RJ opened the first ever Autonomy and AI Day explaining why Rivian believes it is positioned to lead in this next phase of the industry. The company is leaning hard into compute, custom hardware, large scale AI systems, and a shared data foundation that touches every part of the ownership experience. Let鈥檚 break it all down. Meet the Rivian Autonomy Processor One of the biggest announcements was RAP1, Rivian鈥檚 first in house processor built on a 5nm multi chip module. It delivers 1600 sparse INT8 TOPS and can push 5 billion pixels per second inside the new Gen 3 Autonomy Computer. Rivian even built its own AI compiler and platform software to support it. This shows Rivian is no longer just integrating off the shelf chips, it is now designing silicon specifically for its autonomy roadmap. Click to enlarge Click to enlarge Autonomy Computer and LiDAR on R2 The ACM3 (Autonomy Compute Module 3) autonomy computer will debut on R2 starting at the end of 2026, but Rivian made it clear that R2 will launch initially without LiDAR. What Rivian confirmed today is that LiDAR will be added later in the program. This lines up with what we explored back in May when we spotted early signs that Rivian was evaluating LiDAR as a redundancy and ground truth layer for future autonomy. Rivian has now officially validated that LiDAR is coming to R2 down the road, where it will join cameras and radar to create a richer, more resilient perception stack. Click to enlarge Click to enlarge Large Driving Model and Rivian鈥檚 Data Loop Rivian explained how its autonomy stack is powered by a self improving data loop feeding the company鈥檚 Large Driving Model, which is trained similarly to an LLM. Reinforcement learning distills high quality driving behavior into efficient onboard models. Every release improves the system, and Rivian laid out a trajectory that moves toward point to point, eyes off and eventually personal Level 4. Universal Hands Free Coming to Gen 2 Rivian confirmed that a major soft...
First seen: 2025-12-11 18:38
Last seen: 2025-12-12 18:47