Tesla has shut down its in-house Dojo supercomputer project, marking a major turn in the automaker’s artificial intelligence ambitions. The decision comes after the departure of key project leaders and signals CEO Elon Musk’s focus on developing Tesla’s next-generation AI chips.
Bloomberg first reported that Dojo’s lead, Peter Bannon, is leaving Tesla, and the remaining team members will be reassigned to other data center and compute projects. The news follows the earlier exit of about 20 Dojo engineers, who have since launched DensityAI, a startup developing chips, hardware, and software for AI applications in robotics, autonomous driving, and data centers.
DensityAI was co-founded by Ganesh Venkataramanan, the former head of Dojo, along with ex-Tesla engineers Bill Chang and Ben Floering.
Musk highlights AI chip focus
Elon Musk responded to the news on X, explaining that Tesla will now focus on its next-generation AI chips rather than splitting resources between two chip designs.
“It doesn’t make sense for Tesla to divide its resources and scale two quite different AI chip designs,” Musk wrote. “The Tesla AI5, AI6 and subsequent chips will be excellent for inference and at least pretty good for training. All effort is focused on that.”
He added that using these chips in a supercomputer cluster could “reduce network cabling complexity & cost by a few orders of magnitude” and hinted that such a system might be considered a future “Dojo 3.” Musk also emphasized that the performance leap from AI4 to AI5 was unlike anything he had seen before.
A sudden turn for a long-hyped project
Launched publicly at Tesla’s first AI Day in 2021, Dojo was built around Tesla’s custom D1 chip and was designed to process huge amounts of video data to train self-driving algorithms and power the Optimus humanoid robot. In 2023, analysts at Morgan Stanley estimated it could contribute as much as $500 billion to Tesla’s market value through potential robotaxi and AI software revenue.
However, the project faced challenges, including delays and high-profile departures. With Dojo gone, Tesla will lean more on suppliers like NVIDIA, AMD, and Samsung for AI hardware. Last month, Tesla signed a $16.5 billion deal with Samsung to manufacture its AI6 chips through 2033, while Taiwan Semiconductor Manufacturing Co. (TSMC) will handle production of AI5 chips.
Strategic shift amid market pressures
The shift comes at a sensitive time for Tesla, as its board has approved a $29 billion pay package to keep Musk focused on the company’s AI and robotics push.
While investors may welcome the cost savings from ending the in-house supercomputer initiative, Tesla still faces questions about its ability to retain top talent. Key executives, including leaders from the Optimus robot and software engineering divisions, have exited this year.
The move also comes as Tesla works to strengthen its position as an AI and robotics leader while grappling with increasing EV competition, declining sales, and political controversy surrounding Musk.
The age of humanoid robots may be closer than you think. Read how Tesla’s Optimus is stepping into the spotlight.