Elon Musk stated over the long weekend that Tesla goals to start over work on Dojo3, the electrical vehicle corporation’s formerly discarded third-generation AI chip. Only this time, Dojo3 won’t be targeted at training self-driving models on Earth. Instead, Musk stated it is going to be committed to “space-based AI compute.”
The move comes 5 months after Tesla efficiently shut down its Dojo effort. The corporation dismissed the team behind its Dojo supercomputer following the departure of Dojo lead Peter Bannon. Around 20-Dojo employees also left to join DensityAI, a latest AI infrastructure startup founded by former Dojo head Ganesh Venkataramanan and ex-Tesla employees Bill Chang and Ben Floering.
At the time of Dojo’s shutdown, Bloomberg mentioned Tesla deliberate to boost its reliance on Nvidia and other companions like AMD for compute and Samsung for chip production, instead of ongoing evolving its very own custom silicon. Musk’s recent comments assist the strategy has shifted once more.
The billionaire executive and Republican megadonor stated in a post on X the decision to restore Dojo was primarily based on the state of its in-house chip roadmap, observing that Tesla’s AI5 chip design was “in good shape.”
Tesla’s AI5 chip, made by using TSMC, was designed to power the automaker’s computerized driving functions and Optimus humanoid robots. Last summer time, Tesla signed a $16.5 billion deal with Samsung to construct its AI6 chips that promise to power Tesla vehicles and Optimus, aw well as allow high-performance AI training in data centers.
“AI7/Dojo3 may be for space-based AI compute,” Musk stated on Sunday, locating the resurrected venture as more of a moonshot.
To attain that, Tesla is now preparing to rebuild the team it demolished months ago. Musk used the same to hire engineers directly, writing: “If you’re interested in operating on what will be the very highest volume chips within the world, send a note to AI_Chips@Tesla.Com with 3-bullet factors on the toughest technical issues you’ve solved.”
The timing of the declaration is extremely good. At CES 2026, Nvidia disclosed Alpamayo, an open source AI model for self autonomous driving that directly demanding situations Tesla’s FSD software. Musk commented on X that solving the long tail of uncommon edge in using is “super tough,” including: “I genuinely hope they succeed.”
Musk and numerous different AI executives have claimed the future of data facilities may also lie off-planet, since Earth’s power grids are already strained to the max. Axios recently reported Musk rival and OpenAI CEO Sam Altman is likewise excited by the possibility of placing data centers into orbit. Musk has an part over his peers because he already controls the release vehicles.
Per Axios, Musk plans to use SpaceX’s upcoming IPO to assist finance his vision of using of Starship to release a constellation of compute satellites that could function in steady sunlight, harvesting solar power 24/7.
Still, there are numerous roadblocks to making AI data centers in space a opportunity, no least the challenge of cooling high-power compute in a vacuum. Musk’s comments of Tesla constructing “space-based AI compute” fit a acquainted pattern: go with the flow an idea that sounds a far-fetched, then try to brute-force it into reality.
