The headlines sound dramatic: Tesla dissolves its Dojo supercomputer team, project leader Peter Bannon leaves the company, and a central AI project is terminated. At first glance, this seems like the end for a project that analysts once believed could add up to $500 billion in value for Tesla.
But a closer look – including comments from Elon Musk – reveals: This is not about failure, but about a realignment that was long part of the plan.
What has been reported so far
According to reports, the Dojo team is being dissolved, Bannon is leaving Tesla, and remaining employees are being distributed to other data center and compute projects. Additionally, Tesla will increasingly rely on external partners like Nvidia in the future.
Elon Musk: Convergence instead of termination
Elon Musk confirmed the changes but presented them as a strategic focus. Background: Tesla previously developed two separate AI chips – one for training (Dojo Gen 1) and one for inference in the vehicle (AI6). This dual structure is now to be replaced by a unified architecture.
AI6 as the next Dojo generation
The upcoming AI6 chip design integrates the Dojo architecture directly into the vehicle. This will make it possible to perform training and execution of Full Self-Driving (FSD) on the same chip basis in the future – shortening development cycles and increasing efficiency.
Important: Tesla is currently still bringing the Dojo V2 into mass production to bridge the transition period until the AI6 launch. Dojo will also remain as a supercomputer in the form of a cluster of AI6 chips.
No end, but restructuring
The “end” of the Dojo team mainly means the reallocation of personnel from the completed Dojo V1/V2 project to the new, unified AI6/Dojo V3 approach. Bannon's departure coincides with this strategic shift.
Tesla is not saying goodbye to custom AI chips – on the contrary: By converging training and inference on one platform, development is intended to become simpler, faster, and more focused.
Conclusion
The era of Dojo as an independent supercomputer may be over – but its technology, architecture, and developers are decisively shaping Tesla's AI future. AI6 will be the centerpiece of a new, integrated chip strategy that provides computing power for both training and vehicle deployment.