Elon Musk said over the long weekend that Tesla aims to resume work on the electric car company’s previously abandoned third-generation AI chip, Dojo3. Only this time, Dojo3 is not intended to train self-driving models on Earth. Instead, Musk said it would specialize in “space-based AI computing.”
The move comes five months after Tesla effectively shut down its Dojo effort. The company has disbanded its Dojo supercomputer development team following the retirement of Dojo leader Peter Bannon. Approximately 20 Dojo employees also left and joined DensityAI. DensityAI is a new AI infrastructure startup founded by former Dojo director Ganesh Venkataramanan and former Tesla employees Bill Chan and Ben Floring.
At the time of Dojo’s closure, Bloomberg reported that Tesla planned to rely more on Nvidia and other partners such as AMD for computing and Samsung for chip manufacturing, rather than continuing to develop its own custom silicon. Musk’s latest comments suggest another shift in strategy.
In a post on X, the billionaire executive and major Republican donor said the decision to revive Dojo was based on the state of the company’s internal chip roadmap, noting that Tesla’s AI5 chip design is in “good shape.”
Tesla’s AI5 chip, made by TSMC, was designed to power the automaker’s self-driving capabilities and Optimus humanoid robot. Last summer, Tesla signed a $16.5 billion deal with Samsung to develop the AI6 chip that powers Tesla cars and Optimus and enables high-performance AI training in data centers.
“AI7/Dojo3 is for space-based AI computing,” Musk said on Sunday, positioning the revived project as something of a moonshot.
To accomplish that, Tesla is currently preparing to rebuild the team it disbanded several months ago. Musk used the same post to directly recruit engineers, writing, “If you’re interested in developing the world’s most produced chip, please send a note to AI_Chips@Tesla.com with three bullet points about the most difficult technical problems you’ve solved.”
tech crunch event
san francisco
|
October 13-15, 2026
What should be noted is the timing of the announcement. At CES 2026, Nvidia announced 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 edge cases while driving is “very difficult,” adding, “I honestly hope we’re successful.”
Musk and several other AI executives argue that the future of data centers may lie outside Earth, as Earth’s power grids are already stretched to the limit. Axios recently reported that Musk’s rival, OpenAI CEO Sam Altman, is also excited about the prospect of getting data centers off the ground. Mr. Musk already controls the launch vehicle, giving him an advantage over his colleagues.
According to Axios, Musk plans to use SpaceX’s upcoming IPO to fund his vision of using Starship to launch a fleet of computing satellites that can operate in constant sunlight and collect solar power 24/7.
Still, there are many obstacles to realizing AI data centers in space, not least the challenges of cooling high-power computing in vacuum. Musk’s comments about Tesla building “space-based AI computing” fit into a likely pattern. It’s about coming up with an idea that sounds far-fetched and trying to force it into reality.
