Tesla shuts down Dojo, the AI training supercomputer that Musk said would be key to full self-driving

Tesla dismantles the team behind Cojo SuperCanter, and ends the car industry play in developing driver -home flakes for technology, according to Bloomberg.

Peter Bannon, Dogo’s leadership, leaves the company, and the remaining team members will be reset to another data center and the project account inside Tesla, in Bloomberg’s reports, which were martyred with unknown sources.

The solution of the Tesla Dojo efforts follows the departure of about 20 workers, who left the auto company to start their artificial intelligence company called Mensei. According to the new startup starting from the hidden soon, it builds chips, devices and programs that will operate the AI’s data centers that are used in robots, by artificial intelligence agents, and in car applications. Dengeenyai was founded by former Dojo president Gunish Venkataramannan and former employees of Tesla Bill Chang and Ben Fliring.

It also comes in a decisive time to Tasla.

The CEO Elon Musk prompted the shareholders to look at Tesla as the AI and Robotics, despite the launch of Robotaxi limited in Austin last June that appeared in the Y Model Vehicles with a person in the front passenger seat and resulted in a number of incidents reported in vehicles that show a problem driving behavior.

Tesla’s decision to close Dogo, which Musk has been talking about since 2019 is a major shift in the strategy. Musk said that Dojo will be the cornerstone of Tesla aspirations in artificial intelligence and its goal is to reach full self -driving due to its ability to “process massive quantities of video data.” He talked about Dogo, albeit shortly, recently as the company’s profit call in the second quarter.

In 2023, Morgan Stanley expected Dogo to add $ 500 billion to the company’s market value by opening new revenue flows in the form of robotics and software services. Only last year, Musk indicated that the Tesla’s Ai team will double “on Dojo in the period before revealing Tesla’s Robotaxi, which occurred in October.

But he talked about Dojo stopped around August 2024, when Musk began describing Cortex instead, “Tesla New Ai Traaining Supercluster Supercluster in Tesla was built in Austin to solve real artificial intelligence.”

The Dojo project was part of the super computers, and one part of the company. Tesla unveiled her D1 chip when she officially announced Dojo on the first day of AI in 2021. Venkataraman presented the slide, which Tesla said will be used alongside the NVIDIA graphics processing unit to operate the Dojo SuperCant computer. The auto industry also said it is working on the D2 chip of the next generation that would solve any convenience of information flow to its predecessor.

Sources told Bloomberg that Tesla is now planning to increase its dependence on NVIDIA, as well as other external technology partners such as AMD for Compute and Samsung to manufacture chips. Last month, Tesla signed a $ 16.5 billion deal with Samsung to make its AI6 inference chips, which is the design of chips promising to expand Humanoid Optimus Humanoid from FSD and Tesla to high -performance AI training in databases.

During the Tesla profit call in the second quarter, Musk alluded to possible repetition.

“Thinking about Dojo 3 and the AI6 inference slide, it seems intuitively, we want to try to find rapprochement there, as it is basically the same slide,” Musk said.

The news comes at a time when the Tesla Musk Board of Directors provides a $ 29 billion payment package to keep it in Tesla and help push the company’s efforts forward, rather than being united by its other companies, including the start of the most formulated AI operating from artificial intelligence.

TECHRUNCH contact Tesla for more information.

Do you have sensitive advice or secret documents? We report the internal business of the artificial intelligence industry – from companies whose future is to people affected by their decisions. Contact Rebecca Billan on ribecca.bellan@techcrunch.com and Mexuel Zeef on MaxWell.zeff@techcrunch.com. For safe contact, you can contact us by reference to @Rebeccabellan.491 and @mzeff.88.