1 hour ago
124
Rivian’s CEO walked straight into the conversation everyone keeps avoiding, saying the US EV market can’t grow when one company holds half of it.
RJ Scaringe made that point while showing off Rivian’s Autonomy and AI program on Dec. 11, where the company laid out its full tech stack and doubled down on its plan to build every core system in-house.
RJ founded Rivian in 2011, took it public in 2021, watched the stock jump to $120 a share, then saw it fall to $18 as of press time.
Rivian has seen gross profit in recent quarters, but net profit continues to slip away, and the company recently pulled in nearly $6 billion from a joint venture with Volkswagen.
Rivian builds the next autonomy systems using its own hardware and software
RJ said the autonomy effort started right after Rivian launched its first generation of vehicles in late 2021. In early 2022, the team realized they needed a full reset.
“We wanted to do a clean-sheet approach,” he said. So the company rebuilt the camera system, redesigned the compute hardware, and shaped the entire stack around an AI-first design.
Those choices powered Rivian’s Gen 2 vehicles, which rolled out in mid-2024 with nearly 10 times more compute than the first generation, 55 megapixels of cameras and multiple radars, creating a massive data stream that trains Rivian’s model.
The Gen 3 platform uses an in-house chip that processes 5 billion pixels per second, about five times faster than the best chips on the market today.“This lets us build the model more efficiently and more quickly,” RJ said.
Rivian already offers Universal Hands-Free driving, similar to GM’s Supercruise, and plans to expand it across more roads. In 2026, Rivian will add a “point-to-point” mode for full supervised trips. After that comes “eyes-off,” where the driver becomes a passenger.
The final step is personal Level 4, which lets the vehicle run entirely on its own, even with no one inside. RJ said the goal is to cover things like school pick-ups, airport drop-offs, and grocery runs.
He also explained why Rivian builds its own silicon instead of using chips from Nvidia. “We made the decision years and years ago to build our entire vertical software platform in-house,” he said.
The company invested hundreds of millions and hired thousands of people to build its internal systems. Rivian partnered with TSMC to manufacture the chips.
RJ said the setup gives Rivian better performance for vision-based robotics and supports a training loop that needs huge GPU power.
Rivian sets its own path while comparing its approach to Tesla
Yahoo Finance asked whether Rivian could catch Tesla’s FSD program.RJ said the goal is to be world-class and agreed that Tesla’s approach uses the right tools.
He said both companies use neural networks, end-to-end training, live reinforcement learning, and huge data streams from customer vehicles.Rivian still believes in a mix of sensors instead of only cameras.
“By including radar and lidar, it lets us turn the entire fleet into a ground-truth fleet,” he said. Every Rivian on the road sends data to help train the system.
On the industry, RJ said the loss of the EV tax credit in Q4 made things tougher. Many automakers are pulling back, which he said reduces customer choice and hurts the market. That drop in competition lets Tesla hold around 50% of the US EV segment for vehicles under $50,000. “That’s not a reflection of a healthy industry,” he said.
RJ argued that the US can’t move from 8% EV adoption to 25%, 30%, or even 100% without more than one strong option. He said Rivian’s R2 will be one of those options, but he hopes other companies step in too.
Get $50 free to trade crypto when you sign up to Bybit now
English (US)