Rivian has announced a strategic shift toward vertically integrated automotive technology at its inaugural Autonomy & AI Day in Palo Alto. The company introduced proprietary silicon designed in-house and outlined a roadmap for next-generation vehicle autonomy. These developments support a new software architecture underpinned by artificial intelligence.
Highlights
- Custom Silicon: The new RAP1 processor features a 5nm architecture and 1600 sparse TOPS.
- Sensor Strategy: Future R2 models will integrate LiDAR for redundant sensing and spatial data.
- Universal Hands-Free: Hands-free assisted driving expands to 3.5 million miles of road.
- Rivian Assistant: An AI-powered voice interface launches in early 2026.
Proprietary Silicon and Processing Power
Rivian is transitioning to in-house silicon designed specifically for vision-centric physical AI. The first-generation Rivian Autonomy Processor (RAP1) utilizes a custom 5nm architecture. This design integrates processing and memory onto a single multi-chip module to improve efficiency and performance.
The RAP1 powers the company’s third-generation Autonomy Compute Module 3 (ACM3). This hardware is currently undergoing validation. It is expected to ship on R2 models starting at the end of 2026. The module features the following specifications:
- 1600 sparse INT8 TOPS (Trillion Operations Per Second)
- Processing power of 5 billion pixels per second
- RivLink low latency interconnect technology
- In-house developed AI compiler and platform software
Sensor Integration
Rivian plans to integrate LiDAR into future R2 models alongside the ACM3. This addition supports a multi-modal sensor strategy. LiDAR provides detailed three-dimensional spatial data and redundant sensing capabilities. This integration aims to improve real-time detection for complex driving scenarios.
Advanced Autonomy Software Architecture
The company detailed a software-first approach powered by the Rivian Autonomy Platform. This system utilizes an end-to-end data loop for training. Rivian introduced its Large Driving Model (LDM), a foundational autonomous model trained similarly to a Large Language Model.
The LDM uses Group-Relative Policy Optimization to derive driving strategies from large datasets. Advancements in this software are scheduled for second-generation R1 vehicles in the near term. This includes the introduction of Universal Hands-Free (UHF) capabilities.
- Available on over 3.5 million miles of North American roads
- Capable of off-highway operation on marked roads
- Autonomy+ subscription priced at $2,500 (one-time) or $49.99 (monthly)
Rivian aims to continuously improve autonomy capabilities for Gen 2 R1 and future R2 vehicles. The developmental trajectory includes point-to-point navigation and Level 4 personal autonomy.
Rivian Unified Intelligence Integration
The company is implementing AI across its business operations through Rivian Unified Intelligence (RUI). This shared data foundation helps develop new features and improve service infrastructure. It also powers predictive maintenance capabilities.
Rivian Assistant
A key component of this architecture is the Rivian Assistant. This next-generation voice interface is scheduled to launch in early 2026 on Gen 1 and Gen 2 R1 vehicles. It uses edge models to understand vehicle status and context.
- Connects vehicle systems with third-party apps
- Integrates Google Calendar as the first agentic framework application
- Utilizes frontier large language models for natural conversation
The RUI platform is designed to assist service technicians by embedding AI into diagnostics. The system scans telemetry and vehicle history to identify issues. Similar intelligence will be integrated into the mobile app to improve self-service diagnostics for owners.
Stock It To Me
Sign up for our popular daily email to catch all the latest EV news!
