Tesla Shuts Down Dojo AI Training Supercomputer Project, Elon Musk Calls It Crucial Step Toward Full Self-Driving

Elon Musk's Tesla has terminated its internal development program for autonomous driving chips, according to Bloomberg. The decision to disband the Dojo supercomputer project marks a strategic pivot for the automotive manufacturer.

Peter Bannon, head of the Dojo initiative, will exit the company while remaining team members will be redeployed across Tesla's data centers and computational projects. The move follows the departure of approximately 20 engineers who established DensityAI - an AI startup focusing on data center chips for robotics applications and autonomous vehicles.

This restructuring occurs during a pivotal period for Tesla as CEO Elon Musk seeks to reposition the company as an AI and robotics entity. Recent autonomous vehicle trials in Austin revealed operational challenges when Model Y test vehicles demonstrated anomalous driving behaviors with human occupants in passenger seats.

The Dojo shutdown represents a significant reversal from Musk's earlier vision. First mentioned in 2019 as the "cornerstone" for full self-driving capabilities, the project had previously been positioned to process massive video datasets through its hybrid supercomputer-chip development model.

Earlier in 2023, Morgan Stanley estimated Dojo could potentially add $50 billion to Tesla's valuation through robotaxi and software revenue streams. However, development discussions ceased by August 2024 when Musk shifted focus to Cortex - Tesla's new Austin-based AI training cluster.

Tesla now plans to strengthen partnerships with external tech providers including AMD for computing power and Samsung for chip manufacturing. The automaker recently signed a $16.5 billion agreement with Samsung to produce AI6 inference chips capable of supporting FSD systems, Optimus robots, and data center operations.

During Q2 earnings calls, Musk indicated potential convergence between Dojo's 3.0 iteration and AI6 chips. "We want to find a convergence point essentially creating the same chip," he stated, signaling reduced emphasis on in-house silicon development.

The announcement comes alongside Tesla's board offering Musk a $29 billion compensation package to maintain his focus on AI initiatives rather than diverting attention to other ventures like xAI - a more specialized AI startup he co-founded.