Future. And to do that, we need highly compelling products, products that are deeply exciting to consumers, that deliver attributes and design characteristics, driving characteristics that are exciting and pull people out of internal combustion vehicles because they're looking at something that isn't just better for the environment, but also really enjoyable and really desirable to use every day. And so the only way for us to really realize the full potential of our vision, of course, is to also drive towards profitability. And this is something that I've talked to many of you about. We talk about in our earnings calls. We're going to spend a lot of time on that today.
With the transition from our Gen 1 to our Gen 2 vehicle, what's coming with our R2 vehicle and beyond the R3 as well is vehicles and a platform and an architecture where the level of vertical integration we've driven into the product will really see that translate to structural cost advantages. We're going to talk about how that's done. We're going to talk about our network architecture, our highly efficient ECU topology. We'll talk about that in the context of how that enables incredible software and the software to continue to get better and better over time, more and more feature-rich. But we'll also talk about what that represents for us in terms of managing our production system and our supply chain.
This focus that we talk about in vertical integration, my hope is that everyone leaves following today's discussions more deeply understanding why we've been so principled from the earliest days. From the very beginning of Rivian, we said we need to own the electronics in the vehicle. That allows us to own the way we think about the network architecture. And that importantly allows us to own not just the software stack, but allows us to own how the hardware within the vehicle evolves. And our Gen 1 to Gen 2 transition on R1 went from 17 in-house ECUs to seven. And that wouldn't have been possible in the time that we did it if we had to coordinate across 50, 60 different ECUs across a variety of tier-one suppliers.
It was possible because we built the capability from the ground up, the ability to design computers, build the software that runs on those computers, and then the ability to consolidate those computers down to a much smaller number of platforms in the vehicle that introduces thousands of dollars in cost savings, but also massively simplifies how we run the fleet, massively simplifies how we update software across the vehicle. Importantly, as we saw evidenced by our partnership with Volkswagen, creates flexibility of a platform to be applied to very different form factors. Within Rivian, and even on our Gen 1 platform, we saw this between R1T, R1S, and our commercial vans, showing the same network architecture, same topology of ECUs. That's only better with the Gen 2 platform, with this consolidation.
So we'll see our Gen 2 platform be leveraged in R2, in R3, and now that it's public across a wide range of vehicles within Volkswagen Group. Beyond the vertical integration that's enabled by the hardware, this also allows us to look at some of the really important features in a really powerful way over time. One of the big ones we'll spend time on today is autonomy. When we talk about autonomy and vehicles being able to drive themselves, there is a lot of noise around this. There's a lot of questions. A question we get asked all the time is, is this something we should be doing ourselves? This is for reasons we'll spend a lot of time on today. It's really important, we believe, to control the full stack.
So on the perception side, by controlling the entirety of the camera set, the entirety of the radar set, and then allowing ourselves to fuse that information very early, and that early fusion process embraces leveraging AI to train the system. Leveraging the fact that one of our big assets is a large fleet of vehicles, a growing fleet of vehicles, that as those vehicles are on the road, they continue to get better. We can use offline training of those platforms with very powerful compute and a really robust sensor set that's in our Gen 2 vehicles. We'll see that sensor set continue to improve to ensure that this platform is something that's really powerful. Now, beyond electronics, software, our autonomy platform, we've also taken the approach of looking at a high-voltage system.
So battery modules, battery pack, drive units, and power electronics, and are building those in-house as well. Key among them is thinking about the powertrain. When we launched, we had launched with a Quad-M otor. We developed the inverter in-house. We developed the gearbox in-house. We built the assembly in-house. We purchased the rotor and stator. About a year and a half ago, we launched a dual-motor system, one motor in the front, one motor in the back, with our own platform, something we call Enduro. We'll talk about that more later. That was a significant reduction in costs. We were able to deliver really high performance with a two-motor system. But we now have updates to our twin-motor per-axle system, so in a quad configuration and a tri-configuration. The quad is, I mean, it's silly. It's so quick.
It's 0 to 60 in under 2.5 seconds, as you heard me say in the video, less than 10.5 seconds for a quarter mile. But amazingly, it's more efficient. It's much lower cost. And the way it's been set up is it packages beautifully. And so this vertical integration across the full drive unit, so winding our motors, building our inverter, building our gearbox, assembling the whole unit. For those that go on the tour today, the plant you'll get to see that production equipment. This is something that creates really meaningful and structural cost advantages for us as well.
