← Cheeky Pint
Loading summary
Transcript228 lines
- [00:00]
Lex Fridman
Cars used to have cool names, you know, you have Mustang, Thunderbird, now we have R1, R2.
- [00:07]
RJ Scaringe
The code name for Gen 2 is Peregrine.
- [00:08]
Lex Fridman
Okay.
- [00:09]
RJ Scaringe
Which is beautiful. Super fast Following Rivian.
- [00:11]
Lex Fridman
Would anyone here buy Rivian Peregrine? Yeah, I've seen a lot of takers.
- [00:14]
RJ Scaringe
With the exception of two companies, Rivian and Tesla, every car on the road has anywhere from 50 to maybe 150 little computers that run the car.
- [00:21]
Lex Fridman
This is why you in a normal car cannot adjust, say the side mirrors from the center console.
- [00:27]
RJ Scaringe
It's sort of three different companies. In 2013, 2014, I would go meet with suppliers and CEOs of very large suppliers. I'd say we're working on an electric SUV and electric truck as a flagship product. To say laughed out of the room is probably accurate. So you know how to pour these things and everything.
- [00:46]
Lex Fridman
These are actually really easy. This is good. This is a new now you gotta go all the way.
- [00:49]
RJ Scaringe
All the way.
- [00:50]
Lex Fridman
It's perfectly sized to fit.
- [00:51]
RJ Scaringe
That is impressively sized.
- [00:53]
Lex Fridman
RJ Scarringe is the founder and CEO of Rivian, makers of electric trucks and SUVs. You've probably started to see Rivian's out and about on the road. It's a delightful product. And how they built an independent new car company is a really interesting story. Welcome. Cheers.
- [01:07]
RJ Scaringe
Yeah.
- [01:09]
Lex Fridman
The Rivian story makes no sense to me. It shouldn't have worked because, you know, you grew up passionate about cars, nature enthusiast, you did a mechanical engineering PhD. And Rivian was not only the first company you started, it was your first job or I mean, your first, you know, career. Like hardware companies are startups on hard mode, auto companies are hardware companies on hard mode. So everything about this just seems like it shouldn't have worked out.
- [01:40]
RJ Scaringe
Yeah, it's really hard.
- [01:42]
Lex Fridman
Explain yourself.
- [01:43]
RJ Scaringe
Yeah, I mean, one of the things that makes starting a car company so difficult is you need a lot of hard to obtain things all at the same time. So you need billions of dollars of capital, you need a team of thousands of engineers, you need hundreds of suppliers that would want to work with you and provide components and parts manufacturing facility that's enormous and takes years to build. And if you had all those ingredients at the start, it would take maybe four or five years to launch the first product. But in the beginning you have none of those ingredients. And so it really is a challenging exercise. And for me, I think the part that was probably harder than I expected was just the path to go from. We started with no capital, so to go from no capital, it's raising a very small amount of capital, enough to prove out like some semblance of an idea, but then to start to raise more and more capital to scale it. But it was a low probability of success.
- [02:34]
Lex Fridman
So what were the moments where, you know, if you're following the prediction market, if you're looking on polymarkers, will, you know, will Rivian succeed? What are the moments where the probability would have spiked upwards at Some point?
- [02:44]
RJ Scaringe
Probably 2019. 2019, Amazon invested in the company, so that would be a big spike in probability.
- [02:50]
Lex Fridman
How much do they invest?
- [02:51]
RJ Scaringe
Well, in total, they invested several billion dollars before it went public. Today they're our largest shareholder, but in addition to them. So they invested, I think around 2 billion prior to our IPO, roughly because it invested in a few rounds. But the round they Invested was a $700 million round. It was a big first round of a number of strategic and really well known investors came in at that point. And that was after we'd shown the product at the LA Auto show. And prior to that we'd raised a few hundred million dollars, but we still needed quite a bit more money to launch the company. But importantly, beyond just them investing, we also put a commercial deal together where we build commercial vans for them. And so you now see them all over the place. These are the big electric vans that we build with Amazon as a lead customer. But there's other non Amazon customers.
- [03:34]
Lex Fridman
Why do they care about electric vans? Was it their climate commitment? Was it something to do with. Operationally it works better. Like what do they want?
- [03:40]
RJ Scaringe
There's a true cost advantage to running electric vans for delivery. So the when you think about a delivery van, it spends a lot of its time sitting still. And so not having an engine idle, not having the same service frequency and all the engine maintenance and repairs, oil changes, brake replacements, everything just goes away. So there's a real cost advance to running EVs, but there weren't any great electric van products in the market and certainly nothing that was designed with robustness or with their use case in mind.
- [04:08]
Lex Fridman
Okay. So they thought for their logistics operation, having a proprietary source of electric vehicles would lead to a long term operating cost advantage.
- [04:15]
RJ Scaringe
Yeah, I mean it was wild. From the early meetings with Amazon, we sketched an idea on the whiteboard and that was in 2019, early part of 2019, we were sketching the ideas for the vehicle and by end of 2019, we'd signed a deal to supply commercial vans. And then roughly exactly two years later, we launched the product. So it was an extremely Fast program to go from like whiteboard to like vans delivering packages within about two years.
- [04:45]
Lex Fridman
Yeah, yeah, yeah. And where are those, like where is production those vans now in terms of volumes or kind of how you work.
- [04:51]
RJ Scaringe
With Amazon, it's about 10,000 or so a year. It's going to be growing. So we're bullish on the van market in the long term. I think in the short term there's been a lot of things we didn't plan for. We had Covid supply chain eruption. We've seen a number of companies just be much more capital constrained in how they deployed assets to invest in, let's say electrification or electrification infrastructure. And so the whole commercial van space has moved slower than we thought. But the economics are so advantaged that we're really clear on it long term being a big space.
- [05:21]
Lex Fridman
Do you already sell those vans to others? Will you sell those vans to others?
- [05:24]
RJ Scaringe
We just started about a year ago selling to others. But the ramp up has been slow. A number of. When you think of large operators of vans, very often these are not the companies that are going to take the biggest risks on technology or be on the front edge of early adoption. So we think it's like they take smaller steps with pilots, prove the model, prove that they can support the charging infrastructure and then once they see the economics or so advantage, then I think many will make the switch. Yeah.
- [05:52]
Lex Fridman
Going back to the Rivian story being a bit unlikely. The other thing that seems to me is getting reliable production going is really hard. And again, it feels like the first cars that roll off the line should be super janky and unreliable and everything like that. In preparation for this episode, I've been stealing my wife's Rivian. I've been really enjoying it. And it is. You were complaining because you were making fun of me because it's a used version that we bought, not a new version. So it's the 23 model, which is one of the earliest ones and it's awesome and it works great and the build quality is incredible and things like that. And so again, that's the other part to me that feels unreliable of the. It feels like that's not how hardware companies usually go. So how did you guys get the quality and reliability good out of the gate?
- [06:36]
RJ Scaringe
Yeah, I mean one of the things that's so different about building a car company versus most software, most technology companies, and for that matter even how venture capital has been, how it's evolved and how it takes place is the first product of most startups like usually a tech or Software centric startup is pretty loose. It's there to demonstrate that the idea holds promise. And it may not have scalability built into it, but it shows there's a product market fit. And then following that, there could be more capital invested to grow it, to add robustness to the product.
- [07:10]
Lex Fridman
The minimum bet size for really big.
- [07:12]
RJ Scaringe
Yeah. So in a vehicle, your first dollar of revenue takes many billions of dollars of development. And the thing you launch needs to be incredibly robust. And so that means like multiple winters of testing, multiple hot summers, we testing vehicles of the desert, very large durability and reliability teams set up to run this. And there's like cultural challenges when you think about building a new company. One of the strengths of a new company very often, and I think what allows startups and entrepreneurs to excel is they're willing to try new things as businesses. That it gets embedded into the culture. There's a certain move fast and break things mindset. And that's great for technology, it's great for pushing boundaries. But it's very different than the skills and the cultural mindset. You need to run stable operations where let's say if you're running a manufacturing plant, you need repeatability, you need a lot of process orientation. And so just for us, this step of going from a product development company, developing new technology, making prototypes which are wonderful, and demonstrating that those prototypes were capable and then transitioning that into a production system that can scale and has high quality and robustness built into it, that was. It took a lot of work. It was hard.
