Tesla Disbands Dojo Supercomputer Team, Shifts to External Partnerships

Tesla (TSLA.O) is reportedly disbanding its "Dojo" supercomputer team, with the project's lead departing, according to sources familiar with the matter. This decision signifies a complete halt to Tesla's plan to promote in-house chip development in the field of autonomous driving technology. The Dojo head, Peter Bannon, will leave the company, and Elon Musk has ordered the project's closure, reports indicate. Approximately 20 team members recently jumped to the newly established DensityAI, while the remaining members will be reassigned to Tesla's other data centers and computing projects. Tesla plans to increase its reliance on external technology partners, including using Nvidia (NVDA.O) and AMD (AMD.O) solutions in the computing field, and relying on Samsung Electronics for chip manufacturing. This marks a significant turning point for a strategic project that has been planned for many years. Dojo was once regarded as a core element in Tesla's efforts to enhance its computing power in the artificial intelligence race, and an important pillar of the company's multi-billion dollar strategic layout. The system was designed independently by Tesla and is used to train machine learning models for autonomous assisted driving, full self-driving, and the humanoid robot "Optimus." It can process data collected by vehicles at high speed and optimize the company's algorithms. Analysts once believed that Dojo could become a key competitive advantage for Tesla. Morgan Stanley predicted in 2023 that it could add $500 billion to Tesla's market capitalization. According to media reports, DensityAI, which will soon be operating publicly, is developing chips, hardware, and software for artificial intelligence data centers, which will be used in robotics, AI entities, automobiles, and other fields. The company was founded by former Dojo head Ganesh Venkataramanan and former Tesla employees Bill Chang and Ben Floering. This year, Tesla has faced increasing competition, declining sales, and negative consumer reactions due to Musk's political activities, and is experiencing a loss of core talent. Optimus engineering director Milan Kovac and software engineering vice president David Lau resigned earlier this year; media reported in June that Musk's long-time confidant Omead Afshar also suddenly left the company. Last month, Tesla signed a $16.5 billion agreement with Samsung to secure a supply of AI semiconductors through 2033. A Texas factory about to begin production will produce Tesla's next-generation AI6 chips to add a supply source in addition to TSMC. Musk hinted at a strategic adjustment on a July 23 earnings call, saying that future versions of in-house developed technology may merge with partner solutions: "Considering Dojo 3 and AI6 inference chips, intuitively, we would like to find a merging path to make them essentially the same chip." In fact, Musk stated in January 2024 that the company may not continue Dojo indefinitely, but will rely more on external partners. "We are moving forward on both Nvidia and Dojo paths at the same time, but Dojo is more like a gamble - it's long-term, but the potential return is very high, worth a try." The implications of this strategic shift are significant. While some may see it as an acknowledgement of the challenges in competing in the chip manufacturing space, others may view it as a necessary step to focus company efforts on other areas such as software development and expansion into new markets. How this move will impact Tesla's future in the electric vehicle and artificial intelligence market remains to be seen.