Tesla has officially disbanded its Dojo AI training supercomputer team, according to a confirmation from CEO Elon Musk, who stated the project had become an “evolutionary dead end.” This decision comes just weeks after Musk had indicated that Tesla’s second Dojo cluster was expected to be operating “at scale” in 2026.
Musk explained the rationale behind the shutdown on X, the social media platform he owns, on Sunday. He posted, “Once it became clear that all paths converged to AI6, I had to shut down Dojo and make some tough personnel choices, as Dojo 2 was now an evolutionary dead end.” He suggested that a form of “Dojo 3” might persist as a large number of AI6 Systems-on-a-Chip (SoCs) on a single board.
The original plan involved powering Tesla’s first Dojo supercomputer with a combination of Nvidia GPUs and in-house-made D1 chips. Subsequently, Tesla had intended to construct a second Dojo facility, referred to by Musk as “Dojo 2,” which would have utilized a second-generation D2 chip. However, it appears that the D2 chip, along with the broader Dojo project, has been shelved. Tesla is now re-focusing its efforts on its AI5 and AI6 chips, which are being manufactured by TSMC and Samsung, respectively.
The AI5 chip is primarily designed to power FSD, Tesla’s driver assistance system. In contrast, the AI6 chip is engineered for both onboard inference, intended to facilitate self-driving capabilities in vehicles and autonomous functions in humanoid robots, and for large-scale AI training. Musk emphasized the strategic shift, stating on Friday evening, “It doesn’t make sense for Tesla to divide its resources and scale two quite different AI chip designs. The Tesla AI5, AI6 and subsequent chips will be excellent for inference and at least pretty good for training. All effort is focused on that.”
He further elaborated on the benefits of consolidating chip designs for supercomputer clusters. Musk suggested that integrating “many AI5/AI6 chips on a board, whether for inference or training,” would significantly reduce network cabling complexity and cost by “a few orders of magnitude.” He mused, “One could call that Dojo 3, I suppose.”
Musk has been an advocate for Dojo since 2019, consistently positioning it as a foundational element in Tesla’s pursuit of full self-driving capabilities and the commercialization of humanoid robots. However, discussions about Dojo began to subside around August 2024, when Musk started promoting “Cortex,” which he described as a “giant new AI training supercluster being built at Tesla HQ in Austin to solve real-world AI.” The current status of Cortex remains unclear, and inquiries to Tesla regarding its progress and the fate of the $500 million Dojo facility built in Buffalo, New York, have not yet been publicly addressed.
This strategic pivot in Tesla’s AI development coincides with a challenging period for the company. Tesla is currently experiencing a downturn in electric vehicle sales and has faced significant brand damage, partly attributed to Musk’s increased involvement in politics. Despite these challenges, Musk has continued to assure investors of Tesla’s future in autonomy. This commitment comes after a slow and limited robotaxi launch in Austin in June, which reportedly led to numerous incidents of problematic driving behavior by the autonomous vehicles.
