Tesla Ends Dojo Supercomputer Project, Shifts to Nvidia and Samsung

Tesla, Inc. has decided to shut down its Dojo supercomputer project, a move that marks a significant shift in the company's strategy for AI and autonomous driving technology. The decision involves dissolving the Dojo team and reallocating its members to other data center and compute projects within Tesla.

The Dojo project, which included the development of the custom D1 chip, was initially intended to process video data from Tesla vehicles and support the company's autonomous driving and robotics ambitions. However, Tesla will now increase its reliance on external partners such as Nvidia, AMD, and Samsung for its computing needs.

The head of the Dojo project, Peter Bannon, has left the company, and around 20 former team members have joined DensityAI, a startup focused on developing data center chips. This transition follows Tesla's recent $16.5 billion agreement with Samsung to produce AI6 chips for its vehicles and robots.

Tesla's decision to end the Dojo project comes after previous optimistic projections that it could significantly enhance the company's market value. The company had been deploying Nvidia GPUs in its data centers and had announced a large-scale Nvidia GPU training cluster known as Cortex. This strategic pivot underscores Tesla's ongoing efforts to streamline its AI initiatives and leverage external expertise in chip manufacturing and computing power.