Tesla Ends Development of In-House AI Supercomputer
In a significant strategic pivot, Tesla has reportedly disbanded its Dojo supercomputer project team. The initiative, once hailed as critical for achieving full self-driving capabilities, involved developing custom chips to process massive volumes of video data. Former team leader Peter Bannon is departing, while remaining members shift to other Tesla projects.
Exodus of Talent Sparks New Venture
Approximately 20 former Dojo engineers have launched DensityAI, a startup focused on AI infrastructure for data centers. Co-founded by ex-Tesla executives Ganesh Venkataramanan, Bill Chang, and Ben Floering, the company plans to offer AI hardware and software solutions for automotive, robotics, and AI-driven applications.
Redirection Toward External Partnerships
Tesla now appears to prioritize collaborations with established tech firms like Nvidia, AMD, and Samsung. A recent $16.5 billion deal with Samsung aims to produce Tesla’s AI6 chips, designed for autonomous systems and data center-scale training. This follows CEO Elon Musk’s recent emphasis on the Cortex supercluster as Tesla’s new AI focus.
Impact on Autonomous Driving Ambitions
The shutdown raises questions about Tesla’s self-driving roadmap. Despite Musk positioning Tesla as an AI leader, the company’s June 2025 robotaxi pilot saw mixed results, including reported operational challenges. Analysts previously estimated Dojo could unlock $500 billion in value through robotaxis and AI services.
Leadership Decisions Under Scrutiny
The move coincides with Tesla’s board offering Musk a $29 billion compensation package to retain his focus amid competition for AI talent. Meanwhile, technical challenges and Musk’s prioritization of xAI, his standalone AI venture, continue to influence Tesla’s trajectory in autonomous technology.
