for Jan and June to be redirected to Tesla."
Business Insider did not review the documents. Nvidia declined to comment.
The news comes after Musk said in the carmaker's first-quarter earnings call that the company would double the number of H100s by the end of the year. The chips, which cost upwards of $40,000, power AI technologies like ChatGPT and Anthropic. Delaying the GPUs could set back Tesla's supercomputer development and hinder progress on self-driving cars and robots.
The billionaire previously posted on X that Tesla would spend around $10 billion this year in "combined training and inference AI."
But an Nvidia email from April said that the chip projection and spending estimate "conflicts with bookings," according to the report. The email also said job cuts could further delay an "H100 project" at Tesla's Texas Gigafactory, the report said.
The billionaire responded to a post on X with a screenshot of the CNBC report and said about half of the $10 billion in AI-related expenditures would be internal. That includes Tesla-designed AI inference computers and sensors in all of the cars, as well as Tesla's supercomputer project Dojo, he said.
"For building the AI training superclusters, NVidia hardware is about 2/3 of the cost," Musk said in the post on X. "My current best guess for Nvidia purchases by Tesla are $3B to $4B this year."
Musk previously posted on X that he was "uncomfortable" expanding the EV company's AI and robotics capabilities without 25% of voting control. He said he would "prefer to build products outside of Tesla" until he had that control.
The diversion of chips from Tesla highlights the ongoing tension between Musk and shareholders who question his commitment to the EV company.
One analyst asked Musk during the first-quarter earnings call if the CEO was spread too thin. Musk also owns and is the CTO of X, and serves as the CEO of SpaceX, Neuralink, and The Boring Company.
But, according to Musk, the diversion wasn't about prioritizing X.
"Tesla had no place to send the Nvidia chips to turn them on, so they would have just sat in a warehouse," Musk said in a post on X in response to a screenshot of the original report.
He added that the south extension of the Giga Texas factory is almost complete and it will house 50,000 H100s for Tesla's Full Self-Driving software training.
The news comes just ahead of Tesla's annual shareholder meeting on June 13th. It adds pressure as Musk faces an already contentious battle for his proposed $55 billion pay package, with several investment firms advising against it. Earlier this year, a Delaware judge blocked the package, citing concerns about Musk's undue influence over the board because of his close ties to several members.
The executive pay plan, first approved in 2018, involves a 10-year grant of 12 tranches of stock options that are vested when Tesla hits specific targets. When the company hits each goal, Musk gets stock equal to 1% of outstanding shares. The EV giant says it hit all 12 targets as of 2023.
Tesla did not respond to a request for comment.