Okay, here we go. RJ Scaringe doesn't need really any introduction, CEO of Rivian. Thanks for joining us.
Yeah, of course.
Thanks for being here.
Happy to be here.
I was just gonna say to you privately, but I might as well just say it here: I love going on the test drives and sitting in the back seat with people that have never experienced it before.
Yeah.
Just seeing their reaction, not just to the performance, but, you know, just the quality.
Yeah.
The quality of the vehicle and the suspension and the materials you use. It's super, super interesting. Well, thanks for joining. Wanted to give you just a chance at the top to kind of tell us what's key messages for the audience. You asked me-
Yeah
How was my summer, what I get up to, and I'm talking about my vacation. I don't think you had as much time off.
Yeah.
You're kind of cranking.
Yeah.
Yeah.
Yeah.
Good. You signed up for this stuff, so-
Yeah, that was the idea.
Yeah. But so key messages, and then we'll get into a good discussion. And please, guys, tell me... You know, participate. So your questions are better than mine, but RJ, please.
Yes, it's, as you said, it's been a busy summer, a busy year for us. And we've got a few big objectives, which all ladder into, I'm sure some of the core questions you're thinking about. But number one, we continue to focus on ramping production. And the ramping of production for our first set of products are our flagship consumer products, the R1T, R1S, which Adam is referring to, along with our commercial vans. That's really critical to achieve the level of fixed cost leverage, or fixed cost absorption we need, on the pathway to profitability in our Normal facility, our first production plant.
That's playing out as we simultaneously are working towards a big set of updates that we're rolling into the vehicle in the first half of next year, that not only provide a number of cost savings, but create a number of enhanced features for consumers, and also lays the technical foundation for what will go into our R2 product line. And that's really the third big area, is we've been really focused on making sure we leverage all the learnings from the ramp-up of our first set of products into this next platform, what we call R2. And just defining the characteristics of it, how it's built, the sequence how it's built, how we consolidate parts, how we simplify the architecture, has been a huge focus for the last year, and we're really seeing that culminate.
We'll be showing that product publicly early part of next year. So, we're in that, like, final sort of phase of, like, what we show is what we're going to build.
Yeah.
There's a lot of a lot of work happening on that as well. It's a fun part of a program because we're working on something so much, and we've shown nothing. It's, we're excited to reduce the asymmetry, so to speak, of, of what we know we're doing and what the world has yet to see that we're doing.
RJ, you've been through production hell.
Mm-hmm, and supply chain, too.
Supply chain hell. Yeah, and pricing hell-
Yeah
... and other hell. These are learning experiences, and, you know, you still have a strong balance sheet. You still got-
Yep
the time to make the decisions to become a self-funding company at some point. But how would you compare your level, maybe anxiety is the wrong word, but you know, real concern about being on top of the issues that were coming up as sometimes faster than you were solving them?
Yeah.
Kind of where, where are you now on kind of that line of scrimmage? And when you compare it to how, let's say, a year ago-
Yeah
-which was pretty gnarly.
Yeah. I think the way I'd look at this is there's three big questions-
Mm-hmm
that we think about, and I would certainly think all of you think about, in determining sort of like where our anxiety level should be. Number one, and the most important, is: Does our brand and the ultimate, ultimately, the set of products we build, does it resonate with consumers? So it—are we building something that people want? And if the answer to that is no, then I should get off the stage and, you know, there's no point to talk about the other categories, but that's really foundational. And the objective of R1 as a platform was to establish the brand, really open that brand umbrella, if you will, and, you know, create the platform, the brand platform by which we'd introduce other products over time. And in that regard, it's been enormously successful. So the...
We continue to see very strong demand backlog. We have market-leading residual values. If you buy a R1S today, you can sell it tomorrow for more than you bought it for. You know, if you order an R1S today, you're not going to get it till the end of next year. That's a problem, but a high-class problem. It's a good problem to have. And what we've found is the level of resonance we're achieving with the brand philosophy and the brand vision, and as that then manifests into the product features and the trade-offs we've made, it's, it's been very, very strong.
Mm-hmm.
