Rivian CEO on CarPlay, Lidar, and Affordable EVs  ## Metadata - Author: [[Decoder with Nilay Patel]] - Full Title: Rivian CEO on CarPlay, Lidar, and Affordable EVs - Category: #podcasts - URL: https://share.snipd.com/episode/f0bd2274-717a-401b-bdb2-3fb0c2ff6791 ## Document Notes ## Highlights - **Flagship To Mass-Market Strategy** - Rivian positioned R1 as a flagship to define brand and user experience around active lifestyles. - R2 brings those brand attributes to a much lower price and smaller package without losing perceived quality. Transcript: RJ Scaringe I mean, the goal is definitely to be much broader than an expensive electric vehicle company. So when you launch a new company, it's always an important question, what do you launch first? And so we made the decision, really, even before we made the decision on products, we realized we wanted to build a brand around enabling active lifestyles. And following that, we made, it became really clear we wanted to launch with a flagship set of products. And so we launched the R1 product with, it's a sort of thing, it was like a brother, sister, sibling set of products. There's an R1T, which is a truck, and there's the R1S, which is an SUV. And those two vehicles were intended to launch the brand and really be our handshake with the world and enable us to tell a story on what we stand for, the types of experiences, the trade-offs We make in terms of attributes, let's say like on-road and off-road performance, performance and efficiency. And R2 is really just the continuation of that. So whereas R1 has an average transaction price of around $90,000, R2 starts at $45,000. It captures a lot of those same brand elements. So the vehicles, they can fit your gear, your pets, your kids your stuff you know they can go on road and off road but it's much smaller and then of course as i said much more affordable and For us it's the first product that'll take us from a flagship which is what we started with to something that is broadly accessible in terms of pricing so we're uh we're super excited About but i mean the thing about r2 that I think is so amazing is a lot of the trade-offs we've had to make to get the cost down a lot. So we've taken around half the cost out of the vehicle, a little bit more. Don't manifest in a loss of perceived, let's say, features or quality. It's the execution in the interior is just beautiful. The way the car rides and drives is amazing. It's obviously smaller, but it really feels like a Rivian and it's really enjoyable to drive. And I think package-wise for many users, if you don't need a seven row, it's sort of like just the right size. It's a little bit shorter than a Model Y, but feels a lot bigger. And I love the way, you know, all the little details, the front trunk works great. The rear lift gate's beautiful. ([Time 0:05:33](https://share.snipd.com/snip/acda2fab-d092-452f-a351-7aba53570987)) - **Plan Sourcing For Geopolitical Risk** - Build supply chains aligned to expected geopolitical shifts to reduce tariff risk and cost surprises. - Localize manufacturing content when possible to align with likely future policy and reduce exposure. Transcript: RJ Scaringe With R2, we've actually had the benefit of making a lot of sourcing decisions, recognizing some of the, I characterize them as global geopolitical shifts. And one of the biggest global changes in mindset that's not just unique to the United States, but we're seeing across the Western world, is an increasingly domestic-centric approach To manufacturing, meaning in the case of the United States, and I'd say broadly in the case of Europe, there's a desire to retain more manufacturing content local to the markets in which The vehicles are sold. And so in our case, because we're producing all of our products in the United States, we knew that going into the sourcing of our twos. So we had built a U.-centric supply chain around that. Of course, there's some things that we don't make in the United States, certain metals, especially heavier earth metals, certain materials just globally don't really aren't born Out of the United States. Things like nickels are a very good example. 92% of the world's nickel supply comes out of Indonesia. And so in certain aspects of the vehicle, we've had to build cross-border trade relationships. And some of the changes we're seeing in trade policy do add cost, but we've had the benefit of being able to be quite planful around that. ([Time 0:22:10](https://share.snipd.com/snip/6fef7f38-d7a0-4b3b-a3d3-8b577115d0c3)) - **U.S. Tech Exports Strengthen Position** - Depoliticizing EVs and emphasizing domestic tech and manufacturing can align a company with broad political goals. - Rivian sells technology abroad, using U.S. software expertise as an exportable asset. Transcript: RJ Scaringe And we think of this just not for me, but for the whole leadership team, looking at all the changes