Tesla Disbands Dojo Team and Shifts to External AI Chips

Tesla Cuts Dojo Team

Tesla is disbanding its Dojo supercomputer team as part of a major strategic pivot away from in-house chip development for its full self-driving ambitions. According to Bloomberg, Dojo’s lead, Peter Bannon, has left the company, and remaining members are reassigned to other compute and data center projects.

The move follows an exodus of around 20 engineers who departed to form DensityAI, a stealth AI startup building end-to-end chips, hardware, and software for robotics, AI agents, and automotive data centers. DensityAI was founded by former Dojo head Ganesh Venkataramanan and colleagues Bill Chang and Ben Floering.

Pivot to External Partners

With Dojo shelved, Tesla plans to deepen partnerships with Nvidia for training GPUs, AMD for compute, and Samsung for manufacturing. Just last month, the automaker signed a $16.5 billion deal with Samsung to produce its AI6 inference chips, designed to scale from Tesla’s Optimus robots and FSD systems to high-performance data center workloads.

Elon Musk has briefly revived Dojo in earnings calls but now emphasizes Cortex, Tesla’s new AI training supercluster in Austin. This reflects a shift from designing custom silicon to deploying best-in-class clusters to manage Tesla’s rapidly growing video data streams for autonomy.

Industry Implications

Tesla’s Dojo breakup highlights a fundamental question for AI-driven industries: build custom hardware in-house or leverage external vendors? Much like choosing between crafting your own engine or buying a proven powertrain, teams must weigh performance gains against development costs and time to market.

Key takeaways include:

• Compute economics – balancing R&D spend against the scalability of cloud and GPU-based solutions.
• Vendor partnerships – leveraging Nvidia, AMD, and Samsung’s scale to accelerate AI projects.
• Talent and innovation – retaining top AI engineers versus startup spin-offs like DensityAI.
• Strategic focus – aligning compute strategy with core business goals and timelines.

As organizations recalibrate their AI compute roadmaps, QuarkyByte provides data-driven analysis to benchmark in-house versus outsourced architectures, optimize vendor portfolios, and forecast total cost of ownership. Our insights help automotive and tech leaders navigate strategic inflection points, accelerating decision-making while controlling risk.