But ultimately, as we look across all those systems that we vertically integrate, which we'll go deep on today, what that enables is products that are highly compelling and a product set that we think will help accelerate and draw people towards electrification and products that bring a level of differentiation that we haven't yet seen in the market. So we started that with the R1T and the R1S. Those are our flagship products. The R1S is the best-selling premium large vehicle in the United States today, best-selling EV large vehicle in the United States. In California, it's the best-selling premium large vehicle, EV or non-EV. So the reaction to that product has been outstanding. It's the market-share leading product in the segment. When I say segment, I mean premium segment.
To be able to take that brand strength, the excitement we see for our products, and now translate that to much lower price points with R2, R3, we could not be more excited about that. Now, we spent a lot of time talking about our consumer products because we all drive cars on a daily basis. It's how we sort of naturally relate with the brand. But we've also taken our technology and applied it in the commercial space. And of course, we started that with a large partnership with Amazon. Amazon's also our largest shareholder. And a question I get asked all the time is, how quickly will we start to see some of the other customers beyond Amazon? And what I'd say here, which is really important to recognize, is the scale of these commercial fleets and the transition that they'll make to electrification takes time.
So Amazon today has on the order of 20,000 vehicles in their fleet that we've built. This is something that has been slower than we would have anticipated, but it's a reflection of how difficult it is to build out infrastructure and to convert standard operating procedures and your business to electrification. But we're seeing really clear driver benefits, and the full cost of running these vehicles is significantly lower than internal combustion. And the large fleets know this, and we're working those pilots to drive that. And so we're really excited about continuing to lean into our commercial business as well and starting to see more vans on the road with different colors and different logos on the side of them.
Now, both for the consumer side and the commercial side, a part of our business, which we also haven't, I think we haven't done enough to really clearly communicate how powerful this is over time, is we're a direct-to-consumer business. And so just as we've taken the tier-one supply base and the margin stacking that occurs and all the, what I would call, the massive inefficiencies of trying to coordinate across multiple tier-ones within the electronics set of the vehicle and the network architecture and software stack in the vehicle, we've also removed the margin stacking that occurs from having to pay a third party to sell our vehicles and pay a third party to service our vehicles in the form of dealers.
And so that direct-to-consumer relationship not only takes a layer of margin out that typically exists outside of Rivian, but it also allows us to create a much more connected experience with customers. And so the types of relationships we can build, the emotional connection to the brand by having complete control over what those experiences feel like is really powerful. And so we see that manifest in what we call our Rivian Spaces. These are our physical sales location. We also see it manifest in how we approach pop-ups, how we approach customer events, and the frequency of how we're now conducting those. Importantly, this also allows us to conduct service on our own. And when we think of service, all of us probably naturally think of a service like a service station with lifts like what you see in this image.
This is historically how service has been done on vehicles. It's an important part. Today, we have close to 60 service centers that we continue to build out that brick-and-mortar network. But importantly, the physical service infrastructure that we've known for so long is really being supported in a powerful way with a mobile fleet of service vans. What enables these to be so effective, where more than 70% of our service actions happen with a Mobile Service, is that the diagnostics platform built into the vehicle allows us to be highly predictive with what the vehicles need. If something's wrong in your vehicle, we know about it. We can make sure we have the parts for it. And we can then go to you, go to your house, go to your business, and take care of whatever service action is needed.
Now, today, our service network is ramping. A number of you have vehicles. So you know that in some areas, we don't have enough service infrastructure or service team members. That is growing quickly. But over time, this becomes a very powerful part of our business as well. And being able to control what is, as many of you know, within the auto industry, this is a very powerful part of the auto industry. Being able not only to control that from a customer experience, but also benefit from the installed car park over time from a profitability point of view is really important. So we look at this today. While we may think of it as a cost center long term, this becomes a profit center for us as well. And similar to that, we've also built out a charging infrastructure.
We're in the early days of deployment infrastructure, but a number of years ago, we realized how significant of a gap charging is for customers. Frankly, today, there's really one great network in the United States with the Tesla Supercharger network. We have a relationship and a partnership with them that allows our vehicles to now access that network. But we're also building our own DC fast charger. We built that actually right around the corner here in Normal. And we're now deploying it. One of the most important metrics in charging is uptime. If any of you are driving EVs, you're using charging networks, public charging networks, you'll quickly experience the frustration of going to a charging station and have it not work. So this is something we track very closely.