- [08:31]
Lex Fridman
But the first version of many hardware products have a lot of teething pains. Like, did the R1 have a lot of teething pains? Did you guys do something different to avoid that?
- [08:41]
RJ Scaringe
Well, so for us, we launched, we made a few. If I could go back in time, I probably would change one thing. We launched a truck and SUV. So the R1T and the R1S, it's a sibling set of products. At the same time, we launched a commercial van. So we launched three different products at the same time. And we were building out our plant starting in the beginning of 2020 and then launched in 2021. And so like trying to build out an industrial ecosystem, turn on a supply chain and then ramp a product is hard. Doing that in the middle of COVID is really hard. And doing that across three different product lines was like extremely hard. And so I probably would have inserted more sequencing between all those different products. But nonetheless, when the R1 launched, we had, we won vehicle a year we had, which was awesome. And so Motor Trend does this. A few different organizations, we had like multiple truck reschedule of the year awards. And then Consumer Reports does this annual brand survey and they look at customer satisfaction, customer appeal, and they look at likely to repurchase. And so in our first year of production with our first product we've ever produced, we came out by far number one on that, which was amazing. And in our second year, we came out by far number one. So like the closest, we beat the next closest by like 15 points. So that was amazing. It was really just a reflection of how much work the teams have put in to make all these product level decisions. Trade offs around. You know, you think of a vehicle, it's like many tens of millions of trade offs you're making around cost versus performance versus a feature. And so we'd really done a nice job of getting the product market fit right. So it really connected with consumers. But saying all that, the ramp was hard. So launching late 21, it was difficult because we were coming off the back of COVID And then we had this awful supply chain crisis that hit, you may remember like 2022, where getting anything from drywall to semiconductors was hard. But in our case, we were last in line with many of the suppliers because we were new. And so you can imagine we went from having a bill of materials that was already inflated relative to existing large incumbents to something where if we just wanted supply, they said, look, you're going to have to pay us an extra 20%. And so it was like this across thousands of parts in a car. All these discrete decisions of do we stop production for this part that should cost $2, but we're being asked to pay 4. And you do that several thousand times and it adds up. So we had a cost structure that was really challenging. We chose to actually ramp.
- [11:15]
Lex Fridman
Yeah. And I presume you mostly just ate those cost increases because it doesn't make sense to spike the cost of the vehicle. Did you debate whether you should pass along the cost increases or eat them?
- [11:25]
RJ Scaringe
Well, in parallel to this, of course, inflation went really wild in 2022 into 2023. So the, the price of vehicles grew quite a bit in that same time frame. And so we did add in reaction to that, we did have adjustments to our pricing, but we've tried to be really stable on pricing because it's always hard. If your pricing is moving around a lot, it wreaks havoc on the residual market. It's not great for consumers. This is really hard for consumers. So we've tried to be really methodical with that, but saying all that, we had to sort of grow through 23. We resourced a lot of our billet materials. We launched what we call Gen 2 of the vehicle, which you don't have, but is a meaningful step forward in terms of technology, but also a considerable reduction in the overall cost of the vehicle. And so finally, end of 24, beginning of 25, we started to hit positive gross margin. So Q4 of 24, Q1 of 25, positive gross margin. We're extremely bullish on long term profitability, but this business is really tough if you have disruptions to supply. Just the fixed costs are really tough.
- [12:33]
Lex Fridman
What's the Gen 2 pitch like? What's most exciting from a tech point of view? Is it autonomy? Is it something else?
- [12:38]
RJ Scaringe
Boy, yeah, there's a lot there. We didn't change the exterior of the car much, so really we redid a lot of the interior, like the guts of the car. And the biggest change was on Gen1. We designed and built all the electronics in the car, but we allowed for a number of the compute platforms to be specific to an area like specific to a domain.
- [13:03]
Lex Fridman
The zonal architecture where you move to.
- [13:05]
RJ Scaringe
Yeah, the reason we did that at the start is we needed to move fast. Let's say you're on the body controls team. Your timeline isn't necessarily the same as let's say the powertrain controls team or the vehicle dynamics team. And so we allowed the proliferation of a few different domains that we knew long term would consolidate. But we made the decision because we said we own the whole software stack, we own the electronics, the consolidation will come in step two. And then with gen two, we went from 17 down to three. And so now we have a, what we call an east zone controller, a west zone controller and a south zone controller.
- [13:34]
Lex Fridman
When I say controller, these are the computers and the integrated systems management.
- [13:39]
RJ Scaringe
Yes, there's three real time computers in the vehicle that run the whole platform. So they run everything from propulsion to body control to your lighting, all the real time systems. And that was a big lift to move all the software into this new architecture. But we also use that as a chance to completely change out the way we approached autonomy. And so that effort started in actually about late 2021, where we recognized we needed to move entirely to a platform where we own the entirety of the perception stack. So all the cameras in our case also the radars, a much more powerful compute platform for autonomy and launch that in a way such that we can build this large data flywheel where we can train the system and what we often call now end to end training, where we have cameras, radars that feed in, that's building a large parameter multi billion parameter foundation model that governs how the vehicle drives. Then deploy that in a distilled state back into the vehicle. So that's now in the car. And the slope of the curve in terms of progress and additional features is going to be really steep. Your Gen1 won't have that. Gen2s will, but that was a huge part of the lift. And all that laid the network architecture, the consolidation of ECUs, this refactor software stack, it laid the foundation for us to do a lot of other things. So we're about to launch our lower priced vehicle which we call R2. It's the coolest thing we've ever done. It is awesome. It's a smaller suv, five passenger. As I was telling you before, you will love it. If you like the R1, you'll really like the R2. But it's starting price of $45,000. It has all the essence of a Rivian that you see in R1, but you know, squeezed into a slightly smaller and higher volumes, I presume. Much higher volumes.
- [15:24]
Lex Fridman
Can you give us the comparison between R1 volumes and R2 volumes?
- [15:27]
RJ Scaringe
So R1, the average selling price is just over $90,000. And so it's the. In terms of these things are always subject to boundary diagrams. But if you draw the boundary diagram around electric premium vehicles. So like premium SUVs that are over $70,000 or by far the best selling vehicle in that class of vehicles. So we have about 35% market share as we draw that boundary diagram. Said differently, we have the best selling electric SUV over $70,000 in the United States. Interesting. In California and the state of Washington we have the best selling suv, electric or non electric in California. In the state of Washington.
- [16:06]
Lex Fridman
That fits with what I see out on the roads.
- [16:08]
RJ Scaringe
Yeah. So if you're here in the Bay Area, like everywhere. So it is the best selling premium suv. But the size of the market for someone who's buying a $90,000 vehicle is just naturally more collapsed.
- [16:17]
Lex Fridman
So kind of how many R1s do you ship a year?
- [16:19]
RJ Scaringe
This year we'll do call it 40ish.
- [16:21]
Lex Fridman
40,000.
- [16:22]
RJ Scaringe
Okay.
- [16:22]
Lex Fridman
And so what are you tooling up for in terms of R2 production?
- [16:25]
RJ Scaringe
So R2 we're going to build first in our plant in Illinois, just south of Chicago, a town called Normal. And there we'll be building up to 175,000 R2s in addition to R1, in addition to our commercial vans. But we're also building a plant in Georgia which will be like the further expansion and the size of the market for R2 and then R2 variants. So there's an R3, there's other things we haven't yet announced, but there's a whole host of products that sit on that platform. You know, it's many millions of units. It's just the whole, if you think of like the whole automotive demand space. The average price of a new car in the United States, $49,000. The most popular configuration is a five seat SUV or crossover. And so it's just a huge space.
- [17:07]
Lex Fridman
Okay, so you're on the cusp of a very significant ramp up.
- [17:09]
RJ Scaringe
Yeah, so sort of the first step is going from, in total, call it sub, 50,000 units a year. The plant will be capable of 215,000 units with, with just our normal facility, normal Illinois facility. And then we add in Georgia across two phases, another 400,000 units of capacity. We've taken the approach of very heavily vertically integrating technology. And so everything from the low voltage systems we just talked about, so the computers in a car, the software that runs in them, the perception stack, our self driving platform, our whole infotainment platform, and then into the high voltage side batteries, gearboxes, motors, inverters, DC DC converters, DC AC converters, all that power electronics, all is done in house. And so when you're producing less than 50,000 units a year, the engineering fixed cost and the operational fixed cost to run that is very high. So it's a structural cost challenge. But you've each cost to run which, well, just to run an engineering team to engineer all that content, but it's only finding its way into a small number of vehicles.
