IMPORTANT POINTS:
- Tesla plans to put its AI supercomputer, Dojo, into production in July 2023.
- The company expects Dojo to become one of the world’s five most powerful supercomputers by 2024.
- The Dojo technology will be used to train neural networks using video data from Tesla’s vehicles, furthering the advancement in autonomous driving.
In May, Tesla took another step on its path to advanced artificial intelligence (AI), creating a new official Twitter account, @Tesla_AI, dedicated to reporting on the progress of its FSD software and the upcoming humanoid robot, Teslabot.
In a recent update, the dedicated AI account revealed another major development. According to a chart published on June 21, Tesla’s highly anticipated Dojo supercomputer is expected to go into production in July 2023.
According to the graphic, Tesla intends to continue building its computer until it becomes one of the five most powerful in the world by early 2024. The company plans to achieve this by leveraging its vast database of driving videos in the real world, which he has been accumulating for several years. Also, by October 2024, Tesla expects its computer to reach 100 exaflops.
What is Dojo?
Dojo is Tesla’s custom supercomputer platform, developed for AI machine learning. More specifically, the company will use Dojo to train neural networks using video data from its own vehicles. The supercomputer is expected to play a crucial role in advancing the computer vision technology that powers its autonomous driving solution.
Although Elon Musk’s electric vehicle company already boasts a powerful supercomputer based on NVIDIA GPUs, Dojo’s new offering uses chips and an entire infrastructure designed by Tesla.
In a lengthy Twitter thread, Tesla’s AI account described how the automaker is building the fundamental models for autonomous robots. On the other hand, he also discussed the company’s multi-model neural networks, which are already integrated into its fleet of electric cars.
