Tesla's CEO, Elon Musk, made an announcement stating that the company is set to re - initiate the Dojo 3 supercomputer project. Originally, this project was scheduled to be put on hold in August 2025, so this marks a substantial shift in Tesla's chip - related strategy.

The core objective of Dojo 3 has undergone an expansion. Initially, it was centered around training autonomous driving models on Earth, but now it has shifted to "space artificial intelligence (AI) computing." Earlier, Musk held the view that developing two AI chip designs concurrently would spread resources too thin. As a result, he decided to halt the Dojo 3 project and concentrate on the research and development of chips such as AI5 and AI6. However, with the AI5 chip design now progressing without major hiccups, the Dojo 3 project has been given a new lease on life.

The original Dojo project was aimed at constructing a supercomputer specifically for machine learning training. In contrast, Dojo 3 places greater emphasis on architectural restructuring and cost optimization. Industry experts point out that if Dojo 3 can successfully optimize the cost and efficiency of AI training, it will significantly speed up the commercialization of Tesla's Full Self - Driving (FSD) and Optimus.

Musk has plans to utilize the AI7 and Dojo 3 for space AI computing, in a bid to realize his long - held vision of deploying AI computing centers in space. He firmly believes that within the next four to five years, operating AI systems in orbit will prove to be more cost - effective. Nevertheless, the ambition of establishing space - based AI data centers is not without its hurdles. There are numerous challenges to overcome, including potential risks such as orbital debris, the need for regulatory approvals, and compliance with international space policies. One particularly daunting task is the cooling of high - power computing equipment in a vacuum environment.