- [18:11]
Lex Fridman
You're saying the systems are very complex and so you want to amortize that.
- [18:14]
RJ Scaringe
So once we add volume then it becomes this enormous structural cost advantage because we don't have, typically you have a lot of tier one suppliers that provide all these systems and they add a lot of cost and for that matter also a lot of complexity. So because we do it all ourselves, all that complexity and cost is wiped away. But you need the volume to really have it make sense.
- [18:36]
Lex Fridman
It seems interesting to me that there was a gap in the market for an electric suv because when you think about it, you guys were thinking about starting originally with a sports car. That's what you found at the company to start with. And my sense is that you said, huh, the electric Company starting with an electric car company starting with a sports car idea has kind of been taken.
- [18:55]
RJ Scaringe
That's exactly, yeah.
- [18:57]
Lex Fridman
So maybe we'll pivot. But I've driven the Tesla Roadster. It's not a good idea to put a ton of batteries in a sports car. It doesn't handle well, whereas an SUV is heavy. Anyway, it's perfectly suited to being an electric vehicle. And so why were you guys able to arrive into the market and take what feels like the most obvious form factor for an electric vehicle?
- [19:17]
RJ Scaringe
Boy, is that a good question. It's funny how things change over time. So today it seems obvious, but in like 2013, 2014, I would go meet with suppliers and CEOs of very large suppliers and say we're working on an electric SUV and electric truck as a flagship product. Following that, we want to launch more affordable electric SUVs. And the lack of understanding of what an electric vehicle can do in those conditions was so extreme that it's hard to even appreciate this. I mean, I would get to say laughed out of the room is probably accurate. I mean it would be like, why would you make an electric suv? And the questions would be, well, what if it has to go over a bumpy road or what if it needs to go through water? And so there's this perception that you couldn't drive an electric vehicle off road, you couldn't go through deep puddles. And so born out of that, we realized it was actually this amazing opportunity to not only say, yes, you can make an electric SUV or yes, you can make electricity, you know, lifestyle pickup, but you can make it so much better than anything in the combustion world. And so we, we really had fun with that. I mean, the R1, our new quad motor R1, does zero to 60 on the R1T. The truck will do zero to 60 in just under two and a half seconds. It can manage the most complex off road trails, you know, like Hell's Revenge or any like the really hard stuff where you're going up, you know, almost, you know, 45 degree angles, steep rock faces, it can do that no problem. And you can use it every day. And so like the, the comical thing is you can, you pull up to a stoplight, you like look to your left and you see like a brand new Ferrari. Ferrari. And you're actually quicker than that car. Yes, yes. And you've got like groceries in the front trunk, your kids in a car seat and like silently it just out accelerates any of those things and then it can veer off the road. And drive over a mountain. So it's, it's just we wanted to make this product that just completely challenged those incorrect assumptions around technology.
- [21:17]
Lex Fridman
So if you rewind Those suppliers in 2013, 2014 is what they missed. One, they were maybe thinking about the Ford F150 market, which will be slower as adopters, but there's a very large, just general everyday SUV market and kind of outdoor enthusiast market is kind of thing. One, they kind of misunderstood some of the truck markets. And then secondly, you can actually make a better vehicle when you have all this abundant ship's power available to you. Because especially as you get into the outdoorsy stuff, having just an abundant source of power is actually pretty nice for a lot of the other functionality.
- [21:51]
RJ Scaringe
Yeah, I think with electrification we have found that particularly early on this has shifted, but there was this perception of what an electric vehicle buyer was. And if the description of a Rivian was it's electricity and then you describe all these compromises that sort of fits the historical, call it 2014 description of electric vehicle buyer. You're buying a vehicle, accepting a bunch of compromises. But we have customers today that are not buying it because it's electric. They're buying it because it's great to drive, it's fun, it's capable, it fits their lifestyle.
- [22:23]
Lex Fridman
No one ever got super excited about a Chevy Volt. They're like, this is the car I wanted.
- [22:27]
RJ Scaringe
We wanted a vehicle that was like, if you're cross shopping this with, you know, with a Porsche Cayenne Turbo or like a Wrangler Rubicon or, you know, you'd be like, wow, this is a better vehicle. And it actually does the things that the Rubicon can do and it does the things that the Cayenne Turbo can do in one car.
- [22:44]
Lex Fridman
Where did the culture of surprise and delight features come from? Because that's one thing I associate with Rivian. We have here your gear guardian is that this fellow's name is the kind of animated fellow who kind of makes clear to any ne' er do wells that they are being monitored if they're trying to break into the car. You also have things like the flashlight.
- [23:05]
RJ Scaringe
Is a very cute feature.
- [23:07]
Lex Fridman
And then I also noticed it's not quite surprising to light, but there are often features that other cars will have that I find they're just executed really well in the Rivian. And so the functionality where you pair it with your phone and then it kind of locks and unlocks based on your coming and going. Lots of cars claim to have that, but unless it works super reliably. It's actually more annoying than not having it. Right. Because you walk away from the car and you're like, huh, did it lock?
- [23:27]
RJ Scaringe
Or whatever.
- [23:28]
Lex Fridman
And again, the execution of that feature on the Rivian is really, really good.
- [23:31]
RJ Scaringe
And so it's better on Gen 2.
- [23:33]
Lex Fridman
I will, I promise that someday I'll upgrade. Where did this come from? Why have a flashlight in the car?
- [23:40]
RJ Scaringe
I'm a huge car enthusiast and I think that's really helpful in that I love, I love cars. I've grown up loving them. And so you like, as a kid, when you're growing up reading about cars, it connects with this emotional, almost dreamlikes part of you where you're like imagining and you're dreaming of things. Like, I have three kids now and I ask them to sketch. Hey, sketch me what an R12 looks like or an R9 or an R23. Like the ideas they come up with are so wonderfully removed from the practicalities of what's possible. And they have all these like surprising features. Sometimes they're like, wildly not possible. But we wanted to have some of that like, kid like joy built into the product and even the aesthetics of how the vehicle looks. It's, you know, it's one of the fastest cars you could buy today. But it doesn't look mean, it looks friendly.
- [24:33]
Lex Fridman
Yeah.
- [24:34]
RJ Scaringe
And so we wanted the vehicle to be lovable. We wanted like a lot of when you fall in love with something, it makes like the little things nice.
- [24:41]
Lex Fridman
Yes.
- [24:42]
RJ Scaringe
And so, I mean, I was at. My kids and I were at a farm the other night and it got dark and we're like walking around in the cornfields and we needed a flashlight. So it was like, perfect. We walked back to the car, pull the flashlight out of the car, and everybody that was with us was like, how'd you get a flashlight? I'm like, oh, I have one in the car. It's built in. And so there's like those little features that they just unlock like that made the whole night different for myself, my kids and a bunch of other people. And so we thought about those kinds of features really carefully.
- [25:08]
Lex Fridman
What are fun features that you left in the cutting room floor that you would love to have put in or maybe you'll do in a future car.
- [25:14]
RJ Scaringe
But on R1. Yeah. I shouldn't say this here because everybody's going to say, bring it back up. Until very, very late. In R1, we had two roof options. One's the glass roof, which is what you guys have in your car. And the Other was this beautiful carbon fiber roof that like you could take it off the top and it fit like perfectly puzzled into the front trunk.
- [25:36]
Lex Fridman
Oh wow.
- [25:36]
RJ Scaringe
And so you get this like now when you look up the glass, imagine there's no glass there and it's open. So you get this really nice open arch. Yeah, that was really nice. Especially in a car like this to have that capability. Future R1. Someday we'll bring that back. That was really cool. We all lamented losing that, but it was just a lot. Again back to complexity and launching. We decided just to manage our own complexity in that. But that was one. Now R1 has just about everything we wanted. The new quad we brought back this we call kick steer feature where we can use the fact that there's four motors in the quad and the four motor version to introduce torque vectoring to an extreme degree. Where you can.
