Rivian founder and CEO RJ Scaringe sat down with Behind the Business for a long conversation that ranged well past cars. He walks through the company's origins, the near miss of raising money for a car startup in 2009, and the strategy now taking shape around the R2, the mass market SUV Scaringe says is about to start deliveries at roughly $45,000 and just over $57,000 at the top spec. From there the interview turns to the bigger bets: an in-house AI inference chip revealed in December, a software licensing tie up with Volkswagen, a new robotics company, and a claim that the real shift in transportation is not electric power but vehicles that drive themselves.
Rivian was founded in 2009 and showed its first two vehicles, the R1T pickup and R1S SUV, in November 2018. Scaringe says the R1 is now the best selling premium electric vehicle in the United States, though he notes that segment is small at around 90,000 units. The R2 aims at the part of the market that actually moves volume, the five passenger SUV, where the Tesla Model Y has been the default choice. That framing matters for a reason the interview only partly addresses: the in-house chip and the autonomy stack are expensive, and they only pay off at a scale Rivian has not reached yet. The company's own path to covering that spend runs through selling the technology as well as the cars. Rivian's $5.8 billion software deal with Volkswagen Group, announced last year, is the clearest sign of that supplier ambition, and it lines up with a broader industry move toward centralized, software defined vehicle electronics. Scaringe frames the choice starkly for every automaker: ignore the shift and shrink, build the technology in house, or buy it from someone who did. He is betting Rivian ends up on the selling side of that trade.
On the chip, Scaringe describes an 800 TOPS part, with two per vehicle for what he calls 6,900 sparse TOPS, capable of processing five billion pixels per second and tuned for vision based robotics. He argues the expensive part of self driving is not the cameras or sensors but the compute, which is why Rivian wanted that piece in house. He frames autonomy in levels: today's consumer systems such as Tesla FSD are Level 2, where the driver still watches the road, while a Waymo is Level 4 and can run empty. Rivian's view, he says, is that most vehicles will need Level 4 capability by the early 2030s, and that falling short would cut market share sharply. He also makes a case about vehicle architecture, contrasting the 75 to 150 small control units in a typical legacy car with the smaller number of central computers Rivian and Tesla use, which he says is what makes real over the air updates possible. On the raw materials side he flags supply risk, noting that a large share of the world's nickel comes from Indonesia, which he says makes trade policy a real exposure. He describes a separate robotics venture called Mind, aimed first at repetitive plant tasks like pick and place, plus a micromobility spin out focused on form factors smaller than a car.
Bottom line: This is a coherent plan, but it is a plan that only works if two things land on time: R2 volume and Level 4 autonomy. The R&D and vertical integration make sense at millions of units a year, and they look heavy at Rivian's current scale, which Scaringe more or less concedes. If you are watching Rivian, watch whether it becomes a technology supplier as much as a carmaker. The Volkswagen deal suggests that is the real hedge, and it may end up mattering more to the balance sheet than any single vehicle.
Commentary on a third-party video. Figures and claims are as presented in the source and have not been independently verified. Spotted an error? Tell us and we will correct it.