Photo: Tesla
Tesla has officially shut down its long-running Dojo supercomputer initiative, CEO Elon Musk confirmed over the weekend, marking the end of a project first unveiled during the company’s Autonomy Investor Day on April 22, 2019.
Dojo, built around Tesla’s custom D1 chip, was designed to train the company’s AI models for Full Self-Driving and other applications at a massive scale. The project took a major leap forward in January 2024 when Tesla announced plans for a $500 million Dojo-powered supercomputer in Buffalo, New York. However, less than two years after former Dojo lead Ganesh Venkataramanan’s departure, the effort has now been scrapped entirely.
Musk explained that the pivot came after the introduction of Tesla’s upcoming AI6 chip, which he described as the “convergence architecture” for both inference (in cars and data centers) and training.
“Once it became clear that all paths converged to AI6, I had to shut down Dojo and make some tough personnel choices, as Dojo 2 was now an evolutionary dead end,” Musk said. He added that “Dojo 3 arguably lives on” in the form of large boards packed with multiple AI6 system-on-chips.
While some may see the shutdown as a failure, it’s really a fast pivot, with lessons from Dojo informing Tesla’s future AI hardware. Industry observers noted that Dojo’s unique system-on-wafer design — arranging chips in a 5×5 grid for ultra-fast interconnects — will likely influence AI6’s architecture.
Tesla last month announced a massive $16.5 billion deal with Samsung for AI6 chip production. Samsung currently builds AI4 chips for Tesla, while AI5 — which recently got out of the design phase — will be manufactured by TSMC.
For Musk, the decision to wind down Dojo was about resource focus. “It doesn’t make sense for Tesla to divide its resources and scale two quite different AI chip designs,” he said last week. With AI6 promising strong inference performance and “pretty good” training capabilities, Tesla is consolidating around it entirely.
Dojo’s legacy may be over in name, but in technology, it will live on — just in a new form.