And so, of course, the question then becomes: How do we make sure we repeat that with R2 at a lower price point? But so first and foremost, we're very pleased with where that is. We continue to invest in that. We're going to be ramping up awareness through additional physical locations, and making the products more accessible to future customers. The second-
You're going to have one in Laguna Beach, I think, too.
Yeah, right around the corner. Yeah.
All right.
Yeah.
All right.
That's going to be cool.
Second one.
Next year, we'll do something there.
Great.
So the next question is: If provided we've done a really good job of creating a product portfolio that customers want and a brand that's interesting and clear and holistically understood, we then need to be able to produce products. So the ability to ramp, and this is what you're referring to in your question, Adam, is such an important part of this. And we've built a large production facility just south of Chicago, in Normal, Illinois, and we've had a lot of challenges ramping that. There were supply chain challenges. We launched three products at the same time. And when you think about launching a product, there's thousands of components that need to come together from hundreds of suppliers.
They need to come together in perfect sequence, meaning, if you're missing one part, the vehicles can't be built. If you are missing one part, you have, you know, inventory accumulation issues, so it's, it has to be really well synchronized. And so doing one launch is difficult. Doing three launches is really difficult. Doing three launches where you're establishing and setting up the plant, you know, where we started this in February, March of 2020, you know, with a backdrop of work from home pandemic, it was extremely challenging. Layering on top of that then some of the supply chain challenges, which I've talked about a lot in 2022, it just made for a more complex and more difficult ramp than certainly what we envisioned. Now we've learned a lot from that.
Sort of chief among those learnings is to not launch three products at the same time in the same plant. But we've come out of it, and we're now. We've rounded the corner. You're starting to see that in our numbers. The production numbers are going up, and along with that, our cost structure is dropping very quickly, largely because we're running the facility. The manufacturing doesn't work well if you have these enormous fixed costs and you're producing at a fraction of the intended output. So the fixed cost absorption we're now seeing was one of the core drivers to. We saw Q1 to Q2, $35,000 gross margin improvement. We're gonna continue to see similar level improvements as we go through the path to very strong, positive gross margins for R1.
Then the third thing that I'd be thinking about is, is there a pathway for growth? So provided we have a product that people want, and provided we have, you know, an ability to ramp production and operate the plant in a strong, positive gross margin way, can we replicate some of those successes into a broader, larger market and a bigger, larger addressable market with our R2 product line? And in that regard, of course, I'm very bullish on it. I think we're set up extremely well. The way that product is coming together looks great, but that's been a core focus for a lot of our team, and it's something we're beyond excited to show as we get into early next year.
Okay, let me launch off the point three, the growth. You don't share order book information anymore, and I totally respect that. So how can you convey, you know, where demand is? What can you say that indicative of-
Yeah.
as you have more vehicles in the wild, you don't advertise, and so you get the word of mouth.
Yeah.
You obviously gave the wait time.
Yeah.
That's fine. I don't know if that's moving out or some other indication of-
Yeah
... of that on the forward demand.
I mean, we just as you're probably looking for metrics for our demand, we also look for metrics for our demand. And it's, you know, reservation backlog is a bit of a challenging one because you essentially end up with glass ceilings, where once you build up enough of a line, it's sort of a deterrent for additional orders. So if someone in here said, "Boy, I really like the vehicle. I saw it out front. I'd like to buy one. When can I get it?" And I say, "Well, early 2025." They say, "Okay, fine, I'm gonna buy something else." So that's... It's easy to misinterpret that number.
Mm-hmm.
What we actually find to be really useful is looking at the used marketplace, and the used marketplace represents-
Yeah
A healthy check of pricing. So if vehicles are selling used for more than our MSRP, that means there may be a little bit of pricing room. It also means there's a little bit of, You know, the demand is exceeding supply, which is good. And then we look at not just, like, a 1-month-old, because there are people that still buy a Rivian and sell it the next day to make a little bit of money. It's frustrating, but it happens. It's actually useful for us to watch and observe that. But actually, what does a 6-month-old Rivian sell for? What does a 12-month-old Rivian sell for?
Mm.