in the system and trying to understand what are the things that are going to be persistent? Like, what are the changes that are going to exist two years from now, four years from now, five years from now? And what are the things that are somewhat in the noise? And when we look at it through that lens, there's a number of really clear things. So one thing that's very clear, regardless of whether you're looking at it from a Republican point of view or a Democratic point of view, building independence from China is going to Be a theme that's important in the United States. And so that was really important for us as we sourced R2. We built very little dependency or as little dependency as possible on Chinese supply chain. That carries with it assumptions around costs and different cost structures. But that was one key takeaway. And so in that way, we've aligned the business to the intent of, I'd say, both all political parties in the United States. The other is that there's a big push for U.S. Capability both technology and U.S. Capability in manufacturing. And so putting aside some of the, I'd say, noise and short-term changes that have impacts, real impacts on us around electric vehicles and some of the policy for electric vehicles, We find ourselves very aligned in terms of building deep technology expertise in the United States. That technology expertise that's so strong that, in fact, we're selling that to other companies outside the United States. So, you know, this is our deal with Volkswagen is one of the largest software licensing deals in the history of the auto industry. So, which went a U.S. Company and a European company. ([Time 0:24:12](https://share.snipd.com/snip/f56a2c18-9ac0-4732-b870-c4f580b6ff51)) - **Ramp Challenges And Supply Shocks** - Rivian's early ramp hit pandemic and parts shortages, then achieved positive gross margin after Gen 2 updates. - A China export control on magnets caused a Q2 production drop, but the restriction later lifted and production recovered. Transcript: RJ Scaringe Better improve. And I say that with a smile. So if we look at our trajectory, when we launched, we had a whole host of challenges that were not predicted. We launched right into the pandemic. It's probably hard to imagine a more challenging environment to launch an industrial business into. So it's hard to build a plant and launch a plant with remote work and work from home. We then had a huge supply chain crisis and given us as a new manufacturer, we had real challenge that persisted well into 2023 around just our ability to get enough parts. And so we were starved for parts, which manufacturing really benefits from predictability and being steady. And our on-off nature just created a huge amount of unexpected fixed cost overhead. And then as we came into 2024, things started to really come together. We launched some updates to the vehicle. We call our Gen 2 version of R1. And so Q4 of 2024, we were positive gross margin. Q1 of 2025, we had over $200 million of positive gross margin, which was awesome. And then this past quarter was a tough quarter for us. Q2 of 2025, we had a big drop-off in volume, largely driven by some of the challenges around getting enough heavy rare earth metals to make our motors. So in our case, of our vehicles are, of course, electric. We produce all of our vehicles in the United States. China, as part of the trade negotiations that have been happening, put in place an export control where they didn't allow export of magnets into the United States. And so we just couldn't make motors, because motors needed the magnets. And so we had this, we went from 14,000 units produced in Q1 to around 6,000 in Q2. And that was a singular sort of phase shift, if you will, that we felt we saw in the numbers. But we built solutions around that, that export control has been lifted. And so we're quite bullish on the rest of this year is going to be, we'll produce and sell more vehicles in the second half of the year than we did in the first half. And then, as you said, the launch of R2 really ignites the business. It allows us to get to a level of volume that helps cover our operating fixed costs. It's like our plant fixed costs. ([Time 0:28:54](https://share.snipd.com/snip/f2a5ccfd-e90a-42f9-a04b-8fadcbe644f1)) - **Kid Test Drives And Key Card Reaction** - Joanna's four-year-old asked why a Rivian had a flat tire and loved the key card, showing real household engagement. - RJ explained phone-as-key is the intended primary access and described kids liking R2's lower height and rear glass drop. Transcript: Joanna Stern I drove it many miles, actually. It was almost like, I think, 500 miles in your car. So thank you for that. But I had a pretty tough reviewer in the car for this whole time. And we have a quick question from him. I'm going to play it. Speaker 3 Get a flat tire Rivian Man. Joanna Stern And what else do