Tesla has shut down its in-house Dojo supercomputer team in a major shift of its artificial intelligence strategy. According to Bloomberg News, CEO Elon Musk ordered the disbandment of the team, with project head Peter Bannon leaving the company.
Launched as a highly ambitious initiative, Dojo was designed to train Tesla’s self-driving neural networks on a massive scale, pushing the boundaries of AI training performance. However, the company is now pivoting toward AI technologies with more immediate commercial impact. The focus will shift to the development of advanced inference chips—specialized processors that run AI models and make real-time decisions in vehicles—critical for enhancing Tesla’s Full Self-Driving (FSD) capabilities and other live AI functions.
Musk acknowledged Dojo’s significance but stressed that accelerating progress in inference chip technology offers greater near-term benefits for Tesla’s product lineup. These chips directly influence real-world driving performance, enabling faster, more reliable AI decision-making in autonomous systems.
As part of the restructuring, former Dojo engineers will be reassigned to other AI and hardware teams, ensuring their skills remain within Tesla’s innovation pipeline.
The move underscores Tesla’s pragmatic approach to AI investment—prioritizing solutions that can deliver immediate improvements for customers over long-term, capital-intensive research projects that may take years to realize commercial returns.
