Tesla refocuses chips after Dojo

Tesla ends the Dojo project and pivots to AI5 and AI6 chips to power self driving and robots, reshaping its hardware strategy.

Musk Shifts Tesla Focus to AI5 and AI6 Chips After Dojo Closure

Tesla has disbanded its Dojo supercomputer team as it shifts to a unified AI chip strategy. Musk says AI5 and AI6 chips will handle training and inference for Autopilot and Full Self-Driving, and will also power consumer products like Optimus robots. Dojo was meant to train Tesla's autonomous driving AI using a 5x5 wafer design. Tesla will now rely on Nvidia, AMD, and Samsung for external chips.

The move coincides with Samsung's Texas fabrication ramp and Musk's pledge to speed AI6 production. Shares rose about 2.3% on August 8 as investors weighed the shift against looming challenges, including the end of EV tax incentives on September 30. Wall Street researchers express cautious sentiment, noting potential headwinds for Tesla's AI roadmap and demand for EVs.

The shift reveals a broader change in how Tesla builds AI hardware. Rather than maintaining separate Dojo and in-house chip lines, the company now pursues a single architecture intended to serve both driving software and consumer devices. That simplification could speed development but increases reliance on Samsung's fabrication and external suppliers. The risk is not just technical; it affects how investors evaluate Tesla's growth plan amid tax incentive sunsets.

If AI6 ramps on schedule, Tesla could accelerate its software updates and robot ambitions. If not, the plan may create delays and capital drag. The real test will be material availability, chip yields, and regulatory signals around subsidies.

Highlights

The hardware path Tesla chooses will show if the company can translate ambition into reliable performance.