you like about the car? Speaker 3 That we have the car for it and it opens. That's cool. Joanna Stern So did you hear the beginning of the question? RJ Scaringe I couldn't quite hear the beginning. Joanna Stern It's why did we get a flat tire Rivian Man? RJ Scaringe I heard the Rivian Man, which I really like that. Joanna Stern Yeah, yeah, that's your name. I told him I was going to talk to Rivian Man. And so that's what he said. He wanted to know, why did we get a flat tire, Rivian Man? RJ Scaringe I actually don't know. Did you run over a nail or something? I don't know either. Joanna Stern I don't know. It's just like, anytime now he sees a Rivian, he's like, that's a Rivian. It has a flat tire. And I had to explain to him that actually all cars can- Get flat tires, yeah. And then the second part of his question was it wasn't really a question. He just wanted you to know that he loves the key card. He says that's really cool. And I wanted to actually tell you I don't like the key card. RJ Scaringe The key card, actually, it's not supposed to be used other than like sort of an emergency. You put it in your wallet or you put it somewhere and hold on to it. But the idea is your phone's really key. So I don't know if the team, when you loaned it, set you up with phone as key, but that's the idea. Joanna Stern It's true. I did have it set up on my phone for a little bit after we had the key card. Oh, okay. Any other things you have to say to my four-year son in terms of why he should tell me to get this car? RJ Scaringe Oh, why? So I have a six, a seven, and a nine-year I have a perspective of what the kids really enjoy with it. So I'd say in the case of R2, because my boys have spent a lot of time now on R2, a lot of the fun little features we have on R1, like the flashlight and the door are still there. There's tons of storage. The front trunk's great in R2, but we put in a few more features. So the rear glass drops completely in R2. It makes for like a really nice open air experience where you can drive with the back open and it's not too noisy. You can fit stuff into it. If you have pets, they really like it, but the car has all the same performance like trade-offs. It's still very capable on-road and off-road. It's a great adventure mobile. It fits all your gear. So I think he'd really enjoy the R2, but it's also easier to get into for kids because it sits a little lower than the R1. ([Time 0:31:58](https://share.snipd.com/snip/be613556-9565-42d4-8672-b09ffd95feca)) - **Maps Switched To Google** - Rivian switched its mapping to Google for a significant usability improvement. - The car's architecture allowed a seamless maps update without changing the underlying platform vendor. Transcript: RJ Scaringe Yep. We launched, we had a great partnership with Google. So the maps are now in partnership with Google. And it was an area that there was a lot of requests for a different solution in our mapping. So the beauty of our architecture is we can update all sorts of things, including map applications. So we have a whole new mapping environment, which is beautiful. It's just, it's a really nice, I mean, it's just great working with the Google team to develop this. It's really, it's really special. It's a wonderful application. Joanna Stern And did you work with the Google team on sort of the underlying infotainment system that's based on Android, right? RJ Scaringe It's a good question. We didn't work with Google on the platform. We built it to really work seamlessly with the Android ecosystem so that some of the apps that we integrate can be more seamlessly pulled in. But we did that entirely without Google. And in the case of the Maps, we did work very closely with Google. ([Time 0:34:58](https://share.snipd.com/snip/3f6ca7a6-f1ef-43e2-8da4-e59f25261af6)) - **Own The In-Car Experience** - Don't rely on third-party screens like CarPlay when integrating deep vehicle context and future AI features. - Own the UI glue so vehicle state and AI can coordinate apps and features seamlessly. Transcript: RJ Scaringe That's a funny way to ask it. No, we definitely don't hate CarPlay. We've taken the decision, which I'm very confident in the fullness of time customers will appreciate, which is we wanted to have a seamless digital experience and not have the need To jump between, let's say, CarPlay, which feels like it feels obviously like CarPlay. So it feels the same in every car, that environment, and then what we create as a Rivian environment. And rather, think of it more a la carte, create all the same applications. So you have a YouTube you can go to, you can go to Spotify, you can go to Google Maps, you can go to Apple Music, you can go to everything. You can have all those integrations. But for us to hold the glue for putting that all together, this becomes even more important as we start to integrate AI into the vehicle. And so over the next 