- [26:13]
Lex Fridman
Like an F22 Raptor, you have crossfactoring.
- [26:16]
RJ Scaringe
Yeah. You can pivot it on its axis. You can. So it's really fun off road. It's not an on road feature you're not going to use on the highway, but it. And it's quite useful. Not with that attitude. Yeah, not without it, but. So we brought that back. I think as I look at R2 there's some features on it that I. We didn't put in R1, but that really make R2 special. So R2 still has a flashlight, but the rear glass on the liftgate now it drops. And I've been using an R2 with the kids to do like school drop offs. And it is so cool to come up with the back glass open and the kids just throw their backpacks in through the back of the car. So I love that. And then we've added a few other like really thoughtful features around. Like all the storage compartments are thought about from the perspective of like do you have kids? Do you have pets? Do you have a backpack? Do you have a handbag? Do you have something that you need to put somewhere and where does it go? And increasingly I found in a lot of modern cars you just don't have places for all this gear. And so we do Personas. So imagine you're a backpacker. Where does all your gear go? Imagine you're commuting to the office. Where does your gear go? Imagine you've got two kids and a diaper bag. Where does that go? And so we really have thought about it from the perspective of users and that user centric design just manifests into all these things being thought about really.
- [27:35]
Lex Fridman
The last time I saw you, I was asking you about CarPlay and I was saying why you guys needed that. But then I realized afterwards that I think when people say they want CarPlay or when I said I want CarPlay, what I actually want is navigational and traffic data from a source that I trust and the inbuilt navigation system in a car like, you know, from a Mercedes or BMW or something like that. No one thinks that they have the best traffic data. And then you guys launched Google Maps and it's awesome.
- [28:04]
RJ Scaringe
It's great.
- [28:04]
Lex Fridman
How did that come about?
- [28:06]
RJ Scaringe
Well, you just nailed it. This is a decision that's generated. I said there's many millions of decisions. Many of them will never get noticed and they're just under the surface. One of those decisions that's been noticed quite a bit is the fact that we've intentionally not included CarPlay in the vehicle. And that's not to say we don't think a close partnership with Apple is important. So we have Apple music integration, we have a bunch of Apple integrations that are yet to come, but a great relationship with the team at Apple. But there's more to say. We just felt and continue to feel very strongly around creating a consistent, fully integrated digital experience where you're not jumping between apps, let's say from like a CarPlay app, back to the vehicle app. And it's quite jarring when you don't have, let's say vehicle level controls when you're in the carplay environment. That view we've had since the early, early days, I think that's become going to become even more important and more true in a world of integrated AI. And so suddenly when you can talk to your vehicle and say, please open the front trunk, please unlock the doors, lower the windows, set the temperature, take me to this place, I'm hungry, I feel like tacos. Find something along the way to the Charger. All those immersive experiences really need the full breadth of context from your mobile device, from your vehicle, from the vehicle back to you. What's the state of charge? And trying to do that through a homogenized single layer app that works across many, many brands and by definition in some ways has to play to a lowest comm denominator is really challenging. And so there's a few areas that by not having carplay it generates friction. The first is maps. And so until we launched this partnership with Google, we had built our own mapping platform where we took open street maps and we layered in a routing platform on top of that and we had information that we would buy or on traffic, but it was not going to be as good as Google.
- [30:00]
Lex Fridman
Yeah, I don't even know if it's irrational for me to not trust us, but I just noticed I don't, you know, it's better.
- [30:06]
RJ Scaringe
It's a big step forward with Google, how we've done it. So that I'd say fully eliminated the concern on mapping and then the next is text and voice to text and being able to interact with your device. That's equally important for Android and for Apple customers. And so that's coming very soon and then there's really no gap. And what we found is working directly with Apple on the applications that folks care about, like Apple Music, like these are the kinds of things we'd rather do and give customers this more highly curated choice of different platforms, but within the Rivian ecosystem.
- [30:40]
Lex Fridman
No, again, I was struck by, after we discussed it that again, I don't actually want CarPlay. And it's kind of funny when you think about it. There's like the screen with all the icons and like the audible app and like no one wants to go browse through their audiobook library on the car screen. What they want is navigation data from a source they trust to play audio from their phone. Those are the things they actually want. Can I give you my pitches for my suggested Rivian features? You do the worst job ever.
- [31:04]
RJ Scaringe
Bring it on.
- [31:05]
Lex Fridman
Have you seen that episode of the Simpsons where Homer gets to design his own car?
- [31:08]
RJ Scaringe
Somebody gave me the picture of that car.
- [31:10]
Lex Fridman
Exactly. Yeah, the Homer car. But I feel like it's a classic bike shed issue where I don't know, no one's that familiar with what stripe does at the innards level. And so they just assume that stripe is doing all the sensible things. Whereas everyone drives a car. And so everyone could be like, here's what you guys should do. So with that, here's what you.
- [31:28]
RJ Scaringe
Family members, friends. Totally.
- [31:29]
Lex Fridman
Exactly. Why do all cars offer memory of the driver? Either the old fashioned way that other cars do it, where you press the position one or two for the seat. You guys do it, connect it to the phone. It's really nice. That moves the seat, that moves the side mirrors. No car moves the rear view mirror. Why is that the final frontier of adjustability?
- [31:55]
RJ Scaringe
Well, there's no actuator on the mirror.
- [31:58]
Lex Fridman
Sure. But that's your choice.
- [32:00]
RJ Scaringe
It's definitely our choice. We could. It's a cost trade off.
- [32:04]
Lex Fridman
Yeah, but you have actuators on the side mirrors.
- [32:07]
RJ Scaringe
Yeah, they're harder to Reach, though the.
- [32:09]
Lex Fridman
Rear view mirror is harder to reach.
- [32:10]
RJ Scaringe
So, like, the side mirrors, if you go back in time, if you go back 25 years, those are the heart.
- [32:15]
Lex Fridman
Of the driver's race. Yeah, yeah, yeah.
- [32:16]
RJ Scaringe
They used to be manually adjusted. And then like a big. Yeah, the little knobs. And then in the old, like, way, way back, you'd, like, have to push on the glass to, like, move it around. Then we had little, like, like, knobs.
- [32:27]
Lex Fridman
Advanced knob technology.
- [32:28]
RJ Scaringe
And then we moved to electrically controlled. And, you know, in those cases, we moved to electrically controlled because it's, like, hard to reach out the window or reach across the car. In this case, it's just right there.
- [32:36]
Lex Fridman
Yeah. But you guys are the surprising delight.
- [32:38]
RJ Scaringe
So here's the question we always have to ask ourselves. If we have a certain pool of dollars we can spend on the car are like, we're essentially, we're the arbiters of how those dollars get spent. And we can't make everyone happy. Somebody will maybe wants that, but to do that, we can't. Somebody else. Something else either has to fall out or the price has to go up.
- [32:57]
Lex Fridman
Okay. If you're worried about the bond.
- [32:59]
RJ Scaringe
So we end up taking, like. Those are the kinds of things we just.
- [33:02]
Lex Fridman
You could also do it, like, you know, when you see the really big SUVs, they actually just have a video feed projected up there. You could also do with a video feed, which then would not require any adjustment completely.
- [33:11]
RJ Scaringe
Yes, we actually do have. We have had a lot of debate around this. This is. This is very actively debate where you see the mirror. It's not actually a mirror. It's just a screen.
- [33:18]
Lex Fridman
Yeah.
- [33:19]
RJ Scaringe
And we've not done that yet. If I'm honest. Partially not partially because of me. I've not loved.
- [33:27]
Lex Fridman
Oh, the team is coming and pitching you on.
- [33:30]
RJ Scaringe
There's a team that would like to do it with the mirror as a screen. And I've been against it in part because if you have good visibility. So if you can see out the back, it's a little disorienting. And so my concern is you make 50% of people happier with it. But a lot of people would say, just give me a mirror. And so we just made the trade to say we're going to take those dollars and spend them elsewhere. There's not anything wrong with those products. It's just a decision.
