Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Bloomberg: Tesla replaces Head of Dojo

This site may earn commission on affiliate links.

Tesla Inc.’s Dojo supercomputer project lead Ganesh Venkataramanan has left the company, according to people familiar with the matter.
Venkataramanan, who has lead the Dojo project for the last five years, departed the EV maker last month, the people said, asking not to be identified discussing confidential information. Peter Bannon, a former Apple Inc. executive and director at Tesla for the last seven years is now leading the project.

Elon Musk and representatives for Tesla didn’t immediately respond to requests for comment.

The Dojo supercomputer is being designed to handle massive amounts of data, including video from Tesla cars, which are needed to train autonomous-driving software. At the core of the hardware is Tesla’s proprietary D1 chip.

In recent weeks, Tesla also installed hardware for Dojo at a centralized location in Palo Alto, California, two of the people said. Dojo has relied on multiple data centers in different locations.
As of Wednesday, Venkataramanan was no longer appearing in Tesla’s internal directories, one of the people said.
 
Tesla Inc. has lost another key member of the team working on artificial intelligence and a supercomputer to develop autonomous technology.

Bill Chang, a principal system engineer working on AI, left Tesla in October. Chang updated his LinkedIn profile on Thursday to say he’s stepping back from the Dojo supercomputer project. He had appeared on stage at Tesla’s 2022 AI day, explaining the infrastructure, technology, power and cooling challenges associated with Dojo.
 
  • Informative
Reactions: diplomat33
It looks like the AI Infra team are the folks building the AI cluster. One of their last positive news stories was TSLA's $300 Million AI cluster going live Aug 2023. That was courtesy of TSLA purchasing 10k NVDA H100 GPUs. Now with Dojo likely being delayed it's off to greener pastures.
 
Tesla currently has a very small amount of AI infrastructure in-house, with only ~4k V100s and ~16k A100s. Compared to the other large tech companies of the world, this is a very small number, given firms like Microsoft and Meta have 100k+ GPUs, and they look to double those numbers over the short to medium term. Tesla’s weak AI infrastructure is partially due to multiple delays with their in-house D1 training chip.


It's difficult to fly with the eagles when ....