We actually introduced in our software stack a charging score, which relates to that. So as your map navigates through different charging locations, we rate locations based upon their efficacy. But today, there's two charging networks that operate close to 100% of the time: Tesla's network and ours. Now, our network is still building. We have just over 500 chargers deployed. But the uptime is just under 99%. So what we've built in terms of the hardware and the capability is phenomenal. And we're going to be opening that network up this summer. And one of the reasons to open it up is, of course, it drives more visibility into the brand, but also turns this into a profit center as well. And the utilization of the network is really important for driving the profitability of these charging locations. And we've been very thoughtful in how we've placed these.
We actually purchased a company, A Better Routeplanner, to help us identify where we see charging deserts. This is something that we feel is going to be a real enabler for electrification as we go forward. Now, all of this really, when we think about the suite of things that we're doing—product, our go-to-market, what I just talked about in terms of charging—helps create a community around the products we're building and helps create a level of excitement that we think is among the most important aspects long term of what we're creating as a company. If I wind the clock back, I think 10 years ago, we would be sitting in a room. We'd be talking about the brand we want to create, the types of customer experiences we want, how we hope customers are going to respond.
I remember many of those sessions, we'd say, wouldn't it be great if there'd be user groups that emerge and fan clubs that emerge and people that were enthusiastic about the products we're building, the technology we're creating. Fast forward to today, and it is remarkable the level of engagement we have on the brand, the level of enthusiasm for us as a company, for what we represent as a company. We see that. There's a very active user base. There's groups of Rivian owners that get together, go for drives. There's a group of Rivian owners that get together and participate in community events. And we see them at our locations. We see them setting up their own locations. So the level of love that we feel for our brand and for our company is phenomenal.
And to have that backdrop as we now look at our much lower price products coming with R2 and R3, we couldn't ask for a better starting point. Now, all of that is driven by our team. And one of the most important parts of my job is to make sure that we have an incredible leadership team, but importantly, a highly engaged organization. And the team across all functions, whether it's the product teams, the technology teams, our go-to-market teams, our service teams, need to be thinking about Rivian from a one-company mindset. And the alignment to our mission, the alignment to really helping to drive the shift away from fossil fuels, and importantly, helping to drive a level of excitement around the shift away from fossil fuels is born out of this team.
Hopefully, you get a chance to experience this for those that are on site today as we walk around the plant. You'll get to meet some of the technical leaders and leaders within our go-to-market business, our go-to-market side of the business. This level of commitment percolates across every part of our business. We're so proud of what we built as an organization. Now, that team and what we're going to be talking about in terms of profitability and the path to profitability, that is the driver that's going to deliver on that. The focus on driving costs down in our bill of materials, the focus on driving efficiency in our OPEX, and of course, the focus on making products that people are really excited about is born out of that team.
Now, with that said, today, you're going to get to meet some of the team. So we're going to have Wassym spend some time walking us through what we built across our software stack. He'll be followed by Vidya, who will talk through how we approach our overall electrical system design. Then we'll have James come up spend some time talking about autonomy and how we believe our approach of vertically integrating the perception stack and owning the compute stack really enables us to have features that will get meaningfully better through training over time. Kjell will walk us through our go-to-market side of the business. We'll finish with Claire. We'll have two Q&A sessions. So after Wassym, Vidya, and James talk, the three of them plus myself, we'll come up, we'll do Q&A around product and technology.
After Kjell and Claire talk, Claire and I will come up and go deeper on any questions we have around the financials of business. Thanks again, everyone, for joining us. I'll invite Wassym to come on up.
Thanks, RJ. Good morning. My name is Wassym, and I lead the software development at Rivian. People often ask me, what's different in Rivian software? Let me give you a small example. We will meet Jordan. Jordan is an R1S customer in the beautiful state of Colorado. She goes to the gym three times a week. She wakes up at 7:00 A.M., walks to the vehicle. The vehicle is configured automatically. The cabin is set to her preferred temperature. And then she can drive with her preferred music service. It's a very simple sequence. All Jordan had to do is pick up her phone. Now, let's see how we would do this in the legacy auto world. This is one example of a legacy architecture. You can see here many ECUs, likely from many different suppliers.
The simple example that I have just shown, which is just 30 seconds, we would need more than 30+ ECUs in order to do it. This is 30+ suppliers, 30+ black box software, different interfaces, different requirements, different integration, different test methodology. Now, imagine as you're developing that experience, you want to change it. You will need to coordinate again with all these suppliers. Even worse, you have shipped the vehicle, and then you have a change of heart on the overall experience. It's even more painful because you don't have OTA capability. How do we do it at Rivian? At Rivian, with the vertical integration that we have between hardware and software, we developed a much cleaner and simpler architecture. With the powerful compute that we have, we simplify the hardware architecture, move that complexity to software.
