Archive — US — techcrunch
Tesla Dojo: the rise and fall of Elon Musk’s AI supercomputer
Tesla has shut down its AI supercomputer project, Dojo, in August 2025 after years of development and expectations from Elon Musk. Initially seen as vital for self-driving tech and robotaxis, the project's failure is attributed to a shift towards partnerships for chip development, marking a strategic pivot amidst falling EV sales and mixed responses from analysts on Musk's promises.
Key points
- Dojo was designed to train Tesla's Full-Self-Driving (FSD) neural networks.
- Despite initial optimism, Tesla ceased development, labeling Dojo an 'evolutionary dead end' in August 2025.
- Elon Musk previously emphasized Dojo's importance but later shifted to promoting the Cortex supercluster.
- The decision followed a $16.5 billion contract with Samsung for next-gen AI6 chips, indicating a pivot from self-reliant hardware.
- Key employees left Tesla to form a new company, DensityAI, raising concerns about talent loss impacting future AI projects.
- Tesla continues with a $500 million investment in a supercomputer that won't be Dojo, indicating a shift in strategic direction.
Why it matters
- The shut down of Dojo can be related to the growing trend of companies shifting from proprietary hardware to collaborative partnerships in AI development seen in 2025.
- The move highlights increasing challenges faced by automakers in the EV sector and the evolving landscape of AI computation.
- Similar trends in AI, where data-centric approaches are being prioritized, can be observed across major tech firms investing in cloud computing capabilities.
The discontinuation of Tesla's Dojo project underscores the challenges in AI hardware development and signals a strategic shift towards collaboration, which could redefine Tesla's approach in a competitive landscape, especially as it aims for leadership in autonomous technology.