Rivian CEO RJ Scaringe on R2 launch day: level 4 autonomy by 2028, 200 build combos vs hundreds of thousands on R1
Jun 9, 2026 with RJ Scaringe
Key Points
- Rivian launches R2 today with 200 build combinations versus hundreds of thousands on R1, a deliberate correction after COVID-era supply chain chaos nearly broke the company.
- Rivian's consolidated software architecture—consolidating legacy automakers' 100-150 control units into zonal controllers—underpins a $5.8 billion licensing deal with Volkswagen Group.
- Scaringe targets Level 4 autonomy by 2028, betting that foundation model approaches will deliver capability legacy manufacturers cannot match within a one-to-two year window.
Summary
Read full transcript →Rivian R2 launch day
RJ Scaringe is launching the R2 today — Rivian's first mass-market vehicle, priced well below the R1 line where average selling prices exceed $90,000. The R2 has been part of the plan since the company's earliest days, conceived as the product that would finally bring Rivian to scale. First deliveries to external customers begin today, after an internal employee rollout.
200 build combinations vs. hundreds of thousands
The R1 launch nearly broke the company. Scaringe says launching the R1T truck, R1S SUV, and Amazon commercial van within three months of each other — during COVID — was the single biggest thing he would do differently. The supply chain complexity was compounding: the R1 supported hundreds of thousands of possible build combinations, each requiring supported configurations across the production system.
The R2 is a deliberate correction. Fewer than 200 build combinations total. A narrower color selection at launch. A launch edition trim only. The R3 platform, which Rivian has publicly previewed, was specifically held back to avoid repeating the same mistake — Scaringe says R2 needs to ramp and stabilize before R3 is introduced.
On range, the 330-mile figure on the all-wheel-drive version is an output rather than a target. The design constraint was exceeding 300 miles on every configuration. The off-road tire variant lands at 307; the base rear-wheel-drive version reaches 345.
“The path to level four in 2028... The rate of progress we're making on our own platform is materially different than what we've been doing for, you know, the last two years... r two has less than 200 build combinations. Wow. And just to put that into perspective on r one when we launched, we had hundreds of thousands of different possible build combinations.”
Software architecture and the Volkswagen deal
Scaringe's explanation of why legacy automakers struggle with over-the-air updates is worth understanding. Traditional vehicles have accumulated 100 to 150 separate electronic control units over decades, each written by different suppliers, often third-party contractors, with no unified OS. Changing something as simple as the sequence of events when a driver approaches the car requires coordinating changes across 10 to 15 companies.
Rivian consolidated that compute onto a small number of zonal controllers running a single OS it built internally. That architecture is what underpins the $5.8 billion software licensing deal with Volkswagen Group, through which VW is paying to use Rivian's platform for its own future EV products.
The R2 carries significantly more infotainment compute than the R1, enough to run larger on-device AI models without routing to the cloud — reducing latency for in-car AI assistants. Scaringe frames R2 as a platform with headroom for features that don't yet exist.
Level 4 autonomy by 2028
Scaringe's autonomy roadmap is specific. Point-to-point hands-free navigation launches later this year — driver in the seat, hands off wheel, eyes on road. Level 3 (hands off, eyes off, but driver available for takeover) follows in 2026. Level 4, where the vehicle drives empty with no human required to intervene, is the 2028 target.
His broader argument is that the rules-based, C++-coded autonomy approaches that dominated from roughly 2013 to 2022 have been fully superseded by foundation model approaches. Multi-billion-parameter neural nets trained on millions of real-world miles produce qualitatively different results, and Scaringe says progress from 2026 to 2031 will bear no resemblance to the prior five years. His view is that by 2030 or 2031, buyers will expect Level 4 capability as a baseline — and that the inflection will come when drivers first experience genuine eyes-off time return, rather than the current FSD model that still requires attention.
The competitive risk for legacy manufacturers is timing. Even a one-to-two year lag behind Rivian and Tesla on meaningful autonomy levels may prove harder to recover from than any gap in interior quality or styling — particularly as early adopters who've experienced Level 3 or above become unwilling to consider vehicles without it.
Every deal, every interview. 5 minutes.
TBPN Digest delivers summaries of the latest fundraises, interviews and tech news from TBPN, every weekday.