Rivian is shipping the R2 now, its first vehicle built for real volume, and in this Masters of Scale interview founder RJ Scaringe walks back through the decision that shaped the company: building its own electronics and writing its own software at a point when Rivian had almost no money and a board that told him it was a bad idea. That bet became a $5.8 billion licensing agreement with Volkswagen Group, whose first Rivian-software car, the VW ID1, lands in Europe next year as that market's cheapest EV. Scaringe also runs a robotics company he spun out, and a partnership with Uber will put 50,000 robotaxi versions of the R2 on the road from 2028. Running through all of it is a 17,000-person company that treats owning the hard parts as the strategy itself rather than a feature to brag about.
What makes the timing matter is where the R2 lands. Scaringe pegs the average new car in the United States at around $50,000, and notes that until the R2, that price band held essentially one compelling EV: the Tesla Model Y and Model 3, which together take 55 to 60 percent of the country's EV sales. He frames that not as Tesla being beatable but as a market starved for choice, and argues the car business is unusual in that the largest automaker on earth holds only about 10 percent of global demand, so there is room for many winners rather than one. The Volkswagen deal is the part worth sitting with. Volkswagen poured money into its own in-house software effort, the Cariad division, which has been widely reported to run over budget and behind schedule, before deciding to license Rivian's stack instead. The obvious worry, that the cheap ID1 could cannibalize the R2, is one Scaringe waves off: Rivian licensed the software but not its vehicle designs or self-driving system, and he genuinely wants more good EVs on the road, not fewer.
Scaringe puts hard numbers on the scale problem. He estimates a single vehicle takes around 40 million decisions and, at full size, roughly 5,500 to 6,000 engineers working in parallel, which forces the company to enable decision-making that is highly distributed yet still produces a car that feels like one mind designed it. To keep early product calls coherent, Rivian now caps the first six months of any program at no more than 50 people, a lesson learned when the full R1 team swarmed the R2 and progress stalled. He uses the glove box as a concrete example: the R1 shipped without one so the space could hold computers and a larger frunk, and enough owners objected that the R2 has two. On software, Rivian consolidated the car down to a small number of zonal computers instead of dozens of supplier-built modules. The self-driving roadmap is aggressive: supervised point-to-point later this year on Gen 2 cars and the R2, unsupervised next year, and eventually driverless operation. The robotics spinout chases industrial labor, which Scaringe pegs at about a third of global GDP.
Bottom line: The R2 is the real exam here, the moment Rivian finds out whether it can survive the jump from boutique volumes to mass production, and Scaringe says it is ramping faster and more smoothly than the R1 did. The software licensing and the robotics spinout are the hedges that make the car business survivable if the ramp gets ugly. If you are watching whether an American EV startup can become a durable carmaker, this is the company to watch, with one caveat: Scaringe is asking the market to fund three hard businesses at once, and that is a lot of conviction to keep selling year after year.