Vidya will talk later on about our zonal architecture and all the great progress that we did with Gen 2. Claire will talk about the impacts on BOM cost saving from that architecture. And around this, we developed an end-to-end integrated and connected platform with more than 80% of software which is done in-house. We orchestrate the choreography of all these features. So when RJ did not like the lighting sequence and he wanted to change that three or four times, easy, we can do it. When our customers wanted to add Kneel Mode, same thing, we could do it. When our customers wanted to have pre-scheduling, we can do it. With the platform that we have, the over-the-air updates, the agility of the in-house knowledge, we can integrate and make the vehicle better and better over time. Now, this is the features and the capabilities.
It's really important to spend some time on how we designed this. We have been extremely intentional to have an architecture which is scalable and modular. You see here some of the core foundations of that software operating system that we have created with diagnostics, security, privacy, over-the-air technology at the heart of that foundation. We created multiple abstraction layers so that we can abstract different hardware configurations. We can have different and manage the complexity of different products, different feature variants. Case in point, we have been able to migrate from a domain architecture to a zonal architecture in about two years using the same software stack. We are using today the same applications between two completely different low-level architectures. When we launched Gen 1, R1T and R1S run on the same binary.
With Gen 2, the reuse of the applications, we are now setting the foundations for our MSP architecture. With the very exciting announcement that we have with the Volkswagen Group, we can now leverage on this architecture to scale up and down into many more variants. We're very excited with the opportunity that we will have to get our electrical architecture, our hardware into many more products, many more variants across the globe. At Rivian, software is obviously at the heart of the product, but is pervasive throughout the entire company. Software is at the heart of the operations with manufacturing, supply chain, logistics. In our commercial direct-to-consumer relationship, as RJ mentioned, some figures, since we launched, we had more than 30+ OTAs with more than 500 features. What's really important is the excitement that we're creating with our customers. 96% of customers install OTAs within two weeks.
In the manufacturing world, with the powerful diagnostics architecture that we have created, we're able to improve first pass yield by more than 30%. Then in service, with the direct-to-consumer digital ecosystem that we have, we have been able to improve service lead time by more than 20%. Now, let's go deep in each of these areas. Product. Our experience, I mean, as you have seen in the introductory video with Jeff, our experience is software-defined. That's a fantastic advantage because it creates that connection that we have with our customers where we receive feedback. We improve the product over time. The product today is very, very different than the product that our customers bought two years ago. The product in two years will be also very different, will be much improved versus where we are today. This is the Drive Mode. It's probably my favorite app.
It's also the favorite application for many of our customers. This is really, it showcases the depth of the software-defined experience. With a touch of a screen, you can change the entire personality of the vehicle from your best sports car to the most capable off-road vehicle. Our deep over-the-air update experience, as you see on the screen, we can not only update the infotainment system or the graphics with this beautiful Unreal Engine designed by Jeff and team, but we can update the dynamics. We can update the suspension. We can update deep-level controls within the vehicle. And the beauty, we can receive feedback from customers. We created Sand Mode based on feedback from customers. We created Snow Mode based on feedback from customers. Very recently, you have seen it in the video, we created Launch Mode based on feedback from RJ.
We will continue improving the core DNA experience of our vehicles. Another example. The charging experience is foundational for peace of mind of our customers. We introduced the reliability score to provide a rating for each of the charging stations, whether they are Rivian charging stations or third-party charging stations. In just one month after introduction of this feature, we have seen an increase of 35% of successful charging sTOPS. Given how now with our integrated solution with Trip Planner, with navigation, we are steering customers to more reliable charging stations. Now, behind this beautiful UI and application that you see here, there's a very sophisticated data platform. And the connected data platform is at the heart of our entire software stack, as you have seen. It's probably as important as the embedded software. What this allows us is we can actually run smart workloads in the cloud.
In some cases, once they mature through machine learning, we can move them to the edge. We can arbitrate on that constantly. What it also allows us is we're building this data foundation with security, with privacy in mind. With the anonymized data, we can continue to influence the product over time. As our fleet grows, and now with the exciting collaboration that we have with the Volkswagen Group, this will provide us much more scale. This data platform will become even more important. Now, let's go through some examples of how software impacts operations. We use the same software operating system when the car moves through the line. We created a clean sheet diagnostics solution to detect misbehaviors and then ensure reliability and safety of the entire system. We use that same architecture. We use it in service, and then we use it in manufacturing.