18 months, we're going to see a lot of new features that by necessity are performing tasks or doing decisioning to connect different applications. And so like knowledge of what's the vehicle state, knowledge of, you know, is it in drive? Is it parked? What's the conditions outside the vehicle? What's your driving history? What's your preferences are? Knowledge of all of that at an ecosystem level then allows us to present a richer, better experience for you as a driver or occupant of the vehicle. And so I think it'd be really hard to do that if we had to put all that through an application which expands to take over the screen and provides you, you know, essentially with a set of bubbles That feel very much like a, you know, like a CarPlay experience. Piece by piece, everything that someone may have missed from their CarPlay experience, whether it was a mapping, soon we're going to have voice to text, is going to be there and it'll Be beautiful. We're really convicted on this. For some folks, that means they're not going to buy a Rivian. We accept that. It's a decision. I say this all the time. ([Time 0:40:00](https://share.snipd.com/snip/a565607a-2265-4ba8-aa67-8ecb1579bb78)) - **Foundation Models For Driving** - Rivian shifted from rules-based planners to large parameter driving models trained on in-house perception data. - This foundation-model approach accelerates capability growth and makes sensor upgrades additive, not brittle. Transcript: RJ Scaringe Boy, we could spend the rest of the time on this. This is a big topic. But first, it's important to recognize there's been a big shift. So the way autonomous systems were developed up until around, you could say, 2021, 2022, maybe even, is you had a perception platform or stack of perception, which may be cameras and Radars, maybe in some cases, cameras, radars, and lidars, that would see the world or perceive the world, would identify all the objects, it would classify those objects, it would Assign vectors to those objects, so you know how they're moving in the world, and then all those objects would be handed to a planner. And the planner would be a rules-based environment that's essentially a programmed version of interpreting how the world works and then making a whole series of decisions based upon This defined set of rules. And that environment has its obvious limitations. It's very sensitive to changes in the sensor set. It's very sensitive to the location or the application. And so it sort of reached an asymptote of capability. And while that was being developed, a whole new approach to thinking about how to develop self-driving systems emerged, which was to really to use AI, which interestingly, AI as we Know it today, was not really part of the first gen of self-driving systems. It was a very rules-based environment. And so what's now happening and what's happening in our Gen 2 vehicles, which is why there's a significant difference between our Gen 1 and our Gen 2 vehicles, is our Gen 2 vehicles, which We launched in the middle of 2024, is we brought all of our perception stack in-house. So all the sensors we designed ourselves, there's no third-party sensors there. We then built a much more powerful compute platform, so an inference platform in the vehicle. And we're taking all this data and we trigger it based upon a whole variety of things, which we can talk about in a moment, but like interesting events. And we use that to feed the training of a large parameter model. Think of it like a foundation model for driving. So a large, you know, it's ultimately a multi-billion parameter model. And that model's a neural net of how to drive a vehicle. And the beauty of this is as you improve sensors or as you change sensors or as you change vehicle applications, this neural net of understanding of how to drive a vehicle doesn't go away. So a simple analogy is if I wear glasses, if I had learned to drive without glasses and you suddenly handed me my glasses, I would not forget everything I learned. I would just drive better with more accuracy because I have better perception. And so the same is true in this approach now, which is building a large model. You've maybe heard it or people have heard it called an end-to approach where you take this approach of having data come in, you use that to train the model, and then that model determines How the vehicle's operating. And so why this is so exciting is the rate of improvement of these systems is going to be much higher. That's the platform we now have in our vehicles. And we're going to begin to start to see the feature set expand. And so today, as you said, we have a, we call it a highway assist. It's a highway feature where you can be on the highway, you take your hands off the wheel, your eyes stay mostly on the road, but otherwise the vehicle is driving itself. The next extension of that is hands off the wheel, eyes on the road