- [33:56]
Lex Fridman
Okay, next question. California mandates booster seats up to the age of eight. And I've always found it funny that we are boosting the child as opposed to moving. The third point up and down on the seat belt, where you could just have an adjustable seat belt and then ensure that the child is correctly configured where you move the third point down. And now the seat belt is appropriately across their lap and it would just be much less stuff to be carted around. You're not requiring boosters. You just adjust the seat to be correctly configured to the child size. And Volvo kind of does this with the XC90, where they have an integrated booster seat in the middle seat, but it's only one seat, and it's actually still a booster and everything. Why not do adjustable seat belt points as a child restraint device?
- [34:36]
RJ Scaringe
So we do have the adjustability of the third point. We can adjust it. It just only goes down to a fifth percentile female. So it doesn't cover the kid range. And the reason for that is it's classic unhealthy consequences. So when you have. We have to test at different settings. And so we test across from 5th percentile female to 95th percentile male. And that's what that range is set up for. We also, uniquely in the United States, there's no other developed country that has this where we have unbelted tests. So we run a lot of our crash tests with no belts on. And so the kinematics of a body moving in the event of a harsh collision are very different if you have your seatbelt on or your seatbelt off. And so it's a very interesting question. There is no clear answer. But if you were to say we're going to test around all occupants being belted, you would end up with different solutions and different control algorithms for the airbags.
- [35:31]
Lex Fridman
Ah, that's so interesting.
- [35:32]
RJ Scaringe
And so we also plan for unbelted in a way that I think other countries don't contemplate that.
- [35:38]
Lex Fridman
Sorry. And that is when you say we. The United States has a certification.
- [35:42]
RJ Scaringe
United States. Yeah.
- [35:44]
Lex Fridman
I didn't know the United States was an Adler. There are there places where if you could change the certification and safety rules around cars, you would make changes.
- [35:54]
RJ Scaringe
It'd be really nice if the United States and Europe had one set of standards. Yeah. The biggest difference in terms of, like, safety between the US And Europe is in pedestrian production.
- [36:05]
Lex Fridman
Okay.
- [36:05]
RJ Scaringe
So when you think about if you hit somebody on the road, what you want is you want to have them go up over the car.
- [36:10]
Lex Fridman
Yeah.
- [36:11]
RJ Scaringe
But they're going to. They're likely going to have a hard impact with the hood.
- [36:13]
Lex Fridman
Yeah.
- [36:14]
RJ Scaringe
And so in Europe, there's very tight pedestrian production standards. And so the shape of the bumpers have been designed to make sure they get, they get brought over the top of the car and then the hood needs to have some give to it. And so we've designed our hoods. Like a big part of the hood design is to contemplate a head, like a human head hitting the hood. There are certain cars that you see here in the United States that could never pass pedestrian production in Europe. Like for example, a cybertruck could not work in Europe.
- [36:44]
Lex Fridman
Interesting. You wouldn't expect a carmaker like Rivian torture to use Stripe. But one reason they do is they need to accept a huge range of payment values. So when you're putting down a thousand dollars deposit on the car, you probably want to use your credit card. But when you're actually ready to buy and you're spending $90,000, you might run into a few issues with that. So Rivian uses Stripe to securely accept bank payments using ach, the right payment method for the situation. It just depends. For Rivian, they want bank payments and many businesses on Stripe, they want to accept the blossoming variety of international payment methods. If you're expanding to Brazil, for example, you need Pix, which is now how most people pay online there. In Poland, offering the local option Blick can lift conversion by over 40%. And globally, customers increasingly expect point of sale credit using services like Klarna. Stripe's optimized checkout suite handles all these payment methods and their differences for you. We use AI to analyze the customer and the context they're in and then we dynamically surface the payment methods over 75 different options most likely to convert for that specific customer. Learn more@swepe.com payments. Are accident rates going up because of texting and driving?
- [37:57]
RJ Scaringe
Yeah, for sure.
- [37:58]
Lex Fridman
Okay.
- [37:59]
RJ Scaringe
But countering that is as the vehicle increasingly drives itself.
- [38:01]
Lex Fridman
Sure, that's where I was going to go. There's a battle between people are becoming worse drivers, which is kind of bad and scary, but true because they're just increasingly on their phone. And we were walking through the city at one point recently and the car kind of blew past the pedestrian crossing, right as people were kind of in the process of crossing. And just like this old guy next to us as we were crossing the road said, you have to be on your phone to be driving like that. Apt description of much driving these days. But how much are the improvements in semi autonomy, like the fender bender detection and stuff like this helping?
- [38:38]
RJ Scaringe
I mean, I have a nine year old, a seven year old and a six year old and I truly think that by the time my Oldest is able to drive. The environment's going to be so different than what we see today. And we're in this the next five years we're going to see autonomy features where the idea of like hands free or hands free, eyes off go from being something that a very small percentage of customers are looking for or using to something that's essentially expected. And so for that reason, it's by far the biggest investment area for us in our business. It's like it's the most critical area of technology development. But what's changed so much that has me believing deeply that the next five years will look very different than the last five years is just the way that AI is now part of this system. So historically the way self driving systems were created was you'd have a perception stack, it would see the world, it would identify all the objects, it would classify the objects like person, car, tree, assign vectors to those objects, so acceleration velocity for each of those in X, Y and Z. And then it would hand all those objects and their classifications to a planner that would be rules based. So a planner written by humans that would describe the rules of the road and for which would make a whole host of decisions around what the trajectory should be for the vehicle. That whole approach has been completely replaced with what I would call like Gen 2 of autonomous vehicles, which is you still perceive the whole world, but you've used a transformer to encodal that images. You then feed it into a very large model and you train it in the way that you train at LLMs. You're building this very humanistic understanding of the world. And if you have a car park that's large enough like what we do, you can create this amazing training flywheel. And the rate of progress is so much faster than trying to program how the world works because you just have.
- [40:28]
Lex Fridman
Cars going around in the parking lot.
- [40:31]
RJ Scaringe
Not yours, yours, Gen 1, but all the Gen 2 vehicles.
- [40:33]
Lex Fridman
They're gathering training data for this.
- [40:34]
RJ Scaringe
Yeah. So the obvious ones are if you're in self driving mode, if you're on the highway for example, in our vehicle and it disengages, that's a piece of information we train off. That more interesting is when you're driving, you as the human, the model is running in parallel and when you do something different than the model predicts. That's interesting. It's like if you change lanes and the model predicted you would not change lanes or vice versa. If the model predicts we should change lanes now and you don't change lanes on a singular Basis. Okay, it's sort of interesting, but you multiply that across thousands and thousands of vehicles, it becomes a really powerful method for teaching a digital driver how to drive.
- [41:12]
Lex Fridman
Yes. I feel like the big debate that plays out in the public realm on this stuff is Basically the Waymo vs Tesla approaches of more sensors on Waymo's vs fewer sensors on Tesla. And Elon's being quite dismissive of LiDAR. Where do you stand on this sensor debate?
- [41:29]
RJ Scaringe
I think first is I'd say there's alignment amongst Rivian, Tesla, Waymo around the use of AI to train. And so the view that you're building a large scale foundation model, multi billion parameter model, you're building that with the data flywheel of the vehicles that are deployed, there's consistency there. I think for some reason this has been a lightning rod issue in ways that I wouldn't have expected. If you look at sensor theory and even if you have noisy sensors, just mathematically it proves that having more sensors is a better approach.
- [42:04]
Lex Fridman
This always seemed nonsensical to me. Yeah, more sensors obviously better. And we know how to.
- [42:08]
RJ Scaringe
So that's the reason you have more than one camera. So even moving beyond more than one camera, we have many cameras around the vehicle. But then saying we want additional modalities that have a non overlapping set of strengths and weaknesses with a camera. And it's not. There's no arbitration of the camera versus the radar or camera versus the lidar because it's just not how the models are built anymore. They have a worldview and then the addition of different modalities, it actually helps balance out some of the noise and the signals. Because every signal is going to have noise, every camera of noise, every lateral have noise, every radar.
- [42:41]
Lex Fridman
You just want as many sensors as you can afford in the car, then combined.
