Elon Musk has talked about Dojo for years, Tesla’s AI supercomputer that will supposedly be the cornerstone of the automaker’s autonomy ambitions. In fact, not too long ago, Musk said Tesla would “double down” on Dojo.
But lately, Musk has been quiet on Dojo. Instead, Tesla has for the past few months been touting the new “Cortex” supercomputer, which Tesla claimed in August 2024 would be “the world’s most powerful AI-training cluster.”
In Tesla’s shareholder deck, the company noted that it completed the deployment of Cortex in the fourth quarter, a training cluster built at Gigafactory Texas that’s made up of roughly 50,000 H100 Nvidia GPUs.
“Cortex helped enable V13 of FSD (Supervised)1, which boasts major improvements in safety and comfort thanks to 4.2x increase in data, higher resolution video inputs, 2x reduction in photon-to-control latency and redesigned controller, among other enhancements,” reads the report.
Dojo was meant to be Tesla’s custom-built supercomputer designed to train FSD, and eventually, Optimus and other real-world AI. What happened to those plans?
