Tesla plans to offer machine-learning training as a web service with its new ‘Dojo’ supercomputer, according to new comments from CEO Elon Musk.
Project “Dojo” was first announced by Musk at Tesla’s Autonomy Day last year:
We do have a major program at Tesla which we don’t have enough time to talk about today called “Dojo.” That’s a super powerful training computer. The goal of Dojo will be to be able to take in vast amounts of data and train at a video level and do unsupervised massive training of vast amounts of video with the Dojo program — or Dojo computer.
Dojo means “place of the Way” in Japanese and the term is often used for a place to practice meditation or martial arts.
In this case, the Dojo supercomputer will be a place for Tesla to train its Full Self-Driving AI.
Last month, Musk revealed that Tesla’s Dojo supercomputer will be capable of an exaFLOP, one quintillion (1018) floating-point operations per second, or 1,000 petaFLOPS.
This would make Dojo a truly cutting-edge supercomputer considering there’s currently a race to break the exaFLOP barrier in supercomputing with companies like Intel and AMD, along with governments, in the running.
In a series of tweets that started through a response to famed software engineer John Carmack, Musk confirmed that Tesla plans to open the supercomputer to the public as a web service to train machine learning models:
Tesla’s CEO added about the Dojo supercomputer project:
Dojo uses our own chips and a computer architecture optimized for neural net training, not a GPU cluster. I could be wrong, but I think it will be best in world.
The automaker has been building its own chip architecture expertise with a team of chip experts over the last five years in order to develop their onboard self-driving computer, which was released last year.
There’s currently no clear timeline for Tesla to have Dojo up and running, but Musk indicated that it could be next year, and there was a recruiting effort for the program last month.