That is a useful metric for us to gauge sustained robustness of demand. It's also very valuable for us to unlock leasing, which we haven't done yet. We intentionally held off on offering a leasing product because we had a lot of confidence in residual value, and we wanted to prove that, with market-leading measured residual values that'll lead to exceptionally strong leasing packages.
Plus, the IRA application on leasing-
Works great for leasing, yeah.
It does work.
Yeah.
There are ways to do it.
$7,500 does work with leasing.
Yeah.
Yeah.
Okay.
So it creates, we think, some really interesting ways to access more price-sensitive customers.
Mm.
But that was an extension to your question, but the core of it is we really look a lot at how the vehicle is doing in the marketplace.
I'd like to go to the business model. I mean, one of the things that really differentiates you is your level of vertical integration.
Mm-hmm.
It really, really stands out, you know, in the ex-Tesla-
Yeah
... EV startup community. You do your own OS. That's really differentiates you. It's hard, you know?
Yeah.
You do your own inference chip designs.
Yep.
You know, tell us again, for this audience that may not be as familiar, like, why is doing your own OS important?
Yeah.
I have a few other questions as to some other things that you do-
Yeah
internally that are perhaps an advantage.
Yeah.
Yeah.
So if we look at what's gonna play out over the next 15 years, we're gonna witness the vast majority of vehicles being electrified. That means, you know, motor, motor supply is going to grow, battery supply is going to grow, and in our view, that, that will cease to be a differentiation point. Being electric will just be the way. And I'd say, while we talk about electrification as, as sort of a core of, of, of the change that's occurring, I think the much more challenging and much more difficult transition we're gonna witness is the role of software and electronics in the vehicle. And it's much harder to copy a robust, highly integrated, holistically designed software and electronics ecosystem within the vehicle, largely because of how that's been done to date.
So with the exception of ourselves and Tesla, as Adam said, legacy OEMs or the incumbent OEMs leverage a variety of Tier 1 suppliers to provide computers that go into the vehicle. May hear them called ECUs, electronic control units, that have some specific set of functions or some domain that they control. And that sort of set of ECUs, which are outsourced to third parties, and along with those outsourced ECUs, the software stack that sits on them makes it very challenging to cross-leverage compute, to cross-leverage any sort of perception if it's associated with self-driving. And more so, it makes it really hard to do updates. It makes it really hard to improve the software stack.
We took the view, we have the strong conviction that owning the entirety of the electronic stack, so all the computers in the vehicle, and then owning the software stack that sits on them, creates an opportunity to have a much more holistic customer experience, user interface experience, which is great from a demand point of view, and you'll feel it in the vehicle. It's not to say that vehicles may have different features, but the way those features are executed and the way they come across in the vehicle feels much more seamless. But above and beyond creating a seamless digital experience, it also represents an opportunity to have very regular and meaningful updates to the software stack.
A lot of OEMs, a lot of vehicles will claim to have the ability to do an over-the-air update, but they're relatively infrequent, and they tend to be very surface-level. Like, you know, the clock moves on the screen or the color changes on the screen, but they're not fundamentally changing the way the dynamics are in the vehicle or the way the sensor stack is being utilized. We do an over-the-air update every 3-4 weeks, and they're meaningful, and it's in response to features we see coming in. It's in response to what we see happening at a macro level across how consumers are interacting with products and electronics and technology. So the third area that's really important is cost.
When you control the software stack and the electronic stack, you can consolidate compute platforms to a much smaller number of computers within the vehicle, and that's really hard to do if you're trying to work with multiple third parties to consolidate features that may today exist across different ECUs. And so I think this is actually going to be the hardest thing for incumbent players to replicate, in part because the organizations that they've built weren't designed for this. So they have organizations that are designed to procure features or procure ECUs, versus organizations that design computers and develop software to run on those computers. And so from the very beginning, that's just how we were built. We don't. We're not a procurement organization around technology. We go build it.
So it's so embedded into our culture and how we operate that allows us to move very fast, and we think that's going to represent a significant differentiator long term, both on the cost side and the consumer feature side.
RJ, a CEO of an EV startup recently told me... And the CEO can make a mean jackfruit taco. Just gonna leave it at that, you know?
Okay.
