Tesla is disbanding the team behind its Dojo supercomputer, halting its in-house chip development for autonomous technology, as reported by Bloomberg.
Dojo’s leader, Peter Bannon, is leaving the company, and the remaining team members are being reassigned to other projects at Tesla, according to Bloomberg’s anonymous sources.
This follows the exit of around 20 employees who left Tesla to form a new AI company named DensityAI. The startup, launched by former Tesla members including ex-Dojo head Ganesh Venkataramanan, Bill Chang, and Ben Floering, is developing chips, hardware, and software for AI data centers utilized in robotics, AI agents, and automotive applications.
This move comes at a pivotal moment for Tesla.
CEO Elon Musk has been promoting Tesla as an AI and robotics company, despite a limited robotaxi rollout in June in Austin using Model Y vehicles with a human in the passenger seat, which resulted in several reported driving issues.
Ending the Dojo project signifies a significant strategy shift for Tesla. Musk had previously described Dojo as essential to Tesla’s AI goals, touting its capacity to process vast video data amounts for achieving full self-driving capabilities. He only recently mentioned Dojo during Tesla’s second-quarter earnings call.
In 2023, Morgan Stanley estimated Dojo could potentially boost Tesla’s market value by $500 billion through new revenue sources like robotaxis and software services. Last year, Musk mentioned that Tesla’s AI team would intensify their efforts on Dojo before the robotaxi reveal in October.
Techcrunch event
San Francisco
|
October 27-29, 2025
However, discussions about Dojo ceased around August 2024 when Musk began promoting Cortex, Tesla’s new AI training supercluster at its Austin headquarters intended to address real-world AI challenges.
The Dojo initiative encompassed both a supercomputer and in-house chip production. Tesla introduced its D1 chip when it announced Dojo at its first AI Day in 2021, with Venkataramanan presenting it as a complement to Nvidia’s GPU for Dojo’s operations. Tesla also planned a next-gen D2 chip to solve any issues with its predecessor.
Bloomberg reports now indicate Tesla will increase reliance on Nvidia and other external partners like AMD for computing and Samsung for chip production. Last month, Tesla secured a $16.5 billion agreement with Samsung to produce its AI6 inference chips, designed to support everything from full self-driving and Tesla’s Optimus humanoid robots to advanced AI training in data centers.
During Tesla’s second-quarter earnings call, Musk spoke of potential overlaps.
“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 explained.
This development coincides with Tesla’s board granting Musk a $29 billion compensation package to retain him and further the company’s AI initiatives, deterring distractions from his other ventures, including his AI startup xAI.
TechCrunch has reached out to Tesla for more details.
Have sensitive information or documents? We’re covering the AI industry intricacies — from driving companies to the affected individuals. Contact Rebecca Bellan at rebecca.bellan@techcrunch.com and Maxwell Zeff at maxwell.zeff@techcrunch.com. For secure communication, reach us via Signal at @rebeccabellan.491 and @mzeff.88.