Tesla Halts Dojo AI Supercomputer Amid Staff Exodus & Strategy Shift
Facing a significant downturn in global sales, Tesla is reportedly scaling back its ambitious in-house supercomputer project, known as Dojo, which was designed to power its advanced driver assistance systems. This strategic pivot comes amid a notable exodus of key personnel and a broader realignment of the company’s artificial intelligence initiatives.
The individual leading the Dojo project is said to be departing, following approximately 20 team members who have already transitioned to a rival data center firm actively recruiting former Tesla executives. The remaining staff from the Dojo team are expected to be reassigned to other data center endeavors within the company. This development is particularly striking given CEO Elon Musk’s past effusive praise for Dojo, which he once heralded as a “beast.” However, Musk himself admitted last year that he had come to view the project as a “long shot” with a low probability of success. He recently clarified the company’s stance, stating that it “doesn’t make sense for Tesla to divide its resources and scale two quite different AI chip designs.”
Dojo’s primary role was to train the intricate machine learning models underpinning Tesla’s Autopilot and “Full Self-Driving” software. The shift indicates a move towards greater reliance on external technology partners, including industry giants like Nvidia, AMD, and Samsung, for the manufacturing of AI chips. Despite this, Tesla is not entirely abandoning its internal chip development. Musk has indicated that future in-house chips, such as the AI5 and AI6, will be optimized for “inference” (applying AI models) and remain “pretty good for training” (developing AI models), with all current efforts focused on these designs.
The challenges facing the Dojo team are symptomatic of a wider struggle for talent in the hyper-competitive AI sector, where major tech companies are offering substantial incentives to poach skilled professionals. Beyond its AI efforts, Tesla has grappled with a significant “brain drain,” seeing numerous key executives depart. The automaker is also contending with the ramifications of Musk’s increasingly public embrace of far-right ideologies, a factor that some analysts believe has contributed to plummeting sales and a substantial financial impact.
Operational hurdles have compounded these issues. The initial rollout of Tesla’s robotaxi service in Austin, Texas, has been plagued by technical glitches and alarming near-miss incidents. Adding to the company’s woes, a Florida jury recently found Tesla partially responsible for a fatal collision involving its Autopilot software, ordering the company to pay nearly a quarter-billion dollars—one of its most significant legal setbacks to date.
Despite these accumulating warning signs, investor confidence appears remarkably resilient, with Tesla’s market capitalization still exceeding $1 trillion and shares having risen over ten percent in the past month. During the company’s second-quarter earnings call last month, Musk had suggested a future path of “convergence” towards a single AI chip design for “Dojo 3.” However, with the reported disbandment of the Dojo team, the future trajectory of Tesla’s in-house supercomputer development remains unclear, raising questions about whether the project was yet another ambitious, yet ultimately distracting, endeavor. For the foreseeable future, it seems increasingly likely that Tesla will continue to depend on externally sourced hardware, a trend underscored by its recent $16.5 billion agreement with Samsung for AI semiconductors.