He told me recently, that when it comes to ADAS, and developing and training your own vision-based, autonomous system, that you either wanna do it all in-house, the way Tesla does, full stack, and try to do everything internally, or get someone else to do it all, as much of that as possible. And at the in-between route of kind of like having your own OS and then working with someone else, that's that gets complicated.
Mm-hmm.
You currently use an outside supplier-
Yep.
for that, but you're public with making efforts to try to-
Yep.
Look for ways to do it yourself.
Yep.
You might even have a side by side for a while, and-
Yep. Yep.
Okay. Tell us about that journey. Why are you doing that?
Yeah.
Why is it so important for you to take control of that, rather than just trust the system, which a lot of my clients say is more than good enough?
Yeah.
and getting better all the time?
Yeah. So as we think about the world of self-driving, there's a few aspects to consider here, so... And there's lots of different sort of levels of self-driving, and there's, frankly, a ton of confusion around what those levels represent. But ultimately, you have a set of sensors that have the objective of perceiving the world, such a perception stack, and the efficacy through which those sensors perceive the world unlock or enable the ability of the vehicle to control and make decisions in a safe manner, brake, steer, accelerate, and so on.
The way a lot of the systems were architected in the beginning, including our Gen 1 architecture, was to have some of those sensors be sensors that are directly controlled by the integrator or the developer, in this case, us, but then a few select sensors come from a third party. The challenge when you have these sort of mixed systems is it leads to a very late fusion of your perception stack. So if you have a front-facing camera that's, let's say, not providing raw data, but rather providing object classification and vector association, so it says, "This thing is a car.
This thing is a bike, and it's got an XYZ vector, you know, of velocity and acceleration of this. You don't get to take the raw information and merge it with all the other cameras in the vehicle to create a better, more accurate, aggregate view of what the world is. And it'd be like, to make it easier to visualize, if you think of how we as humans process information, my eyes, my ears, my sense of smell, they're. I have a really early fusion of all that information. I don't, like, see something, make a determination as to what it is, do that whole analysis, and then my ears hear something, make a determination what it is, and once they're both determined, try to match those. At the very beginning, I merge that mainstream and in my brain, of course.
So in a vehicle, you want to be doing the same thing. You want a really early perception fusion, such that you can get the highest level of accuracy in terms of how the world is perceived by cross-leveraging, particularly in the world of ML, cross-leveraging all the insights that may be gained from each sensor. So our view is to do that well, and to be able to do that iteratively and learning over time, it's really important to control the entirety of the perception stack, and then to control the entirety of the compute stack and the control stack.
What we're going to be launching here shortly is any sensor that wasn't our own is now becoming our own, and it's a part of a broader initiative to make sure that we believe self-driving over time, over the next 10 years, will become a feature that is important to customers in their purchase decision. I'd say today, it's not the primary driver. If you look at folks that are buying Rivians, they're not buying or not buying because of our self-driving feature set. It's a sort of a nice-to-have. We think, in particular with R2, and as we move into the second half of this decade, that becomes a more important consideration. We've invested heavily. We have a very strong ML team that's...
perception team that's been built to ensure we have a what we think of as best-in-the-world solution in terms of perception stack, the way the perception's run, and you referenced it, but the compute stack that supports that.
Does Amazon or AWS help with that effort?
AWS helps in terms of, using it to train our system.
Okay.
So-
So they do the neural net training?
Uh, well-
You use them, or-
Well, we use their-
Use some of their-
We use some of their compute platforms, yeah.
Okay.
Yeah, in the cloud. And so that's another big question, you-
Yeah.
For us to—
On the training side.
Yeah, we saw your report, I mean-
Okay.
How things are-
I'm sorry.
No, but how... You can train, you can have an on-site, on-prem training, you can use cloud. Ultimately, there's a question of cost as to which you do, and I think long term, you'll see people use a hybrid approach. Use sort of base load, you'll have something that's an on-prem-
Mm-hmm
... stack, and then you'll have some cloud to do some peak, peak usage.
Mm-hmm. Questions for RJ? Just wait. If you don't mind waiting for the mic, we can get it to you. Thanks.
