Keep pulling the thread on RJ Scaringe.
Rivian entered into a $5.8 billion software licensing deal with Volkswagen Group to provide its network architecture and ECU topology for use across Volkswagen's brands.
Rivian's upcoming mass-market R2 model will have a starting price of $45,000.
With the exception of Tesla and Rivian, every car on the road uses a domain-based architecture with 100-150 separate Electronic Control Units (ECUs), which makes software updates and feature coordination difficult.
RJ Skirinj identifies Rivian, Tesla, and Waymo as three of the few companies with the necessary ingredients to succeed in developing autonomous driving technology.
In late 2021/early 2022, Rivian made the decision to completely reset its autonomy platform, moving away from a third-party, rules-based system to a new, clean-sheet AI-based architecture.
In an autonomous vehicle system, the onboard inference compute is an order of magnitude more expensive than the entire perception stack, which includes sensors like cameras, radar, and LiDAR.
Rivian developed its own in-house inference chip primarily as a cost-saving measure to enable the deployment of high-level autonomy capabilities in every vehicle it produces.
RJ Skirinj believes there are fewer than five, and possibly as few as two to four, companies outside of China that possess the necessary ingredients (capital, GPUs, car park, data flywheel) to successfully develop a neural net-based autonomous driving system.
Rivian's Gen 2 vehicles, which launched in mid-2024, feature a completely new hardware and software stack for autonomy with no shared code or hardware from the Gen 1 system.
The overall adoption rate of electric vehicles in the United States is approximately 8%.
RJ Skirinj believes that the current rate of EV adoption in the United States is primarily a reflection of a lack of consumer choice, not a lack of demand.
The Tesla Model 3 and Model Y combined account for approximately 50% of the EV market in the United States.