Elon Musk said over the weekend that Tesla plans to restart work on Dojo3, its previously abandoned third-generation artificial intelligence chip. Only this time, Dojo3 won’t aim to train self-driving models on Earth. Instead, Musk says it will be dedicated to “space-based AI computing.”
The move comes five months after Tesla effectively shut down its Dojo effort. The company has disbanded the team behind its Dojo supercomputer following the departure of Dojo chief Peter Bannon. About 20 Dojo workers as well left to subscribe to DensityAIa new AI infrastructure startup founded by ex-Dojo head Ganesh Venkataramanan and ex-Tesla employees Bill Chang and Ben Floering.
At the time of Dojo’s shutdown, Bloomberg reported that Tesla planned to increase its reliance on Nvidia and other partners such as AMD for PCs and Samsung for chip manufacturing, rather than continuing to develop its own custom silicon. Musk’s latest comments suggest that strategy has changed again.
The billionaire executive and Republican bigwig he said to one posting on X the decision to revive the Dojo was based on the state of the internal chip roadmap, noting that Tesla’s AI5 chip design was “in good shape.”
Tesla’s AI5 chip, made by TSMC, was designed to power the automaker’s automated driving features and humanoid Optimus robots. Last summer, Tesla signed a $16.5 billion deal with Samsung to make its AI6 chips that promise to power Tesla and Optimus vehicles, as well as enable high-performance AI training in data centers.
“AI7/Dojo3 will be for space-based artificial intelligence computing,” Musk said he said on Sunday, positioning the resurrected work as more moonlighting.
To achieve this, Tesla is now preparing to rebuild the team it disbanded months ago. Musk used the same post to recruit engineers directly, writing: “If you’re interested in working on what will be the world’s highest volume chips, send a note to AI_Chips@Tesla.com with 3 bullet points about the hardest technical problems you’ve solved.”
Techcrunch event
San Francisco
|
13-15 October 2026
The timing of the announcement is remarkable. At CES 2026, Nvidia unveiled Alpamayo, an open-source AI model for autonomous driving that directly challenges Tesla’s FSD software. Musk commented to X that solving the long tail of rare cases in driving is “extremely difficult,” adding, “I sincerely hope they succeed.”
Musk and many other AI executives have argued that the future of data centers may lie off-planet, as Earth’s power grids are already stretched to capacity. Axios recently reported Musk’s rival and OpenAI CEO Sam Altman is also excited about the prospect of getting data centers into orbit. Musk has an advantage over his peers because he already controls launch vehicles.
According to Axios, Musk plans to use SpaceX’s upcoming IPO to help fund his vision of using Starship to launch a constellation of computing satellites that can operate under continuous sunlight, harvesting solar energy 24/7.
However, there are many obstacles to making AI data centers in space a possibility, most notably the challenge of cooling high-powered computers in a vacuum. Musk’s comments about building Tesla’s “space-based computational AI” fit a familiar pattern: float an idea that sounds far-fetched, then try to violently make it happen.