The beauty of such approach is we can achieve meaningful efficiency improvements. Let me give you an example. This is a video that we shot actually one week ago here in GA in Normal. So you will see the vehicle moving through the line. This is the test that we have created and defined ourselves on the line. We test different aspects and features of the vehicle using that same operating system. And then here, one of the tests has failed. And the system is smart enough to detect which connection has failed. In this case, this is an RCM, so this is a safety system. It displays a picture to the operator on the connection that they need to fix.
The beauty here is instead of having that defect escape through the line and go to end of line, which would take much more time, cost much more money to disassemble the vehicle, we're able to detect that as the vehicle is moving through the line. It's a massive efficiency unlock for us, and it's a higher quality bar for our vehicles. Let's talk about commercial. With our direct-to-consumer business, we developed a very powerful digital ecosystem with our customers. And for service, we're completely changing the relationship of how our customers interact with service. With our telemetry and remote diagnostics, our service technicians don't need to physically inspect the vehicles. They can look at a website, and remotely, they can see what's wrong with the vehicle. With our over-the-air updates, we're able to fix not only software issues, but in some cases, hardware issues.
This is a massive efficiency saving. When you compare that with the legacy auto model where for any issue that you have, you need to bring the car back to service. Actually, you need to bring it back to a dealership. And with the very capable and powerful data platform that we have, with the integration of AI, we can now unlock many more capabilities. I'll walk you through an example. This actually happened to me with this is the new AI system that we are testing. So my key fob stopped working. Actually, it was weird because I can unlock the vehicle, but then I cannot start the vehicle. So I started the chat through the mobile app, and then the system determined could read remotely the telemetry and the logs within the vehicle.
It could associate the actual root cause of the problem, link that to a special batch of defective sensors from the supplier, and even more than that. It could go and read into the supply chain database, know if the part is available. Then to the digital booking system, it could identify that this is a defect that is fixable through mobile service van. It booked a service appointment automatically for me so that the mobile service van comes at home to fix the problem. And then it even suggested an additional tire rotation for me by reading the records of my service logs. This is the type of magic experiences that we will be able to unlock with the rich data platform and the integrated digital systems that we have in the company from the product to service, supply chain, and logistics.
As we have seen, we built an industry reference end-to-end software platform with scalability, modularity, and flexibility in mind. We're very excited with our partnership with the Volkswagen Group where we can take this to many more vehicles in many more places and help accelerate the consumer's shift to electrification. We built our company as a digital-first company. With the powerful data platform that we have, with all the systems that we have, AI will now unlock much bigger value from that data platform. All this is meant for a single purpose, which is continue to improve the customer experience, continue to put a smile on our customers' faces. This is our biggest satisfaction as a company. We cannot have such amazing user experience. We cannot have such amazing software without an underlying amazing hardware. With this, let me introduce Vidya, who leads our hardware team.
Thank you, Wassym. My name is Vidya Rajagopalan, and I lead the electrical hardware team here at Rivian. At Rivian, we build our key hardware in-house. We do that so we can offer our customers the features they want and do it at speed. Our hardware and software teams work very closely together, co-developing the systems, and we're able to bring our systems to market much faster than a traditional tier-one model because of that. Now, while our hardware is bespoke, we design it for our vehicles, it's actually very scalable. We have the same hardware in our SUV, the R1T, the R1S, the SUV, and the truck, and the EDV, a commercial van, which is much bigger. You would imagine, you would think it has different hardware platforms. No, we use the exact same hardware platform on all of them.
Then finally, we're able to do all of this and yet save money and really focus on cost. The reason we can do that is our hardware has no unwanted features. We're not buying something off the shelf from somebody that has features that were meant for a broad audience. We also eliminate the tier-one margin stacking. So Wassym showed you this picture earlier about how legacy OEMs typically serve up features. You saw how there's a ton of ECUs, and all these ECUs are electronic control units. Oftentimes, these legacy OEMs have somewhere between 40 to over 100 such ECUs. ECUs are little mini computers, and they have them at a very low level of granularity, perhaps one controlling the seat.
