Tesla Revives Dojo 3 Supercomputer for Space Applications

Tesla has made a significant change in its artificial intelligence strategy by resurrecting its supercomputing project Dojo3, which was previously considered, but now with the ambitious goal of making AI computing infrastructure above the Earth’s atmosphere. The company’s chief, Elon Musk, has already confirmed this in a message on the social media platform X.

The Dojo project aimed at building up the internal AI training needed by Tesla for its self-driving cars and robots, but it was put on hold a few months ago as a result of the company’s restructuring. At that time, the management of Tesla dismantled the Dojo team working exclusively on it and indicated that they would be relying more on the external API partners for the AI chips.

However, with his latest remarks, Musk is bringing back the importance of the in-house AI hardware. He said that the AI5 chip, which is being designed and manufactured in collaboration with the Taiwanese semiconductor company TSMC, has achieved stability and is ever so “good for Dojo3’s main development.

Importantly, the Dojo3 platform, which Musk also calls “AI7,” will be ready to take off with new capabilities rather than just continuing from where the old ones left off. Instead of building up a stronghold of Earth-based AI training, Tesla wants to open up a corridor for space-based AI computability. Musk hasn’t shared any details about the technical architecture, but the vision is to use outer space as a place for large-scale computing, possibly getting the advantages of uninterrupted solar exposure as well as the unique cooling characteristics available at that orbit.

The news comes along with a broad hiring call, with Tesla recruiting engineers who have experience in chip and AI hardware development. Elon Musk is referring to the upcoming chips as “the highest volume chips in the world,” which shows the significance of semiconductor design in the company’s long-term technological roadmap.