Tesla Inc. is reportedly disbanding its Dojo supercomputer team, with its leader set to leave the company, as per sources familiar with the situation. This move is seen as a significant shift in the automaker’s initiative to create in-house chips for driverless technology. Peter Bannon, who was leading Dojo, is departing, and Chief Executive Officer Elon Musk has directed that the initiative be discontinued, according to sources who requested anonymity due to the sensitivity of the internal discussions. According to sources, the team has recently seen a loss of approximately 20 workers to the newly established DensityAI. The remaining employees at Dojo are reportedly being reassigned to various data center and compute projects within Tesla.
Tesla is set to enhance its dependence on external technology partners, with sources indicating collaborations with Nvidia Corp. and Advanced Micro Devices Inc. for computing needs, as well as Samsung Electronics Co. for chip manufacturing. Last month, the electric vehicle manufacturer finalized a substantial agreement worth $16.5 billion with Samsung, aimed at securing AI semiconductors until 2033. The upcoming plant in Texas is set to produce Tesla’s next-generation AI6 chip, marking a strategic move to diversify the company’s sourcing beyond the prominent chipmaker Taiwan Semiconductor Manufacturing Co.
The decision signifies a significant transformation for a program that has been in development for years, with Dojo previously seen as a key component of Tesla’s multibillion-dollar strategy to enhance its computing capabilities in the competitive landscape of artificial intelligence. Musk subsequently took to X to express that it was illogical to allocate resources between two distinct AI chip designs, thereby confirming the development to a user who referenced the report. The Dojo system, a supercomputer designed by Tesla, is utilized to train the machine-learning models that power the company’s Autopilot and Full Self-Driving programs, in addition to its Optimus humanoid robot. The technology relies on a proprietary in-house chip referred to as the D1, which is utilized for training AI in a manner similar to Nvidia’s accelerators. The AI5 and AI6 inference chips operate independently, powering models in vehicles and robotic systems.
The computer ingests data collected from vehicles and swiftly analyzes it to enhance the company’s algorithms. Analysts have indicated that Dojo may serve as a significant competitive advantage, with Morgan Stanley projecting in 2023 that it could contribute $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. The company was founded by Ganesh Venkataramanan — the former head of Dojo — and ex-Tesla employees Bill Chang and Ben Floering. Tesla’s shares continued to fall in postmarket trading following the report, showing a decline of less than 1 percent.
Tesla is experiencing a significant departure of essential personnel this year amid challenges such as increasing competition, declining sales, and a consumer backlash against Musk’s political activities. Milan Kovac, who served as the head of engineering for Optimus, along with David Lau, the vice president of software engineering, both left their positions earlier this year. Additionally, Bloomberg reported in June that Omead Afshar, a longtime confidant of Musk, had left unexpectedly.
Musk hinted at a strategic pivot during Tesla’s most-recent quarterly earnings call, stating that 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 stated during the call on July 23. Tesla’s CEO indicated last year that the company may not continue to pursue Dojo indefinitely, suggesting a potential shift towards relying more on external partners. In January 2024, Musk stated, “We’re pursuing the dual path of Nvidia and Dojo.” “But I would think of Dojo as a long shot. But it’s a long shot worth taking because the payoff is potentially very high.”
