Tesla has reportedly dismantled its Dojo supercomputer team, a move that signals a significant change in the EV maker’s approach to developing autonomous driving technology.
The team’s leader, Peter Bannon, is set to leave the company, and CEO Elon Musk has directed the closure of the project, sources familiar with the situation revealed, as reported by Bloomberg.
The Dojo supercomputer, which has been a cornerstone of Tesla’s plans to enhance its AI capabilities, is known for its custom D1 chip designed to train machine-learning models for Tesla’s autopilot and full self-driving features.
The project has faced setbacks, with around twenty team members recently departing for DensityAI, a new venture.
Remaining Dojo personnel are being reassigned within Tesla to work on other data centre and computer initiatives, according to sources.
This strategic pivot also includes increasing Tesla’s dependence on external partners for technological expertise.
According to the sources, Tesla is expected to collaborate with Nvidia and Advanced Micro Devices for computing solutions and with Samsung Electronics for chip production.
The decision to shut down Dojo comes after years of development and investment. The supercomputer was once touted as a key element in Tesla’s strategy to assert dominance in the AI domain.
Financial analysts, including those from Morgan Stanley, projected that Dojo could potentially add significant value to Tesla, with estimates reaching as high as $500bn.
In a recent $16.5bn agreement, Tesla secured a long-term commitment from Samsung to supply AI semiconductors until 2033.
A forthcoming Texas facility is planned to manufacture Tesla’s next-gen AI6 chip, which would reduce Tesla’s reliance on Taiwan Semiconductor Manufacturing, another major chipmaker.
The automaker CEO has previously hinted at a possible shift in strategy during Tesla’s latest earnings call, suggesting that the company’s in-house technology might increasingly align with that of its partners.
Neither Tesla nor Bannon immediately responded to requests for comment on the matter.
However, Musk confirmed the Bloomberg News report on the social media platform X, stating that dividing resources between two different AI chip designs was not sensible.