The ECU for a seat would be a little microcontroller that all it mostly does is moves the seat around, heats the seat. As we all know, a microcontroller is really a simple computer. Modern computers are capable of doing much more than that. You don't need to dedicate one to just move a seat. Our approach is very different. When we started Rivian from the get-go, we always wanted to be vertically integrated. We also were very clear that we were not going to follow that legacy philosophy of hardware for really small pieces, small tasks. Because not only did that add to mass cost, but it also created a huge complexity in software land because you have these little, little ECUs coming from multiple tier ones, and you're trying to get them to speak to each other.
Can you imagine how Wassym's job would have been? Really hard. Because anytime you needed a change, you'd talk to all these different people. So from the get-go, we were very clear we would have a much simpler architecture. So we adopted what we called a domain architecture. And a domain architecture, you have an ECU per function category. And so that's the first step we took in consolidation. So instead of one per function, it's one per function category. So for example, we had an ECU for thermal management. We have an ECU for vehicle dynamics. We have an ECU for body controls. The body control ECU controls a number of body activities. So it controls, say, the wipers. It controls lighting. It controls the door handles and closure. So it's the first step of consolidation. We aggregated a number of features.
When we did that, we had 17 ECUs, in-house ECUs, and a very limited number of vendor ECUs. We only use them in areas where we don't add a whole lot of value, and they're very limited. But by doing this already, you can see we're 17 ECUs compared to somewhere between 40-100. So that was the first step. We did that very deliberately because we wanted to get started with our software platform. The domain architecture really helped us get going. It helps software teams focus. The body control team is working on the body control ECU. The thermal team is working on the thermal ECU and got us going, and we launched our first vehicle. So when it came time to look at what to do next for our Gen 2 platform, we knew what we wanted to do.
We knew we had a big focus. We wanted to drive cost down. The one big way to do it is always integration. Integration wins in electronics. Everybody knows that. We chose to migrate to what's called a zonal architecture. The zonal architecture addresses a lot of the limitations that exist in a domain-based architecture. There are really two limitations. One is the domain-based architecture is not the most efficient when it comes to using compute. You could do better. The other really big one is it's really inefficient in terms of the way it uses wiring. We went from a model where we had 17 in-house ECUs down to seven in-house ECUs. We still have a few limited vendor ECUs where we don't add a lot of value. Let's take a look at the in-house ECUs in our Gen 2 platform.
As you can see, the blue boxes represent all the ECUs, and this little walkthrough will show you which one's which. The one highlighted right now is our most powerful computer in the vehicle. It's our autonomy and infotainment computer. It aggregates, as the name suggests, autonomy functions, which James will talk to you about later, infotainment and telematics. And then our three zonal computers that we'll be talking more about. But these control really all the general-purpose functions in the vehicle. And then, most important, in an electric vehicle, we have the battery management system. A single battery management system can handle a variety of our packs. And then finally, as you know, our vehicle has phenomenal vehicle access capabilities. You can use a key fob. You can use your phone as a key. You can use an NFC card.
These are the little ECUs, tiny but powerful, that make it all happen. And in this generation, we actually added our one Gen 2, we added ultra-wideband as another modality. And it actually will result in the customer seeing much snappier responses when it comes to vehicle access. So let me take a step back and talk about we make it sound very simple. It's like we said, we'll do 17 ECUs, and we'll do seven. It wasn't all that simple, right? We built this foundation long ago. What does it really take to build an ECU? You start at the very beginning, and you say, what does the customer need? What does the platform need? And then we have our hardware and software team sit down and break that down into really hardware and software tasks and come up with an allocation of hardware and software functionality.
Then the teams go off, my team, so that we work in parallel. My team goes off and then translates that design into a mechanical design and electrical design because all electrical items are actually enclosed in mechanical boxes. We then go simulate it both electrically and mechanically before we actually send it to production because we want to make sure it works. Then we build it. Then once you build it, you have to test it. Testing comes in many ways. You test it for functionality. You test it for robustness. The automotive is probably the harshest environment for any piece of electronics because it sees the most extreme of temperatures. You could be driving in Death Valley, or you see the most rugged you go over potholes. Our strength is off-roading, so you're rock climbing.
So you really need to have a design that's very robust. Electromagnetic compatibility, RF emissions, all goes to say that it's very complex. I make it sound trivial, but there's a lot of infrastructure and capability that we have built up and driven over the years that actually made it possible for us to do it. And then as we're building these ECUs, we're integrating them into different levels of abstraction. So we start with our lab car, which is sort of a shell of a vehicle. And we get all the electronics working in system. Then we put it in vehicles. And of course, we don't always get it right first. So then we lather, rinse, repeat, do it again, iterate. And when we have the perfect design, we send it to manufacturing. And then we focus on scaling manufacturing, scaling yields, making sure we have very minimal scrap.
