Musk Orders Dojo Team Disbanded, Marking a Strategic Shift in Tesla’s AI Ambitions

TradingKey - Tesla is undergoing a major shake-up in its AI strategy. Multiple media outlets, citing people familiar with the matter, report that Elon Musk has ordered the shutdown of the company’s in-house Dojo supercomputer project. Peter Bannon, head of the Dojo initiative, is expected to leave the company. The move signals a setback for Tesla’s ambitions to develop its own AI chips, pushing it to rely more heavily on external suppliers such as NVIDIA, AMD, and Samsung Electronics.

Originally designed to train Tesla’s autonomous driving models, Dojo was considered a cornerstone of the company's AI competitiveness. Morgan Stanley had estimated the system could potentially unlock up to $500 billion in value for Tesla. Now, with the team disbanded, around 20 members have joined a startup called DensityAI, while remaining staff have been reassigned to other internal compute projects.

Beyond Dojo’s demise, Tesla has also seen a string of high-profile executive departures this year, including Milan Kovac (Head of Engineering for Optimus), David Lau (VP of Software Engineering), and long-time senior executive Omead Afshar. This wave of talent loss has deepened uncertainty surrounding the company’s internal R&D efforts.

In an effort to stabilize its chip supply chain, Tesla signed a $16.5 billion agreement with Samsung last month. Under the deal, Samsung’s Texas facility will manufacture Tesla’s next-generation AI6 chips, helping to diversify the company’s sourcing strategy.

Sources indicate that Musk was “extremely frustrated” with delays at the Dojo data center in Austin, Texas, citing construction setbacks and incomplete tunnel infrastructure that have prevented the facility from becoming operational. These delays ultimately triggered internal shakeups and layoffs.

Despite the winding down of the Dojo project, the market generally believes that Tesla is not retreating from the AI space. During a recent earnings call, Elon Musk hinted that the company’s future in-house technologies will be more integrated with those of its partners. Against the backdrop of still-promising prospects in full self-driving and robotics, Tesla's AI strategy may be entering a new phase—asset-light and collaboration-driven.