everywhere. So on any type of road, the next extension of that is hands off the wheel, eyes on the road with turn by turn. So the vehicle can navigate from an address to another address and that's coming next year as well. And then the next step is hands off wheel, eyes off road in certain circumstances. Highway is a really good example where you're on the highway, your hands are off the road or off the wheel. You want your eyes off the road and you can be on your phone reading a book. That set of features for us is going to be coming in 2027. And that I think is the really big unlock as you start to get your time back. So it's not just reducing the cognitive load on you to have to operate the vehicle and be fully aware, but it also gives you the ability to do other things. And so that's what I just described is all happening over the next 18 to 24 months. ([Time 0:43:55](https://share.snipd.com/snip/fc28c9e9-bd17-47f1-bba1-60c952154d72)) - **Sensor Diversity Improves Perception** - Multiple sensor modalities improve model training and world understanding, making perception more robust. - Rivian uses high-megapixel cameras, corner radars, and 3D imaging radar and may include LiDAR if beneficial. Transcript: Joanna Stern Me real quick, though, what are the sensors? I mean, there's this big debate I feel happening right now, or maybe it's not really a debate, and there's maybe one side on the debate, and no one's really debating it, that there is the LiDAR versus cameras. And Tesla's going all in on cameras saying we don't need LiDAR. What about Rivian? Does Rivian add LiDAR? RJ Scaringe The view of sort of the entirety of the science community believes that having multiple sensors is helpful because you build a more accurate view of the world. And the way that we build these neural nets is you want a broad understanding of the world and you want the highest accuracy. And you're going to have lots, if you have more than one camera, you're going to have multiple cameras that have different signals, which have different signal to noise ratios that Need to be managed. But ultimately the way that that information is fused very early, if you have multiple cameras coupled with radar, coupled with potentially LIDAR, as you said, it gives you a more fulsome And accurate picture. It also allows you to train your model better. So it's analogous to if I had to learn the world with one eye, I would learn a less accurate version of the world than if I learned the world with two eyes. And if you look at the evolutionary tracks of many species of animals, most animals have multiple modalities of sensing. And the ones that have to operate in maybe the most extreme environments, let's say extreme darkness, generally combine some optical perception with some wavelength-based perception, Often like, you know, sound waves or sonar. Bats are an example of this. Our view is it's definitely beneficial. And our approach to sensors has been we need to rapidly build our foundation model as fast as possible to test us a lot of vehicles. And it's made great progress. They have an amazing product. So we have more megapixels of cameras. We have 55 megapixels of cameras in R1. That'll jump to 65 megapixels in R2. We have a really robust set of corner radars and a really beautiful 3D imaging radar in the front. And that's rapidly building a robust foundation model that we're going to start to see these features I just described start to deploy with. ([Time 0:48:42](https://share.snipd.com/snip/df5f9fff-d32e-449e-820b-43697e57dec5)) - **Fix Reliability Through Production Learning** - Prioritize production and quality learnings from early builds to improve long-term reliability metrics. - Use R2's design and manufacturing improvements to raise vehicle reliability and customer satisfaction. Transcript: RJ Scaringe Something we're absolutely focused on. The way, of course, this looks at some of our early builds, our vehicle reliability is now getting much, much better. But even with that, even with the fact that there was more service requests on our early builds than, let's say, other vehicles, we had the number one level of customer satisfaction. So I always say to our team, like when we look at that data, imagine if our reliability gets to best in class, coupled with all the product attributes. And so that's our goal. And R2, you know, R1 is continually improved since we launched. And then R2 is a further step change from that with all the learnings that we've made around our production process, production quality, reliability. And it is tough for us when we first launched in our first year, we came out number one of customer satisfaction, but it was our first time building vehicles at scale. So not surprisingly, we had more challenges with just ramping quality systems to begin with. ([Time 0:52:03](https://share.snipd.com/snip/0b1c3ea2-6d19-426c-96cc-697b8e77bba0))