That's how we got here. All this work took us many years. It was not just we often say we went from Gen 1 to Gen 2 in under 2.5 years. Well, that was made possible by all this work we did over a very long period of time. Let us talk a little bit more about what is a zonal controller, right? A zonal controller, as the name suggests, controls everything in the physical space or physical zone around it. As opposed to the previous generation, you had controls that were function controllers that were function-based. So what is the advantage of this? Let's take an example of lighting, right? In the old world, a body controller would have controlled lighting. And the wires, you have lighting all over the vehicle. You have front headlights, front headlights. You have tail lamps.
You have wires going from one ECU all the way to the back and front. It's all over. That's inefficient user wiring, as we talked about earlier. As I said, that's one of the limitations of the domain architecture. In a zonal platform, the lighting is controlled by the zonal computer that is next to it. So we have three zonal computers in our vehicle, west, east, and south, obviously based on the physical location of the vehicle. So whichever zonal you're closest to, if you're a light, that's who controls you. That means you save a lot of wiring. And by doing this, we were able to save 1.6 miles of wiring in the vehicle. And we got rid of some of the wire harnesses in the vehicle. And here to show you an example is Wassym.
It's heavy.
This is a dash harness that actually doesn't even exist in our R1 Gen 2 platform. It used to exist in our first-gen platform. But that's just it's not the full 1.6 miles. We got rid of a lot of wires. Obviously, that means mass, which always ties to range, and it means cost. The ECU consolidation that we did going from 17 to seven got us 60% fewer ECUs, 44 pounds reduced. Because every time you reduce an ECU, you have the enclosure and all the stuff around it. But all this sounds very simple, but it's actually quite complex. Because what we now did, when you move to the zonal architecture, you add a level of complexity to the software. You previously had in the body controls world, maybe just body controls.
Now you have a zonal computer doing body, thermals, chassis, all of it happening at the same time. So software complexity went up. So today, you'll be driving some R1 vehicles. And what you'll find when you drive it, you think of the chassis, the front axle, and the rear axle are being driven by different zonal computers. And so that puts really demands on sort of timing and synchronization and software complexity. And when you drive it, you'll have no idea that those were driven by two different computers. So let's take a look at another computer that's pretty key in this product. It's our autonomy computer. We showed that earlier. It was in the top right of the vehicle. Autonomy got a huge upgrade in the R1 Gen 2 platform. So first of all, the name of the game, as we've been talking about, is integration.
We talked about how we integrated all the domain controllers and do limited zonal controllers. Even in the autonomy space, we merged four different ECUs down to one. But this one computer is phenomenally powerful. So it uses dual NVIDIA DRIVE Orin processors, gives us 250 TOPS, which is a measure of its AI performance, but it's 10x as powerful as our previous AI capability of our Gen 1 platform. And all of this is there so that we can process all the rich data that comes from our sensors, which we also upgraded. But that means this is actually a very complicated board to design. It's, again, not something you do very easily. There are a ton of high-speed interfaces.
For people in the electrical design world, you'll know that running PCIe Gen 4 at 16 gbps, 11 cameras at 6 gbps, all of that is really tricky high-speed signaling. At the same time, on the same board, we have a very sensitive GPS or GNSS device that has to read 100-nanovolt-level accuracy, whereas all these high-speed interfaces pose a real noise challenge. So designing all this requires a lot of skill. It's not easy, but our engineers did do it. We talked about how we upgraded the compute, but we also upgraded the sensors. We did that for a reason. We really wanted to put the best sensing out there because we know in the AI world, it's really important that you get your data. We really want to keep this data, all the training.
We don't want to keep upgrading sensors year after year because then your data becomes stale. And so we really put high resolution, 8-meg, and mostly 8-meg and some 3-meg cameras in there. And we purpose-designed these in-house. We didn't buy them off the shelf from somebody. These are brand new sensors that have very good high dynamic range, which means you can see really well at night. They also have LED flicker mitigation. And what LED flicker mitigation does for you is helps you actually read traffic lights and signs, which is really important for autonomy, as you can imagine. It's a sort of fun fact. LED lights actually blink, and they're not always on. The human eye cannot see it, but cameras actually catch it.
