Tesla Abandons In-House Chip Development, Shifts to External Partnerships

Tesla is shifting its strategy and disbanding its Dojo supercomputer team, with leader Peter Bannon leaving the company. Around 20 team members have joined DensityAI, while the remaining staff will be reassigned to other Tesla data center and computing projects. Tesla is now relying more on external partners like NVIDIA, AMD, and Samsung for computing solutions and chip manufacturing.

Tesla Inc. has announced a significant strategic shift in its artificial intelligence (AI) development efforts by disbanding its Dojo supercomputer team. The decision follows the departure of Peter Bannon, Tesla's chip and Dojo supercomputer chief, after nearly a decade with the company. The Dojo project, which aimed to build Tesla's own AI training supercomputer, has faced persistent technical delays and talent attrition [1][2].

The disbandment of the Dojo team marks a major strategic pivot for Tesla. The company has reassigned engineers to other computing and data center initiatives within the organization. Around 20 team members have joined DensityAI, a new AI startup co-founded by former Dojo head Ganesh Venkataramanan and ex-Tesla employees Bill Chang and Ben Floering [3]. The remaining staff will be reassigned to other projects, indicating a broader shift in focus.

Tesla is now increasingly relying on external partners for its AI compute needs. The company has entered into partnerships with Samsung and NVIDIA for AI hardware, abandoning its previous strategy of vertical integration in favor of outsourcing compute needs [1][2]. This move reflects broader challenges in retaining top talent and maintaining momentum in large-scale AI infrastructure development [2].

The decision to disband the Dojo team and shift to external partners is seen as a pragmatic response to the complexities of in-house supercomputing. While Tesla has made advances in AI inference computing, such as the AI5 and AI6 chips, it has struggled to maintain progress in large-scale AI training infrastructure [1][2]. Outsourcing production to Samsung and TSMC may allow Tesla to access cutting-edge hardware without the high costs and risks associated with building proprietary supercomputers [1][2].

This strategic shift also aligns with Tesla’s recent expansion of its Robotaxi services in Austin and San Francisco. Although these services still rely on human oversight, the company is accelerating the development of autonomous systems through external compute resources. This move may reflect a broader recalibration of priorities, with Tesla focusing on leveraging industry leaders in chip manufacturing rather than pursuing full self-sufficiency in AI hardware [1][2].

The long-term impact of the Dojo team’s dissolution remains to be seen. While some analysts had previously forecast ambitious growth tied to the success of Dojo, the project’s cancellation could affect those expectations. However, others argue that the move to external partners may allow Tesla to concentrate its resources more effectively and maintain competitiveness in a rapidly evolving market [2].

References:
[1] Reuters, Tesla shuts down Dojo supercomputer team, reassigns workers amid strategic AI shift, (https://www.reuters.com/business/autos-transportation/tesla-shuts-down-dojo-supercomputer-team-reassigns-workers-amid-strategic-ai-2025-08-07/)
[2] Electrek, Tesla Dojo supercomputer looks dead as more execs leave for competing startup, (https://electrek.co/2025/08/07/tesla-dojo-supercomputer-looks-dead-as-more-execs-leave-competing-startup/)
[3] AInvest, Tesla Halts Dojo Chip Project, Shifts to External Partners, (https://www.ainvest.com/news/tesla-halts-dojo-chip-project-shifts-external-partners-2508/)