Tesla’s Dojo, a timeline

Tesla has officially ceased its custom-built Dojo supercomputer project, disbanding its team, despite years of development and over $1 billion in projected investment for Full Self-Driving (FSD) neural network training. CEO Elon Musk confirmed the shutdown, stating Dojo 2 was an "evolutionary dead end" and that resources would converge on the AI6 chip, designed for both inference and training. This strategic pivot follows the deployment of "Cortex," a new supercomputer primarily utilizing Nvidia H100/H200 GPUs, signaling a significant shift in Tesla's core AI infrastructure strategy towards more integrated and commercially available hardware solutions.

Analysis

Tesla has officially terminated its proprietary Dojo supercomputer project, a multi-year, high-risk initiative once touted as central to its Full Self-Driving (FSD) ambitions. This strategic pivot marks a significant shift away from developing custom AI training hardware, with the company now concentrating resources on a dual strategy: leveraging commercially available hardware in the near term and developing a unified chip for the long term. The immediate replacement for Dojo is Cortex, a new supercomputer cluster powered by tens of thousands of Nvidia H100/H200 GPUs, for which Tesla has committed to spending billions. This move implicitly acknowledges the superior performance and scalability of Nvidia's offerings and represents a capitulation in the custom AI chip race. The long-term vision now hinges on the AI6 chip, part of a $16.5 billion deal with Samsung, which aims to converge in-vehicle inference and data-center training capabilities. The shutdown of Dojo, which CEO Elon Musk labeled an "evolutionary dead end," follows the departure of key talent, including project lead Peter Bannon, and represents a material write-off of R&D investment.

Key Decisions for Investors

• Investors should re-evaluate the vertical integration premium in Tesla's valuation, as the Dojo shutdown signals a critical dependency on Nvidia for its near-term AI roadmap, potentially impacting margins and introducing supply chain risk.
• The pivot strongly validates Nvidia's market dominance in AI compute, with Tesla's multi-billion dollar GPU orders providing further support for Nvidia's near-term revenue growth and reinforcing its status as the key enabler of the AI industry.
• Monitor Tesla's execution on the AI6 chip development with Samsung, as its success is now the primary determinant for the company's long-term, vertically integrated AI strategy, distinct from its current reliance on third-party hardware.
• The abrupt and costly reversal of the Dojo project raises governance questions regarding capital allocation discipline and the management of long-term, high-risk R&D initiatives, which warrants closer scrutiny.