Tesla disbands Dojo supercomputer team as leader Peter Bannon exits

Don't want to miss the best from Business Standard?

By Ed Ludlow

 

Tesla Inc. is disbanding its Dojo supercomputer team and its leader will depart the company, according to people familiar with the matter, upending the automaker’s effort to develop in-house chips for driverless technology. 

Peter Bannon, who was heading up Dojo, is leaving and Chief Executive Officer Elon Musk has ordered the effort to be shut down, according to the people, who asked not to be identified discussing internal matters. The team has lost about 20 workers recently to newly formed DensityAI, and remaining Dojo workers are being reassigned to other data center and compute projects within Tesla, the people said. 

 

Tesla plans to increase its reliance on external technology partners, including Nvidia Corp. and Advanced Micro Devices Inc. for compute, and Samsung Electronics Co. for chip manufacturing, the people said. 

 

The decision marks a major shift for a program years in the making, with Dojo once positioned as central to Tesla’s multibillion-dollar plan to gain computing muscle in the artificial intelligence race. Tesla and Bannon didn’t immediately respond to requests for comment. 

Musk later posted on X that it didn’t make sense to divvy resources between two different AI chip designs, confirming the development to a user who shared Bloomberg News’s report.

 

The Dojo system is a Tesla-designed supercomputer used to train the machine-learning models behind the electric-vehicle maker’s Autopilot and Full Self-Driving programs, as well as its Optimus humanoid robot. It’s based on a custom in-house chip known as the D1, used in training AI much like Nvidia’s accelerators. That’s separate from the AI5 and AI6 inference chips, which run models in cars and robots.  

 

The computer takes in data captured by vehicles and rapidly processes it to improve the company’s algorithms. Analysts have said Dojo could be a key competitive advantage, with Morgan Stanley estimating in 2023 it could add $500 billion to Tesla’s value.

 

DensityAI, which is poised to come out of stealth soon, is working on chips, hardware and software that will power data centers for AI that are used in robotics, by AI agents and in automotive applications, among other sectors, Bloomberg reported this week. The company was founded by Ganesh Venkataramanan — the former head of Dojo — and ex-Tesla employees Bill Chang and Ben Floering.

 

Tesla’s shares extended postmarket declines after Bloomberg’s report, trading down less than 1 per cent as of 5:29 pm in New York. 

Talent Drain

Tesla has faced an exodus of key talent this year as it has grappled with rising competition, falling sales and a consumer backlash to Musk’s political activity. Milan Kovac, the head of engineering for Optimus, and David Lau, vice president of software engineering, departed earlier this year, while Bloomberg reported in June that longtime Musk confidant Omead Afshar had abruptly left.

 

The EV maker last month reached a $16.5 billion deal with Samsung to secure AI semiconductors through 2033. The plan is for an upcoming plant in Texas to produce Tesla’s next-generation AI6 chip, diversifying Tesla’s sourcing beyond leading chipmaker Taiwan Semiconductor Manufacturing Co. 

  Musk hinted at a strategic pivot during Tesla’s most-recent quarterly earnings call, suggesting future iterations of the company’s in-house technology could converge with that of its partners. “Thinking about Dojo 3 and the AI6 inference chip, it seems like intuitively, we want to try to find convergence there, where it’s basically the same chip,” Musk said on the July 23 call.

 

Tesla’s CEO last year acknowledged that the company might not pursue Dojo in perpetuity and instead lean more on external partners.  

 

“We’re pursuing the dual path of Nvidia and Dojo,” Musk said in January 2024. “But I would think of Dojo as a long shot. It’s a long shot worth taking because the payoff is potentially very high.”