In recent years, Elon Musk has spoken of Dojo, the AI supercomputer that will be the cornerstone of Tesla’s AI efforts. Tesla’s AI team will “double down” on Dojo in October as the company prepares to unveil its robotaxi. But what exactly is Dojo and why is it important? Here’s a simpler explanation:
Dojo is Tesla’s custom-built giant computer for training their self-driving software. They call it “Full Self Driving” or FSD for short. FSD can do some driving on its own now but still needs a person watching. Making Dojo stronger will help FSD drive itself without a person one day.
Dojo will bring all the data Tesla collects from cameras on millions of cars together. It will run lots of simulations to teach FSD better. Most companies use sensors like cameras, radar, and maps to teach their self-driving cars. But Tesla only uses cameras and thinks their AI can learn to drive just from video like people do.
Building Dojo means Tesla controls its computer chips too. Right now Tesla uses chips from Nvidia which can be expensive. Tesla’s new chips are called D1. They put many D1 chips together in boxes to make a supercomputer tile. Several tiles together in cabinets make a big room of computers called a “pod”. Multiple pods are what make up Dojo.
Having their computer chips could save Tesla money and mean less reliance on outside companies. It also lets them add more power to Dojo faster to train FSD better and quickly. A stronger Dojo could even unlock robotaxis for passengers one day.
Tesla has shown prototypes of Dojo but is still working on getting it fully running. They aim to have half the computers from Tesla chips and half from Nvidia by next year. For now, Nvidia still provides most of Dojo’s power. But this is Tesla’s ambitious plan to build a giant AI machine that could be key to unlocking self-driving cars and more.
