Moderation Team
Tesla will restructure its artificial intelligence chip strategy by ending the Dojo supercomputer project and focusing resources on inference chips for real-time decision-making in autonomous vehicles, according to CEO Elon Musk.
In a post on X, Musk wrote: “It does not make sense for Tesla to split its resources and scale two very different AI chip designs. Tesla AI5, AI6, and future chips will be excellent for inference and at least quite good for training. All efforts are focused on that.”
His comments followed a Bloomberg report that Musk had ordered the dissolution of the internal Dojo team, with project leader Peter Bannon departing and remaining staff reassigned to other data and computing projects within Tesla. Around 20 team members recently joined a new AI startup, DensityAI.
Dojo was built around custom training chips to process large volumes of video and data from Tesla vehicles, aiming to improve the automaker’s autonomous driving software. Analysts at Morgan Stanley, led by Adam Jonas, valued Dojo at US$500 billion in 2023, calling it “a key accelerator at the intersection of hardware and software” that could open a new business area for Tesla beyond electric vehicles. Jonas has not commented on whether the shift will affect Tesla’s valuation. Musk did not mention Dojo by name in his latest remarks.
The restructuring comes amid falling EV sales, intensifying global competition, and backlash in Europe over Musk’s political positions. Tesla has also seen executive departures, job cuts, and a strategic pivot toward autonomous driving and robotics.
The company plans to produce next-generation AI5 chips by late 2026. In July, Musk announced a US$16.5 billion deal with Samsung Electronics to supply AI6 chips, without disclosing a production timeline. The AI6 chips are expected to be used in autonomous vehicles and Tesla’s humanoid robot, Optimus, and could also power broader AI applications across Tesla’s operations.
GIPHY App Key not set. Please check settings