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

Tesla is breaking the team behind her Doso super computerIn-house chips for driverless technology, ending the automaker drama, according to BlombergThe

The leadership of the dose, Peter Bannon is leaving the company, and according to the Bloomberg report, other data centers and counting projects will be re -appointed in Tesla that cited anonymous sources.

Tesla’s duzo attempts to break the departure of about 20 workers who left the automaker to start their own AI company Danceta saysThe The new startup is coming out of the stealth soon and is creating chips, hardware and software that data centers for AI are used in robotics, by AI agents, and automotive applications. Density was established by the ex -Dozo Head Ganesh Venktaramanan And former-Tesla employee Bill Chang and Ben Flooring.

It also comes at an important time for Tesla.

CEO Elon Kasturi pushed To get shareholders as AI and Robotics Company, in Austin last June, despite the limited Robotuxi launch in Austin, a model Wi -vehicle was featured with a man in the front passenger seat and the resulting driving behavior was demonstrated by various incidents of driving.

The decision to shut down Tesla’s doso, which Kasturi has been talking since 2019, is a major change in the strategy. CEO Elon Mask says that the dose is the basis of Tesler AI aspiration and “aim to reach its entire self-driving because of the skill of processing the really huge amount of video data.” He talked about the dose, though briefly, recently as a call to earn a second-quarter of the company.

In 2023, Morgan Stanley predicted that doso could add $ 500 billion to the company’s market value by unlocking new earning streams in the form of robotaxis and software services. Last year, Kasturi mentioned that Tesla’s AI team “Double down“Lead to the dose Tesla Robotuxi reveal, Which happened in October.

However, talking about doses around August 2021, when the touteter started instead of the musk cortex, Tesla “Tesla” was built to resolve the real-world AI in Tesla HQ “Giant New AI Training Superclister. “

The duo project was one part of the supercomputer, a portion of in-house chip. Tesla unveiled her D1 chip when it was officially announced on the first AI day in 2021. Venkater Presenting the chip, which Tesla said that the Nvidia GPU along with the GPU would be used to strengthen the dozo super computer. Automaker also says that it is working on a next Jenner D2 chip that will solve any information flow to any of its predecessors.

Sources have told Bloomberg that Tesla is now planning to increase dependence on NVIDIA, as well as for chip manufacturing for AMD for Compute and Samsung, like other external technology partners. Tesla signed last month With Samsung .5 16.5 billion deals To create AI 6 inference chips, a chip design that promises FSD and Tesla Optimus Humanoid Robot to scale all means to training high-performance AI at data centers.

During Tesla’s second-third income call, the musk indicates possible unnecessary.

“Thinking about Doso 3 and AI 6 infence chip, it seems like intuition, we want to try to find the transformation there, where it is basically the same chip,” said Kasturi.

As the news comes Tesler board musk provides a $ 29 billion salary package To help him move forward in Tesla and to move forward with the company’s AI effort rather than being too separated by his other companies, including more pure-play AI Startup Jai.

TechCrunch has reached Tesla for more information.

Have a sensitive tip or confidential document? We are reporting on the internal effectiveness of the AI industry – starting from companies to its future, starting from the advertising agencies affected by their decisions. Rebecca.belllan@techcrunch.com and Maxwell Jeff Maxwell.zf@Techcranch.com Reaches Belan. For protected communication, you can contact us through the signal on @Rebeccalan .491 and @Mizf .88.