Archive · 10 Aug 2025
Tesla Abandons Dojo: Impacts on AI, Self-Driving Plans
Tesla, the electric car maker, has stopped working on its own AI chips called Project Dojo.
This project was supposed to help cars learn to drive themselves.
Instead of making their own chips, Tesla will now use chips from companies like Nvidia and AMD, and have Samsung make them.
Dojo was designed to train a special computer brain, a neural network, that helps cars see and make decisions like humans.
Tesla wants its cars to drive using cameras, like how people see, not other sensors.
They are now focusing on a new system called Cortex to train their AI for self-driving.
Tesla ended Project Dojo, its in-house AI chip development initiative.
The decision shifts focus to Nvidia, AMD, and Samsung for AI chip solutions.
Dojo was intended to train neural networks for Tesla's Full Self-Driving (FSD) program.
Tesla will use Cortex, a new AI supercluster, for real-world AI challenges.
Tesla's strategy involves vision-based autonomous driving, relying on cameras and neural networks.
- Who
- Tesla, led by Elon Musk.
- What
- Tesla ended Project Dojo, their in-house AI chip program.
- When
- In August 2024, Musk began promoting Cortex.
- Where
- Tesla's headquarters in Austin.
- Why
- To streamline AI chip development, relying on external partners.
Previous Strategy
Current Strategy
In-House vs. External Chip Development
Previous Strategy
Develop its own chips in-house to reduce reliance on external suppliers and potentially increase efficiency.
Current Strategy
Reliance on external partners for chips
- Project Dojo Goal
- Develop in-house AI chips for autonomous driving
- New Supercluster
- Cortex
- Primary Supplier
- Nvidia
- Manufacturing Partner
- Samsung
- Focus
- Vision-based autonomous driving