- [42:46]
RJ Scaringe
Yeah. And then it becomes as many as you can afford and as many as you can process, which ties to how much you can afford. And so a few things have happened in the last five years that have changed things. One is we moved to a neural net based approach. That's what we talked about already. The second is that the cost of sensors is six, seven years ago a lidar was $20,000. Fifteen years ago, lidar was $75,000. Today a lidar is like 200 bucks. So it's a very low cost sensor and it solves certain things like very bright light, very low light, extremely well. And similarly a radar is 100 bucks depending on which one. You can buy the crappy radar for 25 bucks. You can buy a great imaging radar for 100 bucks, 125 bucks. And it gives you rain, snow, fog, in ways that cameras obviously don't perform very well. And if you've ever driven in really thick fog, you probably thought to yourself, boy, wouldn't it be great if I had a radar on my forehead? That's what planes have, which I think all the time. Yeah, but you would navigate those situations with a much higher degree of safety. But then the last piece I'd call out is when you have multiple modalities, it allows you to train your perception stack better. And so, for example, if you're driving in the fog, there are subtle little things your eyes will spot and you're hunting for. Like you're looking for the little red dots of the car in front of you from the tail lights. But you have to hunt for them.
- [44:06]
Lex Fridman
Yeah.
- [44:06]
RJ Scaringe
Imagine your brain had the assistance in training its neural net to immediately know where those objects are vis a vis a radar. So your brain can more quickly identify the visual characteristics. And you build this much more robust neural net to be able to see in hard situations, bright lights and other. Really, you're like, you know, coming, let's say, coming down the 101. You've got that sun right in your eyes. It's really helpful to have something that, that doesn't affect. Yeah, but so I think it's strangely become a debate topic. I think it's not really.
- [44:39]
Lex Fridman
Everyone's actually in agreement on the high level points.
- [44:41]
RJ Scaringe
I think everybody's actually everyone's agreement on.
- [44:43]
Lex Fridman
Kind of the overall AI approach. And then, yeah, actually there isn't that much disagreement on the sensor stuff or maybe less than it would appear. You're describing some of the progress in components, so you're describing how lidar and radar have really changed and becoming a lot cheaper. What else is happening in the land of components that you work with? Are tires getting better, batteries, tires, windshields, anything like that?
- [45:05]
RJ Scaringe
I think on the electric side, we're seeing like the motors become lower cost. So I'll illustrate the point. What's going into R2? It's our fourth generation motor that we've designed. And so previous generations used what's called a hairpin winding in the stator. So you take copper wire, you bend it into a unique shape and you have. You've seen these before, I'm sure there's like a whole ring of wires. And then to bring them together, depending on how you designed the electromechanics on it, you're welding all these little tips Together, that was state of the art. And that's what essentially 99% of the cars in our electric vehicles on the road are now doing on R2. On the drive unit that we called Maximus, which is my middle son's name, which is a great name for a motor. We have a continuous winding, so it's one long wire that we wind. It's sort of like woven into like a mat and then we fold it on each other. And what it does, it takes the number of welds in the motor down by something like 95%. So it's slightly more efficient, but importantly, it's a lot less costly. And so does it materially drive different? No. But does it allow us to continually drive the price of electric vehicles down? Yes.
- [46:14]
Lex Fridman
Okay. So a lot of components that are pretty important for electric vehicles is like a meaningful cost curve improvement happening on them.
- [46:21]
RJ Scaringe
Yeah, yeah, yeah. And then I think if you take the complexities of designing, you know, designing real time operating systems on a compute platform across the vehicle, these are things that car companies don't historically do very well. We've got as an industry, there's this massive supplier base of tier 1s that provide functions and computers alongside of them. So you buy a seat, it historically has come with a little computer. You buy a car with power windows, it has a power window computer. And so with the exception of two companies, Rivian and Tesla, every car on the road has, depending on the amount of content, anywhere from 50 to maybe 150 little computers that run the car.
- [47:02]
Lex Fridman
This is why you in a normal car cannot adjust, say the side mirrors from the center console. Regardless of whether you want three different.
- [47:11]
RJ Scaringe
Companies and different software. It's islands of software written by little supplier on little ecus that go back to suppliers. And so that actually underpins. We did a $5.8 billion deal with Volkswagen where we're providing them, this is the second largest car company in the world. We're providing them a zonal architecture, so a platform that allows us to run the whole car with a very small number of computers. Depends on the size of the vehicle. And then a software platform. I think it's inconceivable that by call it early2030s, that a car company can exist at scale and maintain their market share and not have a software defined architecture. And beyond an architecture for which they have, I think of the software architectures that are incarcerated, it's like a field of weeds. It's like you just have all these little ecus all over the place. They're all little islands of software. You can't make updates to them. None of them are written by yourself. There are so many abstraction layers between the actual code and the manufacturer. I think that must go away for you to be competitive in a world of AI where you want deep contextual understanding of what's happening across the vehicle and being able to create these highly immersive, highly evolving experiences that get better and better over time. So I think every car company is either going to go try to develop it themselves, which is hard because they don't typically have those skill sets, try to source it from suppliers. That's very hard because those companies are precisely the ones that don't want to see all their little computers go away or work with us. Obviously, I have a bias point of view, but we think of the Volkswagen partnership as the ultimate billboard for the idea to be able to deploy this platform in such a wide range of vehicles, from flagship Porsches down to a $22,000 European Volkswagen, demonstrates the scalability of this architecture. And so I think that's going to be a huge change that must come. And if it doesn't, sort of like, if manufacturers don't make that change, they're just going to lose market share and the ones that do have that technology are going to gain a lot of market share in the next 10 years.
- [49:13]
Lex Fridman
Yeah, I think people don't realize how much, like with traditional car manufacturing, the brands are in the car assembly business. And, you know, you're going to car Costco and you're filling your cars with, we need a backup, you know, proximity sensor and we need a rear view mirror adjustment. We're spending a lot of time on a side mirror adjustment system and things like that. They're kind of assembling all these components from sub manufacturers that are not integrated. I think your point is, like in a traditional car, the backup beeping sensor boom can only beep. That's the only thing it does. And so you can't integrate it with the rest of the car and the systems because of that approach. Whereas you think the integrated approach is.
- [49:51]
RJ Scaringe
Just definitely going to win. What's also, it's an interesting moment in time because if you were to look, if you were to build a spreadsheet that looked at all the features of a classic network architecture versus us, you'd say this actually looks like you have all the same features, but then you look at how they're created. And in our case, we have, on a relative scale, a small team of software developers that control everything. And we can make Changes very quickly. Like if you have an idea, I can put it into the next release and it'll be in the car in two weeks. And you compare that with something that's very, very requirements driven. So the manufacturer writes requirements sometimes years out. And it may be things like the seat will have six presets and that gets hardwired in. And it's very hard to change things or evolve things. And so I think that again, it's a moment in time where there's a huge amount of effort to try to have a similar set of features to what we do, but then they can't do cool things like we do. So we have a couple of things we do every year that try to flex this into a very easy to understand consumer.
- [50:52]
Lex Fridman
Well, the simplest version of it, sorry, you're probably going to say some things like this, but the simplest version I see is I'll walk out to the car and it unlocks based on phone proximity and adjusts the seats to me again based on knowing that it's me. Yeah, it's minor, but it's really nice.
- [51:05]
RJ Scaringe
But that's like 15 different companies. So you have to do H vac adjustment, seat adjustment, lighting adjustment, exterior sound, interior sound, vehicle state, battery state. But here's where it's exciting is we can do something like a Halloween mode. So we do one every year and we the cars get dressed up. So last year we had a Knight Rider version, we had a Back to the Future version. And when you walk up to your car and it's in that mode, it does a totally different sequence. It plays, you know, in the case of Knight Rider, like the light on the front goes beautiful. You get in the inside of the car, there's different lighting sequences, different sounds, and it's just for Halloween. But that's only possible if you control all those systems. And so we're at like the tip of the iceberg of being able to do these very immersive, very hard to recreate modes that are just, you can't do with a traditional architecture.
- [51:54]
Lex Fridman
Nightrider was early to having a lot.
- [51:56]
RJ Scaringe
Of on device compute.
- [51:58]
Lex Fridman
You know, the value of a zonal architecture and systems integration. Yeah, speaking of, actually Knight Rider, there was a Pontiac Trans Am. If I recall, cars used to have cool names. You know, you have Mustang, Thunderbirds, all these really nice names. El Dorado, very evocative. Now we have R1, R2, model 3x5.