So RJ, in the context of ownership as your customers, you know, as you have, you know, obviously, you have high-end customers right now because they're enthusiasts and whatnot. As we move down the price scale with R2 and whatnot-
Yeah
... where they become, they'd be more price sensitive to total cost ownership. Obviously, charging is cheaper than gasoline, especially out here.
Yeah.
But like, things like insurance and tire wear, with, given the weight of the thing and given the higher torque, how do you think about managing those? Or, you see anecdotal evidence of people that buy one, they go: "I had no idea insurance would cost so much. I had no idea I'd go through tires every six months," you know, et cetera. How do you... Obviously, weight begets weight, so if you get the weight down, that will help. But, like, in terms of insurance, you know, it seems like insurance companies are quick to total one if the battery gets dinged, because they just don't know how to fix it.
Mm.
You know, how do we get those costs down, and how do we make this more accessible to the mass market as we move forward?
Yeah. Well, specifically on insurance, we think, there's a lot of aspects of insurance that are important to just think about here. So we think this is something that historically has not had a lot of involvement from OEMs. So typically, the OEM produces, delivers the vehicle to a third party, to a dealer. The dealer then takes care of customer interaction. They charge 10%-15% to the OEM to sell it, and then there's another company that does the insurance. And the challenge with that is, the insurance company is then removed from a lot of the core decisions that are made in the vehicle to properly assess both repair costs and, importantly, risk. And the reason I call out risk is, historically, the risk of a vehicle getting into an accident was entirely on the shoulders of the driver.
Over time, the risk of the vehicle getting in the accident is going to increasingly be on the shoulders of the vehicle. It asymptotically be going more and more towards the vehicle. And so we launched an insurance product that's in 49, I think it's 49 states today. We're about to launch the 50th state, but it's a great product. It's a profitable part of our business, and part of that is such that we can work very closely with the back-end partner who's actually you know, you know, providing the capital to provide the insurance, to assess risk and to assess repair costs with complete symmetry of understanding, such that we can achieve the lowest possible rates.
Importantly, on the Rivian insurance product, your insurance becomes cheaper the more that you engage our self-driving feature set, because statistically, our vehicle, our vehicle drives better than the best driver in the world. You know, we don't. Our vehicle doesn't, like, look down for texts. It doesn't get distracted by nice sunsets. So we see that as a growing role for OEMs to play in the insurance world, particularly in the context of autonomy. Yeah, tires, tire wear. Well, tire wear is dependent on a lot of things, including how much you like to accelerate. So, you know, we have the data on this now. It's fun to watch. My dad has really bad tire wear on his vehicles. He drives way too quick, but-
Rotating tires is also under-- people don't, don't do it. I don't do it.
Yeah. Yeah.
But it does help.
So, with a lot of electric vehicles, I should say, they're quick and they're fun to drive, so people drive them a little bit more aggressively, and as a result, you do tend to wear the driven wheels a little bit more. But that's something that is usage-based, and what we've found thus far is that we have customers who are going 40,000 miles on a set of tires, and we have customers who are going 6,000 miles on the same set of tires, and it's just very driver dependent. And what we've messaged is to say, like, if you're driving your vehicle hard, you're going to wear out your tires more quickly.
I think from the weight point of view, as we move into products like R2, the difference in weight, you know, you're talking 500-1,000 lbs. It's not so significant that you're gonna see materially different wear characteristics. And then the last thing I'd point out is, we do a lot of work closely with the tire suppliers, on compounds and the what we call, like, the efficiency tires or the low rolling resistance tires, the compounds tend to be a little bit harder, so less sticky. So less good from a performance point of view, better from a wear and a range point of view. And so as more and more customers at lower price points select the most range-efficient tire, they may not realize it, but they're also selecting the most wear-resistant tire as well.
Mm-hmm. Sure, please.
Okay. Can I ask, as we go towards the latter half of the decade and you see autonomy become more important, how do you guys see the house view on what percentage of vehicle miles traveled, do you think from an autonomous perspective, deserve to be in a owned by consumer-run vehicle versus in more of a fleet, maybe rideshare kind of model? How do you guys see that developing over time?
