Elon Musk has announced that Tesla is set to revive its previously abandoned third-generation AI chip project, Dojo3. However, this time, the advanced silicon won't be dedicated to training self-driving models on Earth, but rather to "space-based AI compute," signaling a bold new direction for the electric vehicle giant's artificial intelligence ambitions.
This strategic pivot comes just five months after Tesla effectively shuttered its initial Dojo efforts. The company had disbanded the team behind its Dojo supercomputer following the departure of lead Peter Bannon, with approximately 20 Dojo workers subsequently joining DensityAI, a new 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, reports indicated Tesla planned to increase its reliance on industry giants like Nvidia and AMD for compute power, and Samsung for chip manufacturing, rather than continuing its custom silicon development. Musk's latest comments, however, suggest a significant shift back towards in-house innovation.
In a recent post on X, the billionaire executive confirmed the decision to revive Dojo was based on the positive trajectory of Tesla's in-house chip roadmap, specifically noting that the company's AI5 chip design was "in good shape." While Tesla's AI5 chip, manufactured by TSMC, powers the automaker's automated driving features and Optimus humanoid robots, and a $16.5 billion deal with Samsung covers AI6 chips for vehicles, Optimus, and high-performance AI training in data centers, Musk clarified the role of the resurrected project.
"AI7/Dojo3 will be for space-based AI compute," Musk said on Sunday, positioning the project as a truly ambitious "moonshot" endeavor.
To realize this vision, Tesla is now actively rebuilding the team it dismantled months ago. Musk directly used his X post to recruit engineers, writing: "If you’re interested in working on what will be the highest volume chips in the world, send a note to [email protected] with 3 bullet points on the toughest technical problems you’ve solved."
The timing of this announcement is particularly noteworthy. It follows Nvidia's unveiling of Alpamayo at CES 2026, an open-source AI model for autonomous driving that directly challenges Tesla's Full Self-Driving (FSD) software. Musk, commenting on X, acknowledged the difficulty of solving "the long tail of rare edge cases in driving," adding, "I honestly hope they succeed."
Musk and several other prominent AI executives have increasingly argued that the future of data centers may lie beyond Earth, citing the planet's already strained power grids. Axios recently reported that OpenAI CEO Sam Altman, a rival to Musk, is also enthusiastic about the prospect of placing data centers into orbit. Musk, however, holds a distinct advantage over his peers due to his control of SpaceX's launch vehicles.
According to Axios, Musk intends to leverage SpaceX's upcoming IPO to help finance his ambitious vision. This plan involves using Starship to launch a constellation of compute satellites designed to operate in constant sunlight, thereby harvesting solar power 24/7.
Despite the grand vision, numerous roadblocks remain for establishing AI data centers in space, not least the formidable challenge of cooling high-power compute systems in a vacuum. Musk's announcement of Tesla building "space-based AI compute" aligns with a familiar pattern: introducing a seemingly far-fetched idea, then relentlessly pursuing its realization.