- [52:19]
RJ Scaringe
Why?
- [52:19]
Lex Fridman
Yeah, have we lost a bit of the sense of adventure? Could we have some?
- [52:23]
RJ Scaringe
Yeah, could we? It's A good question. Evocative alphanumeric versus name. We decided to keep things simple and we had like a very. It's a structured name. Okay.
- [52:32]
Lex Fridman
But you're a surprise and delight company.
- [52:33]
RJ Scaringe
Yeah.
- [52:34]
Lex Fridman
You have a Knight Rider mode.
- [52:35]
RJ Scaringe
Yeah. I don't know, maybe we should something.
- [52:36]
Lex Fridman
That'S befitting of an R1 or.
- [52:39]
RJ Scaringe
We had those debates.
- [52:40]
Lex Fridman
You guys are super outdoorsy, you know, like Apple has the Apple's taken the, you know, Yosemite El Capitan naming scheme. But again, I feel like there's a lot.
- [52:47]
RJ Scaringe
All of our cars end up having code names.
- [52:49]
Lex Fridman
We just use the code names of the publication.
- [52:50]
RJ Scaringe
The code name for gen 2 is Peregrine.
- [52:52]
Lex Fridman
Okay.
- [52:52]
RJ Scaringe
Which is beautiful. Super fast following Rivian.
- [52:54]
Lex Fridman
Would anyone hear about Rivian Peregrine? Yeah, I've seen a lot of takers, but you'll laugh.
- [52:58]
RJ Scaringe
So the first batch of prototypes we built and the first one of those vehicles of our one like production representative, the code name for that vehicle is Avocado. So that may be less than a saleable name.
- [53:10]
Lex Fridman
Okay, I can help you with the names again, back to the bike shed. I'm happy to advise on that. As we talk about the Rivian business, think about a car business 20 years ago, say a U.S. manufacturer. I think regard as a hard business, capital intensive. It's about getting to a significant number of units and market share being disciplined on price generally. Trying to move people up from sedans to higher value vehicles is at least for the American carmakers, how they tried to improve the business. How should investors think about what the Rivian business is today where there's like again the classic car business where you're trying to sell a lot of units, trying to get volumes up and you do other stuff like you do insurance and lending and things like that. But carmakers have always done that to drive more volumes. Or is the business actually very meaningfully different from a classic automaker where you have this big Volkswagen deal that sounds like a capital light more kind of B2B side of the business. Obviously people talk about how ride sharing is going to change things. And so I'm curious how much Rivian is a classic car business done? Well, where it's a P times Q. We're going to sell a lot of cars and make a margin on them and you have the volume ramp versus how much is it? Oh, this is actually a totally different kind of business. And we're going to rent cars to people in the ride sharing world or we're going to sell software to other car manufacturers.
- [54:26]
RJ Scaringe
I would characterize. There's three parts of the business. There's the volume of selling vehicles and the revenue associated with that is quite large. And so even with each, today it's $90,000 for every sale, but let's say it's half that. It still just creates a tremendous amount of revenue and helps cover a lot of the fixed costs and allows you to develop technology that even if you're a small percentage that is being put back into technology, allows you to develop a lot of technology. And so that's core to what we believe. And we built a brand that people really love. So we just need to get to higher volume products to pull that together.
- [55:04]
Lex Fridman
200,000 R2s times 50k per is 10 billion of revenue.
- [55:11]
RJ Scaringe
So you can really start to move up the revenue. So this year, last year I think we were just around $5 billion. So we'll see our revenue really start to climb. And so that is absolutely a part of the business. Then there is a technology sales and there's a lot of different ways to accomplish that. But Volkswagen is an example of that. But we look at what we're doing across our technology stack, there's a bunch of areas that we believe are things that can be monetized to generate incremental revenue and often revenue that has a higher margin profile than let's say the vehicle sales. Vehicle sales, let's say that comfortably sitting at, when we get to the right steady state, 20 to 25% gross margin. And then there's a third piece which is different business models. And the big, like a big unlock here is actually just being D2C being direct to consumer. And so with the exception of us and Tesla, every manufacturer today sells through distributors and distributors then on to dealers or in a few cases, direct to dealers. But nonetheless, there's a middle layer.
- [56:11]
Lex Fridman
I realize the two middle layers. It's, you have to sell to distributors who sell to dealers.
- [56:15]
RJ Scaringe
Yeah, often it's a distributor then to a dealer. Yeah, yeah. That often draws the attention to the sale saying, oh, you're paying a third party 10 to 15% to sell your car. That's true. That's a, that's a far less efficient way to sell the vehicle than having a direct relationship. But I think the unappreciated part of it is you lose out on all the business that happens post sale. And so the service business actually becomes a very large part of our revenue generation. And if you look at car dealers and you look at, if you know any friends that own large car dealer groups, they're very profitable and where they make the vast majority of their money is on profit or is on service. The thing you have to wrap your head around though is in the beginning, service is not a profitable business because you have to build. So you're investing a very large amount in building capacity, so service infrastructure. So a ton of capex, you have to build up a large team of service technicians. This is a huge spend category for us. And in the beginning we're just performing warranty work. We don't really have paying service yet. But it suddenly goes from this thing that's a cost center and either feeding in as part of a warranty expense or feeding in as opex. And suddenly it becomes cogs and profitable cogs. And it's like an afterburner to the business because the car park's then at a big enough size where it just is throwing off billions of dollars a year in profit. That's the goal. The services side of the business we think of as quite interesting in the long term and something we protect. And it becomes even more valuable if you have a software defined vehicle because there's an opportunity. For me, think about used car. Used car is another huge profit center for the auto industry. And today when you buy a used car, if you're a business that's buying used car to then sell to someone else, you buy it, you maybe do a 25 point check, maybe you do a 50 point check, you maybe put some sort of a stamp on it to say this has got an additional warranty. But you actually have very little understanding of what the health of the asset is. And you have no ability to incrementally improve the asset with software improvements. With a software update, we control that. So we can decide, hey, we want to invest in making software improvements that will raise the value of these vehicles. And so it allows us to play very differently in the residual space and uniquely control and have profit built into the used car business. And so that's another arm that today represents a very small percentage of our revenue. There's not a lot of used Rivian transactions. We participate in a lot of them, we buy Rivians and sell them. But in the long term it becomes a very big business.
- [58:44]
Lex Fridman
Okay, this is very interesting. So a traditional carmaker part of why the business is hard is because you sell units and the business is kind of the least recurring revenue you can imagine.
- [58:53]
RJ Scaringe
All the recurring revenue somebody else gets.
- [58:54]
Lex Fridman
Exactly. And so you're always in this dog fight for next year's sales and the following year Sales. Whereas your sales, you guys are selling cars and then you have monthly active users of Rivian out there and you have a direct service business because you have direct relationships with those customers. And that really changes the direct business. Never mind the B2B stuff.
- [59:13]
RJ Scaringe
It depends on the brand. But if you look at it like this, for every car that's sold, whatever its sale price is, the revenue over the life of the vehicle is generally around 2 to 2.5x the initial sale price. But that's over a 15 year time frame. Sure. But there's a lot of revenue that happens associated with the vehicle. And of course we know that. Think of all the businesses that exist that are not car companies, that exist because of those cars.
- [59:37]
Lex Fridman
Yes.
- [59:38]
RJ Scaringe
And so some of those businesses have less relevance in an EV world. So like, do I need to go to a Jiffy Lube to change the oil? No. But would I like to have a software update? Sure, definitely.
- [59:52]
Lex Fridman
And so you see, that's still a complex thing. There's still components that wear out.
- [59:55]
RJ Scaringe
For sure. Yeah, yeah.
- [59:56]
Lex Fridman
My car is telling me to rotate my tires. Why do I need to rotate my tires in.
- [60:00]
RJ Scaringe
Because we were biasing the front tires when you're in efficiency mode. So it's wearing the front tires a little bit faster than the rear tires.
- [60:06]
Lex Fridman
And in Gen 2, it does this dynamically.