That's a great question. Without necessarily putting a percent to it, our own thinking on this is, we have these kinds of debates all the time, and they're really fun, philosophical debates. And we ourselves have moved back and forth on the scale. The way I'd characterize it is we believe there's going to be the types of miles that are purely commoditized and point A to point B, and where ownership of the asset is less important and where even the features of the asset are less important. It needs to be safe, it needs to be clean, it needs to go from A to B. But that doesn't represent a majority of the miles.
And I think early on, like 10 years ago, when a lot of people were doing academic studies in this, we you know, there was wild numbers that said, "Well, 90% of the world is gonna be shared fleet." I think what we see is there's still a significant portion of the world, certainly more than more than half, not world, a significant portion of miles accumulated that'll be in things that are owned. And a lot of that's practical. I have a car seat that needs to stay in the car. I've my gym clothes are in the car. So there's there's practical reasons that customers will want to keep some level of personal ownership. But importantly, and the things, regardless of whether it's a 90-10 or 80-20 split, we we can debate those ratios.
The things that are owned will tend to be more aspirational, not necessarily in price, but in terms of it's something that you connect with. Because if you can get your commoditized transport elsewhere, the thing that you own needs to be the thing that you have energy about, you have passion about. It's like, I like the brand, I like the product. So it sort of comes back to one of the reasons we feel so strongly around building a strong brand identity, with a really crisp and well understood by us perspective on how we make product-level trade-offs, feature-level trade-offs, needs to be really, really, really strong. And then if you do that well, you can build a really strong brand over time that helps create that connection point.
Just with, coming up on time, I just want to kind of fire a couple of rapid-
Yeah
Fire things your way and just kind of get your quick bullet reaction, 'cause we're kind of out of time here.
Sounds good.
Number 1 cost opportunity for you over the next 6 months as you kind of head into the new architecture,
With the new architecture or without it?
Well, or is it the new...? What would... Give me both.
Right. For us, what we have, a big update that's coming next year with-
Mm-hmm
Hardware, I refer to it. We call this our Peregrine program. By far, the biggest is the ECU consolidation.
Sure.
I actually would say, I think this gets way less attention as a... You talk about it, which is great, but I think the ability to consolidate ECUs is not a few hundred dollars. We're talking multi-thousand dollar savings to move to, you know, a fraction of the number of computers in the car than what we've historically seen.
And then outside of that?
Outside of that, we have a number of improvements we're making in how the vehicle's processed through the plant.
Okay. And where do you test these things?
So we have, actually, funnily, a pilot facility about 15 miles from here, and we're building the vehicle I just described.
Mm-hmm.
The reason we do that, it's a big pilot facility, so it's like a mini production line. We build these in a pilot line to, one, we need a lot of them-
Mm-hmm
So for our full fleet. But that line is actually run by our operations team. So we have leaders that come from our Normal facility.
Mm-hmm
when we run pilot builds, they are based here, and they work close with our engineering teams to essentially work through any assembly process issue.
Mm-hmm
sequence issue. Frankly, it's a learning from what we went through when we launched, when we first launched, where essentially our pilot run was when we were ramping production. So now we're gonna be integrating a bunch of product updates into the plant, and the plant can't go down for months. It needs to be-
Mm-hmm
The plant's down for a couple of weeks, and we come back on at full strength. So we're working through, as we speak, all those issues such that it's a seamless, immediate sort of binary plant comes back on and it's running very quickly.
Just finally, on Georgia, kind of how, where, what, what does the site look like now, and when could we actually start to,
Site is looking cool.
Great.
We haven't put pictures out yet, but we're excited. Our team was on site there, and they were sending me pictures this week.
Yeah.
I mean, there's lots of dump trucks, lots of earth-
Right
-moving equipment.
Mm-hmm.
We're getting close to having all the grading complete. It's a big site. It's a 2,000-acre site. We're gonna build it across two phases. And then early part of next year, the building will start going up. And then along with that, that happens, as you all know, once the grading work is done and the building design is completed, it happens very quickly.
Yeah.
And then through the end of next year, we'll start to install some of the equipment.
That's awesome. RJ, thanks for spending time with us.
Yeah.
Really appreciate it.
Gladly.
All right.