Archive — US — techcrunch
Tesla shuts down Dojo, the AI training supercomputer that Musk said would be keyto full self-driving
Tesla disbands its Dojo team, ceasing in-house efforts for AI training chips pivotal to self-driving tech. Lead Peter Bannon departs, and about 20 former employees have launched DensityAI. The shift away from Dojo marks a strategic change for Tesla, moving to external providers like Nvidia, affecting Musk's vision of a robust AI-driven future for the automaker.
Key points
- Tesla ends its Dojo project, dismantling the AI training supercomputer team led by Peter Bannon.
- Around 20 employees left to create DensityAI, aiming to build hardware for AI applications.
- Dojo was viewed as essential for full self-driving capabilities, designed to process large volumes of video data.
- Elon Musk had forecasted significant contributions from Dojo to Tesla's market value, predicting $500 billion potential impact.
- The focus is shifting to external partnerships with companies like Nvidia and AMD for computing solutions.
Why it matters
- This change signifies a broader trend in 2025 where companies rely increasingly on external partnerships for AI technology due to the high costs and risks involved in in-house development.
- The move away from Dojo aligns with the AI industry's transition towards robust collaborations rather than standalone projects, reflecting trends seen in recent tech mergers and acquisitions.
Tesla's closure of Dojo highlights a pivotal strategic realignment, focusing on collaboration with external tech firms, which may affect its AI aspirations significantly and is indicative of broader trends in the tech industry toward partnership-driven innovation.