Tesla disbands ambitious Dojo supercomputer team, shifts compute power to AMD and NVIDIA

Tesla has reportedly disbanded its Dojo supercomputer team, with its leader leaving the company and increasing its reliance on external technology partners like NVIDIA and AMD for compute power.

In a new report from Bloomberg, we have heard that Tesla is disbanding its Dojo supercomputer team, which was working on in-house AI chips for driverless technology. Peter Bannon was leading Dojo, and is departing the company with Tesla CEO Elon Musk ordering the Dojo supercomputer effort to be shut down, according to Bloomberg's sources.

The Dojo team lost around 20 workers recently to newly-formed DensityAI, with remaining Dojo workers being reassigned to other data center and compute projects inside Tesla.

Tesla will reportedly increase its reliance on external technology partners like NVIDIA and AMD for its compute power requirements, and Samsung Electronics for chip manufacturing. Samsung recently signed a $16.5 billion foundry contract with Tesla that lasts through until 2033, where it will manufacture FSD chips for the company on American soil.

Tesla's ambitious Dojo supercomputer was designed in-house, and used to train the machine learning models behind its Autopilot and FSD (Full Self-Driving) programs, as well as its Optimus humanoid robot. The Dojo supercomputer takes in all of the data captured by vehicles and processes it rapidly to improve its algorithms, with analysts saying Dojo could've been a key advantage, with Morgan Stanley estimating back in 2023 that it would add $500 billion to Tesla's value.