- [60:08]
RJ Scaringe
Gen 2 on the. Yeah, it can. If you put it in what we call like the all purpose mode. And then R2 is even better. It's like. Because the way we do it dynamically is we discover. I'll go deep here for just one second. So it's a permanent magnet motor. And so a permanent magnet motor generates losses if it's spinning. And so what we do is when you're in conserve mode, it automatically disconnects the rear axle. And so you're using just the front. On Gen 2, with the dual motor, we dynamically disconnect the rear axle even in all purpose. And then on R2 we've been able to make it so fast that it connects and disconnects. You don't even know it's happening. So you can be in like. It feels really sporty because when you hit the throttle, it all very, very quickly connects everything. But you have to mechanically disconnect the rear axle. So you think of like, well, can you do that in 140 milliseconds? It's a lot of stuff to do in 140 milliseconds.
- [60:59]
Lex Fridman
And going to two wheel drive rather than four wheel drive, that's an efficiency Thing.
- [61:02]
RJ Scaringe
Yeah. Because you turn off, you completely disconnect the motors. There's no losses associated with spinning the permanent magnet machine.
- [61:09]
Lex Fridman
Interesting. Okay, the big EV story of the past 10 years has been the China ramp up. Why? Obviously, the Chinese EVs are cheap, but they're also very good. How did China get so good at EVs? And how does the US continue to be competitive with the global market? Obviously, the US is protected as a market, but as you think about winning.
- [61:30]
RJ Scaringe
Other countries, say Europe, Europe in particular. Yeah. So there's two pieces of the China story, and I think the one that we often are drawn to is the cost story. And so we've. One of the things about the car industry, I should say, is everybody buys everybody's products. And so you can go buy. If it's on the market, you can buy it. So everyone buys them, they take them apart.
- [61:54]
Lex Fridman
This is Jim farley talking about EVs.
- [61:56]
RJ Scaringe
On the mechanical side. There's nothing that's not seen by every car company. So when we launch R2, everyone will have it, everyone will take it apart. And so what I call out is that there's not any magic that's happening in terms of how Chinese vehicles are manufactured, the processes, the design of the components and the systems. There's high degrees of overlap in some cases, very similar to what we have here in the US and very similar to some of the things we've done within Rivian, or if you look at comparisons to Tesla or others, obviously they've clean designs, so they have great part consolidation, part elimination, but there's not something that you look at and say, oh, aha. This is how the cost has come down. And what's driven the cost down is that you have the compounding benefits of much lower cost of capital across not just the manufacturer, the vehicle manufacturer, but across their whole supply chain. So tier one, tier two, tier three, everyone's cost of capital is in some cases zero or close to zero, because a lot of that capital is provided by the local government and in some of the larger companies, cases, the federal government or their equivalent of the federal government. And then you have a much lower cost of labor. And you put those two things together and you multiply it across every part, all the way down to every source of raw material. It just ends up with a cost structure that's very advantaged relative to the Western world, particularly very advantaged relative to the United States. And that doesn't translate. Meaning if you took a car, the exact same design of a Chinese car, and you produce that in the United states, with a U.S. or North American centric supply chain, the North American cost of capital, the cost would be much, much higher. And so we're seeing tariffs in the case of the United States go in place that try to take up the fact that there's difference in labor costs, difference in cost of capital. I think Europe's a big question as to what happens there. But depending on how those regulations fall, if there's zero trade friction between, let's say a country and China, more production in the fullness of time will shift to China because of those advantages, the lower cost of labor, lower cost of capital.
- [63:58]
Lex Fridman
But is it not just the cost structure? Because you've seen the videos of like the Nio, the champagne glasses on the.
- [64:04]
RJ Scaringe
So that's the second part. Okay, so that's the second thing. This is the part that I think. I'm glad you touched on this. This is the part that's actually should be, everyone should be taking note of which is the technology is very good and this isn't true for every Chinese car, but there's a subset of the Chinese brands that have done an excellent job in many ways for the same reasons that Tesla and Rivian have developed our technology different than the incumbent manufacturers, that they started with a clean sheet. And so if you start with a clean sheet, you're going to naturally look at this and say, well, let's design a zonal architecture with as few computers as possible. Let's own our software stack. We're not going to have it owned by 50 different suppliers and abstracted across 100 different islands of software. Let's vertically integrate around the high voltage systems. And so you end up making a lot of these very logical first principles, decisions in ways that are not too different from a Rivian or not too different from a Tesla. And then because the cost structure is such, they've been able to say yes in some cases to more features. They haven't done the powered even they.
- [65:08]
Lex Fridman
Have not done the powered features.
- [65:09]
RJ Scaringe
They haven't done that yet. But they were adjust. But you can see they've said yes to more features. Yes, because they have the cost structure to enable that. And so I think that if I'm a western manufacturer, that would be my biggest concern is you have to very rapidly go build a software centric architecture and a world class electronic stack to support that is sort of obvious, but I think that's a big part of what led to our deal with Volkswagen is we offered an ability to very quickly get to that state. I Think every manufacturer should be thinking how do we build a software defined architecture as quickly as possible? Because to be competitive on a world stage you need to have it.
- [65:53]
Lex Fridman
Yeah, I can buy that. How do you feel about the phasing out of the EV credits?
- [65:57]
RJ Scaringe
In the short term I think it's going to create some headwinds in the longer term. I say this often, I think it's hard to argue that it's going to create. It absolutely creates a less competitive environment. So you see a lot of manufacturers pulling back on their investment. I think that's myopically. It's good for Rivian, but I say it's bad for the world, it's bad for my kids. I think it's unfortunate that we don't have more companies leaning in. The best measure of I think the under investment in this space and the lack of highly compelling product is that there's a single company. Tesla still has around 50% market share in the United States. The United States is at around 8% of new vehicle sales are electric.
- [66:38]
Lex Fridman
Isn't that a surprisingly low number? 8% of new vehicle sales in the United States are electric. If you're to ask someone to guess, I feel like they guess a higher number.
- [66:45]
RJ Scaringe
But here's why. If our products Rivian sells say is that our flagship products, I really believe that R2 will expand the 8% but if I'm buying a vehicle and I want to spend less than 50k, there's hundreds of choices in the ice world. I can get sedans, I can get coupes, I can get convertibles, I can get minivans, I can get crossovers, SUVs, all these different choices, different brands. If I'm buying an ev, I think there's truly. I think there's.
- [67:14]
Lex Fridman
I think it's just a choice issue.
- [67:15]
RJ Scaringe
I think there's two choices. I think it's a Model 3 or Model Y today. And everything else is there's maybe a few that are getting close but everyone.
- [67:21]
Lex Fridman
Clearly likes crossovers judging from the internal combustion world.
- [67:24]
RJ Scaringe
And so I think for the first time with R2 you'll have a choice that's highly compelling. That's not a Tesla. That's not to say anything negative about Tesla.
- [67:32]
Lex Fridman
So we are taking the first affordable crossover.
- [67:35]
RJ Scaringe
There's others. I'd say it's going to be the first highly compelling, affordable. That's the key part is you need it to be highly compelling. And so there's been a number of EV launches from existing manufacturers that haven't gone that well. And I think we get the causality backwards. We say, oh, it's because customers aren't ready to buy EVs. And I say, well, no, it's because the EV is not that compelling. And the existence proof that if you have a highly compelling product, there will be a lot of demand is Tesla. The Model Y is one of the best selling cars in the world. Now we just need to have 20 other choices in order to start to really make a dent. And then if you compare this to China, where there is a lot more choice, China is a three and a half, four times higher level of adoption of electrification than we are. And the slope of the curve is very steep. We have a choice issue here in the US is our view.
- [68:20]
Lex Fridman
Last question. Where did the name Rivian come from?
- [68:23]
RJ Scaringe
So Rivian is a play on words. I grew up on the Indian river, and so it's the first three letters of the word river and the last three letters of the word Indian. I, A, N. We like, came up with all these different ideas, but we wanted something that was phonetically smooth and nice but didn't mean anything in any language. Now it's so natural, but when we first started using, it was like, is that right?
- [68:42]
Lex Fridman
It's a great name. It sounds really nice.
- [68:43]
RJ Scaringe
Yeah.
- [68:44]
Lex Fridman
All right, well, rj, thank you.
- [68:45]
RJ Scaringe
Thanks.
- [68:45]
Lex Fridman
Super fun.
- [68:46]
RJ Scaringe
Yeah, it's fun.