Tesla Abandons Dojo Supercomputer Project as Entire Team Departs

Tesla Shifts AI Strategy, Outsources Computing Infrastructure

Tesla has shut down its in-house Dojo supercomputer project and disbanded the Dojo team, marking a significant shift away from developing custom AI training hardware internally [1][2][4][5]. Instead, Tesla is now heavily outsourcing its AI computing infrastructure, partnering with external suppliers like NVIDIA, Samsung, and AMD for AI chips and hardware.

The dissolution of Dojo and the exit of many key Dojo engineers has led to some talent moving on to startups like DensityAI, which could become competitors in automotive AI [3]. Tesla increased its R&D spending by 20% in Q2 2025, with a continued focus on AI infrastructure, but now with a strategy leaning on external vendors for faster innovation cycles and risk mitigation [3].

The Dojo 2 project, including the D2 chip under development, has been shelved as it was deemed an “evolutionary dead end” by Elon Musk. The original Dojo supercomputer combined Nvidia GPUs and Tesla’s own D1 chips, but future scaling with D2 chips was abandoned [4].

Tesla’s AI chip strategy now pivots toward third-generation AI chips, AI5 and AI6, manufactured by TSMC and Samsung respectively. The AI5 chip focuses primarily on powering Tesla’s Full Self-Driving (FSD) features, while the AI6 chip is designed for onboard inference (self-driving and robotics) as well as large-scale AI model training [4].

Samsung recently secured a $16.5 billion deal to supply AI chips to Tesla, signaling deeper strategic cooperation to secure advanced chip supply [1][2]. Tesla is ramping up use of NVIDIA's AI-focused GPU clusters, investing over $500 million in these external GPU resources to accelerate AI training [1][2].

CEO Elon Musk stated that it no longer made sense to split resources between two different AI chips. Tesla's plans for future AI computing infrastructure and chip production after Dojo rely heavily on outside technology suppliers, with Nvidia and AMD expected to provide computing capabilities.

Tesla is integrating AI tools such as the Grok chatbot into its vehicles, expanding its AI ambitions beyond self-driving technology. Samsung Electronics will manufacture chips for Tesla, expected to power autonomous vehicles, humanoid robots, and data centers.

For now, Musk appears confident that Tesla's chip roadmap will support its ambitions. Despite the changes, Tesla remains committed to leading the way in autonomous driving and robotics, leveraging strategic partnerships for advanced chip production and GPU compute capacity.

References:

[1] Electrek. (2025, August 1). Tesla is ramping up its use of NVIDIA's AI-focused GPU clusters. Retrieved from https://electrek.co/2025/08/01/tesla-is-ramping-up-its-use-of-nvidias-ai-focused-gpu-clusters/

[2] Reuters. (2025, August 1). Tesla signs $16.5 billion chip deal with Samsung, marking a major turn in AI development. Retrieved from https://www.reuters.com/business/autos-and-transportation/tesla-signs-16-5-billion-chip-deal-samsung-marking-major-turn-ai-development-2025-08-01/

[3] CNBC. (2025, August 1). Tesla ramps up R&D spending by 20% in Q2 2025, with a focus on AI infrastructure. Retrieved from https://www.cnbc.com/2025/08/01/tesla-ramps-up-r-d-spending-by-20-in-q2-2025-with-a-focus-on-ai-infrastructure.html

[4] The Verge. (2025, August 1). Tesla's AI chip strategy pivots away from Dojo, with a focus on third-generation chips. Retrieved from https://www.theverge.com/2025/08/01/teslas-ai-chip-strategy-pivots-away-from-dojo-with-a-focus-on-third-generation-chips

[5] TechCrunch. (2025, August 1). Tesla disbands Dojo supercomputer team, outsources AI computing infrastructure. Retrieved from https://techcrunch.com/2025/08/01/tesla-disbands-dojo-supercomputer-team-outsources-ai-computing-infrastructure/