Rivian advances AI autonomy with proprietary silicon
Rivian has taken the wraps off its next phase of vehicle intelligence, unveiling custom silicon, a new autonomy compute platform and deeper AI integration at its first Autonomy & AI Day in Palo Alto. The announcements signal a clear push toward vertically integrated hardware and software as the EV maker chases higher levels of automated driving.
This is significant because it shows how an automotive OEM is moving to design its own advanced processors, AI toolchains and safety-certified compute platforms — reflecting broader shifts already underway across the semiconductor and embedded systems industries.
Custom silicon at the heart of autonomy
At the core of Rivian’s roadmap is a move to in-house silicon tailored for vision-centric physical AI. The company introduced the Rivian Autonomy Processor (RAP1), a custom 5nm device that integrates processing and memory in a multi-chip module designed to meet automotive safety requirements.
RAP1 delivers 1600 sparse INT8 TOPS and can process up to five billion pixels per second. It also includes RivLink, a low-latency interconnect that allows multiple chips to be linked to scale performance. An in-house AI compiler and platform software complete the stack, underscoring Rivian’s intent to control everything from silicon to software.
The processor powers Rivian’s third-generation autonomy computer, the Autonomy Compute Module 3 (ACM3). According to the company, the Gen 3 autonomy hardware — ACM3 combined with LiDAR — will ship in R2 models starting at the end of 2026 and is currently undergoing validation.
“I couldn’t be more excited for the work our teams are driving in autonomy and AI. Our updated hardware platform, which includes our in-house 1600 sparse TOPS inference chip, will enable us to achieve dramatic progress in self-driving to ultimately deliver on our goal of delivering L4,” said Rivian founder and CEO RJ Scaringe.
Software-first autonomy and new revenue models
On the software side, Rivian detailed its Autonomy Platform, built around an end-to-end data loop and a Large Driving Model (LDM). Trained in a similar way to large language models, the LDM uses Group-Relative Policy Optimization to distill driving strategies from massive datasets into on-vehicle intelligence.
Near-term gains will reach existing vehicles. Second-generation R1 models will get Universal Hands-Free driving, covering more than 3.5 million miles of roads in the US and Canada, including some off-highway scenarios with clear lane markings.
Rivian also announced Autonomy+, a paid autonomy subscription launching in early 2026, priced at $2,500 one-time or $49.99 per month — signaling autonomy as both a technology differentiator and a business driver.
Unified AI beyond driving
Beyond autonomy, Rivian is rolling out Rivian Unified Intelligence, a shared AI foundation spanning vehicles, service and apps. The first visible result will be the Rivian Assistant, a next-generation voice interface launching in early 2026 on Gen 1 and Gen 2 R1 vehicles, with Google Calendar as the first third-party integration.
By embedding AI into diagnostics and service workflows, Rivian is betting that tight vertical integration — from custom silicon to cloud-connected assistants — can accelerate innovation and reshape the vehicle ownership experience.
If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :
