Elon Musk: Tesla’s Dojo Supercomputer to Power Space-Based AI
▼ Summary
– Elon Musk announced Tesla is restarting its Dojo3 AI chip project, now dedicated to “space-based AI compute” rather than Earth-bound self-driving models.
– This follows Tesla’s shutdown of its Dojo team five months ago, which led to staff departures and a reported increased reliance on partners like Nvidia.
– Musk stated the revival decision is based on the good progress of Tesla’s in-house AI5 chip, designed for automated driving and Optimus robots.
– Tesla is actively recruiting to rebuild the Dojo team, with Musk positioning the project as an ambitious “moonshot” for space-based computing.
– The concept aligns with Musk’s broader vision, leveraging SpaceX for launch capabilities, though significant technical challenges like cooling in a vacuum remain.
Tesla is shifting its strategy to develop a new generation of supercomputing hardware designed for extraterrestrial applications. Elon Musk announced the revival of the company’s Dojo3 project, stating it will be dedicated to “space-based AI compute.” This marks a significant pivot, coming just months after Tesla disbanded its internal Dojo team and appeared to abandon its custom silicon ambitions in favor of suppliers like Nvidia and Samsung. The decision to restart development is reportedly based on confidence in Tesla’s internal chip roadmap, particularly the progress of its AI5 design.
The original Dojo initiative was effectively shut down earlier this year following the departure of key personnel, including project lead Peter Bannon. Approximately twenty team members left to join a new AI infrastructure startup. At that time, industry reports indicated Tesla would increase its reliance on external partners for both chip manufacturing and computational power. Musk’s latest declaration suggests another strategic reversal, positioning the resurrected Dojo3 as a specialized component for AI processing in space, separate from the terrestrial AI5 and AI6 chips destined for vehicles and robots.
Musk is now actively recruiting engineers to rebuild the team, soliciting applications via a public post on his social media platform. He framed the opportunity as working on what could become “the highest volume chips in the world,” inviting candidates to email Tesla with details of tough technical problems they have solved. This recruitment drive underscores the project’s renewed priority within the company’s broader technology portfolio.
This announcement arrives at a competitive moment. Rival chipmaker Nvidia recently introduced an open-source AI model for autonomous driving, posing a direct challenge to Tesla’s Full Self-Driving software. Musk acknowledged the difficulty of solving rare edge cases in driving and expressed hope for Nvidia’s success. However, his focus appears to be expanding beyond terrestrial automotive AI. He and other tech leaders have publicly speculated about the future limitations of Earth-based data centers, citing strained power grids as a major constraint.
The concept of orbital data centers is gaining traction among industry executives. Musk holds a unique advantage in this speculative field through his control of SpaceX, which provides critical launch capabilities. Reports suggest he envisions using a future SpaceX public offering to help finance a constellation of computational satellites, launched by the Starship vehicle, that could operate on continuous solar power. This aligns with the newly stated purpose for the Dojo3 system.
Substantial engineering hurdles remain for space-based computing, however. Managing the intense heat generated by high-power processors in the vacuum of space presents a formidable cooling challenge that current technology has not solved at the required scale. Musk’s pattern often involves proposing ambitious, long-term visions that push the boundaries of existing technology, and the plan for space-based AI compute follows this familiar trajectory. The revival of Dojo3 represents the first concrete step toward attempting to brute-force that particular vision into reality.
(Source: TechCrunch)
