What happened
Rivian introduced its custom-built Rivian Autonomy Processor (RAP1), a 5nm, vision-centric AI chip integrating processing and memory on a single module, replacing Nvidia technology. This RAP1 powers the third-generation Autonomy Compute Module 3 (ACM3), delivering 1600 sparse INT8 TOPS and processing 5 billion pixels per second, utilising RivLink for interconnectivity. Rivian also launched Autonomy+, a driver-assistance package priced at $2,500 or $49.99 per month, featuring Universal Hands-Free operation across 3.5 million miles in the US and Canada. Future R2 models will integrate LiDAR sensors, with deliveries expected in the first half of next year.
Why it matters
The transition to a proprietary AI chip and the introduction of new sensor types establish a tightened dependency on Rivian's internal hardware and software ecosystem. This increases exposure for procurement and engineering teams to a single-vendor supply chain, raising due diligence requirements for long-term support, component availability, and security vulnerabilities within the custom stack. Furthermore, the integration of LiDAR into future models introduces new data acquisition and processing requirements, increasing the oversight burden for platform operators and IT security regarding data handling and integrity.
Related Articles

Nvidia Tests AI Chip Tracking
Read more about Nvidia Tests AI Chip Tracking →
Rivian Develops Bespoke AI Assistant
Read more about Rivian Develops Bespoke AI Assistant →
US AI Chip Exports Resumed
Read more about US AI Chip Exports Resumed →
China Limits Nvidia H200 Access
Read more about China Limits Nvidia H200 Access →
