Tesla Dojo Supercomputer Timeline Shows Shift to New Cortex AI
Tesla's custom Dojo supercomputer, once central to Elon Musk's self-driving ambitions, appears sidelined as a new system called Cortex takes its place. A timeline of promises reveals the shifting focus of Tesla's AI training strategy.
Tesla faces a shift in its artificial intelligence strategy as the highly anticipated Dojo supercomputer takes a backseat to a new system called Cortex. Elon Musk wants Tesla to function primarily as an AI company rather than just an automaker, and Dojo represents the custom-built supercomputer originally designed to train the neural networks behind Tesla's Full Self-Driving (FSD) software. While FSD currently requires an attentive human driver, Tesla believes that massive amounts of data and computing power will eventually push the technology past the threshold into true autonomy.
Musk first teases Dojo in 2019 at Tesla's Autonomy Day, shortly after the company reveals its custom neural network chips. Throughout 2020, he promotes the supercomputer heavily, calling it a beast capable of processing vast amounts of video data and running massive hyperspace arrays. By the end of 2020, Musk tempers expectations slightly by stating that Dojo is not strictly necessary for self-driving but will ultimately make the system more than ten times safer than human drivers.
Despite years of public promises and a highly publicized ramp-up in 2024, Dojo effectively fades from the spotlight as Cortex enters the conversation. The transition highlights the incredibly challenging nature of developing custom silicon for AI training at such a massive scale. As Tesla continues to chase the elusive goal of full autonomy, the company leaves behind the Dojo timeline and pivots its computational resources toward this new supercomputer infrastructure.