During the recent holiday weekend, Elon Musk announced that Tesla intends to revive development on Dojo3, the company's previously shelved third-generation AI chip. This renewed effort, however, will not focus on training terrestrial self-driving models. Musk stated the chip will now be dedicated to "space-based AI compute."

This reversal follows Tesla's effective shutdown of its Dojo project five months ago. The team behind the Dojo supercomputer was disbanded after the departure of its lead, Peter Bannon. Approximately 20 Dojo employees subsequently left to join DensityAI, a new AI infrastructure startup founded by former Dojo head Ganesh Venkataramanan and ex-Tesla staff Bill Chang and Ben Floering. When Dojo was initially halted, reports indicated Tesla planned to rely more heavily on partners like Nvidia and AMD for computing power and Samsung for chip manufacturing, rather than pursuing its own custom silicon. Musk's latest remarks signal another strategic shift.

The billionaire executive, who is also a major Republican donor, explained in a post on X that the decision to restart Dojo was influenced by the progress of Tesla's internal chip development. He noted that the design for Tesla's AI5 chip was "in good shape."

The AI5 chip, manufactured by TSMC, was created to run the automaker's automated driving systems and its Optimus humanoid robots. Last summer, Tesla also entered a $16.5 billion agreement with Samsung to produce its next-generation AI6 chips, which are intended for use in Tesla vehicles, Optimus robots, and high-performance AI training in data centers.

Musk clarified the new direction on Sunday, stating, "AI7/Dojo3 will be for space-based AI compute," framing the revived initiative as an ambitious long-term goal. To pursue it, Tesla is now preparing to reassemble the team it dissolved months ago. Musk used his social media post to recruit engineers directly, writing: "If you’re interested in working on what will be the highest volume chips in the world, send a note to AI_Chips@Tesla.com with 3 bullet points on the toughest technical problems you’ve solved."

The timing of this announcement is significant. It follows the unveiling of Alpamayo at CES 2026, an open-source AI model for autonomous driving introduced by Nvidia that directly competes with Tesla's Full Self-Driving (FSD) software. Musk commented on X that solving the persistent challenge of rare driving edge cases is "super hard," adding, "I honestly hope they succeed."

Musk and other AI industry leaders have suggested that the future of data centers may be extraterrestrial, citing the immense strain on Earth's power grids. Recent reports note that OpenAI CEO Sam Altman, a rival of Musk's, is also intrigued by the concept of orbital data centers.

Musk holds a potential advantage in this arena because he controls the necessary launch vehicles through SpaceX. Reports indicate he plans to leverage SpaceX's forthcoming initial public offering to help fund a vision of using Starship rockets to deploy a constellation of computing satellites. These satellites would operate in perpetual sunlight, enabling 24/7 solar power generation.

Nevertheless, significant obstacles remain to making space-based AI data centers feasible, including the formidable challenge of cooling high-power computing hardware in a vacuum. Musk's comments about building "space-based AI compute" align with a familiar pattern: proposing an idea that seems ambitious, then aggressively working to turn it into reality.