In a surprising turn of events, Tesla CEO Elon Musk has confirmed the complete shutdown of Dojo, the company’s ambitious AI training supercomputer project, just a year after calling it essential to Tesla’s future. The announcement marks a major strategic shift for Tesla’s AI roadmap, as the company pivots its focus toward its next-generation AI5 and AI6 chips.
Musk revealed the decision on X, the social platform he owns, stating that “all paths converged to AI6,” rendering Dojo 2 obsolete. The second-generation D2 chip, which was meant to power Dojo 2, has been shelved. Instead, Tesla will focus on AI5 — built primarily for Full Self-Driving (FSD) — and AI6, designed for both self-driving and humanoid robotics, as well as large-scale AI training.
The Rise and Fall of Dojo
Originally introduced in 2019, Dojo was positioned as a cornerstone in Tesla’s mission to achieve full self-driving and commercialize humanoid robots. The project involved a mix of in-house D1 chips and Nvidia GPUs, with plans for a large-scale Dojo 2 facility in Buffalo, New York, backed by a $500 million investment.
However, by mid-2024, public discussion of Dojo faded as Musk began promoting a new AI training initiative called “Cortex,” based at Tesla’s Austin headquarters. While it’s unclear if Cortex is still active, the final nail in Dojo’s coffin came this week with Musk labeling the project an “evolutionary dead end.”
Why Tesla Pulled the Plug
According to Musk, maintaining two distinct AI chip architectures — Dojo’s D2 chip and the AI5/AI6 family — would have unnecessarily divided Tesla’s resources. By consolidating efforts into a single, scalable design, Tesla aims to simplify infrastructure, reduce cabling complexity, and cut costs significantly.
AI6, in particular, promises to handle both AI training and real-time inference, meaning it could power Tesla’s self-driving vehicles while also supporting its Optimus humanoid robot ambitions. This dual-function capability makes AI6 more attractive than the specialized approach behind Dojo.
Industry Impact
The shutdown of Dojo is significant in the tech and automotive sectors. Analysts note that while Tesla’s AI ambitions remain strong, the abrupt shift raises questions about its long-term R&D stability. The company had heavily marketed Dojo as a breakthrough in AI training efficiency, with the potential to outperform existing GPU clusters.
Some industry experts believe the decision could slow Tesla’s progress in autonomy training, at least in the short term. Others argue that focusing on AI6 may accelerate Tesla’s AI development, especially if the chip meets its performance targets.
What’s Next for Tesla’s AI Efforts
Tesla now intends to scale AI5 and AI6 across its operations, integrating them into both vehicle hardware and server-based training systems. Musk hinted that clustering multiple AI6 chips on a single board could function as a “Dojo 3,” although this would be more of a conceptual successor rather than a direct continuation of the original Dojo vision.
This change comes amid declining EV sales and brand challenges, as Tesla faces increasing competition from automakers in both electric and autonomous vehicle markets. The company is also under scrutiny following a limited robotaxi rollout in Austin earlier this year, which faced multiple reports of erratic driving behavior.
Stay ahead in startup and tech news — visit Startup News for the latest updates and insights.