Sometimes cameras can catch it at a bad time when the light is off, which is why you need technology for LED flicker mitigation. Now, not only do we put the best sensors, the best optics in there, our team works very hard to make sure we get the best images. So we go down all the way to looking at the brackets and designing them to make sure because your bracket that attaches the camera to the windshield could actually cause stray light to come in. So they actually analyze the material for the brackets and make sure we have the right kind of material in there to minimize stray light because all of those pictures are really key to how your autonomy system performs later on. Now, not only do we upgrade our cameras, we also upgraded the radar.
We have our front radar can see up to 1,000 feet now. And also, our corner radars are much improved. But most importantly, our front radar is now an imaging radar. An imaging radar can see in both dimensions, vertical and horizontal. Our previous generation and most automotive radars that you'll see there are really only able to discriminate in the horizontal axis. Now, why does that matter? It matters because you want your vehicle to know whether that's an overpass that's over your head or is that a truck in front of you. So with our Gen 2 platform, we now have that capability. Finally, another set of sensors that are pretty important in autonomy are the GNSS and the IMU. They work together to help figure out where you are in the world. And our solution can do sub-10 centimeters of accuracy significantly better.
OK, so hand in hand with the autonomy computer, in fact, they live in the same enclosure as our infotainment computer. Our infotainment computer also saw a lot of integration. Name of the game, again, integration drives BOM costs down, drives mass down. We merged three different computers in here together: our original infotainment computer, telematics computer, and the gateway. And this computer also serves up there for all of our connectivity, our LTE, Wi-Fi, Bluetooth capabilities. This is the computer that most people probably interact with the most. It's the computer that serves up your navigation, serves up music, the computer you interact with when you make phone calls. So it's a very powerful and very key computer in our system. It also is the one that serves up the displays. We can serve up to three 4K displays at HD.
But finally, most important, what this computer has is AI capability, which is quite unusual for most infotainment computers. And actually, we've had this since Gen 1. So in Gen 2, we actually upped the AI capability of our infotainment computer 3x. And what this does is it can actually process the data coming from our surround view cameras, which run at 60 frames per second. And they're actually upgraded, pretty awesome, in Gen 2. But it can actually run AI algorithms on them. And we've always had this capability in Gen 1. We have it in Gen 2. It's much improved. It's what serves up our Gear Guard functionality. And it can actually do this. And that's, I think, the biggest power of it without sucking a lot of power because it has a really low-power system that powers it all.
OK, so I made it sound like we have all these little ECUs, little islands that go off and work on their own. Actually, that's not exactly it. What we have, the power of our system, really, is that the entire vehicle operates as one single sort of distributed computer. So even though we have all these little computers, they work very closely together. Every computer knows what the other one's doing. And we have all these distributed sensors all over the vehicle. Every computer has access to them. So you don't really duplicate hardware. You don't say if you compare it to the traditional Tier 1 model, where you might buy different boxes; the boxes aren't aware of what the other boxes have. In our world, it's really one big computer. I lied. I said there were seven ECUs, but it works like one big computer.
We have, for the sake of this example, three different sensors we show off. We have the wheel speed sensor, the IMU, and the GNSS. These sensors, while they may be located, the wheel speed sensors located on the wheels, like you'd imagine, are actually available to a variety of ECUs. And they're used to do a variety of different tasks. So while this picture looks very busy, the wheel speed sensor, along with the IMU and the GNSS, is what tells you where you are located. But it's also, and so that's called localization. It's used in the autonomy world to determine where you are and help drive autonomous driving. It's used by the maps to navigate. But it's also used by the vehicle dynamics system for vehicle dynamics to figure out how much torque to apply to the vehicle.
The IMUs actually are used by our incident camera. So let's say your vehicle gets into some kind of incident, and you want to automatically start recording. Well, the IMUs detect that. And all of this functionality is happening on a variety of different ECUs. It's happening on the e-zonal controller. It's happening on the XMM. It's happening on the, sorry, the infotainment computer. We internally call it the XMM, the autonomy computer. So as you can see, it's a very integrated system. Every computer knows about everything. And so the sum of it all, really, is the net performance you get is much greater than the sum of just the parts. So where do we go from here? We talked a lot about integration. Where are we going to go for R2? The big opportunity we see in R2 is in the power electronics space.
So what we plan to do in R2 is integrate some of the discrete parts that we had in R1, so in power electronics. So we're going to integrate our onboard charger, our DC-AC, and our DC-DC into one single unit, along with a few low-voltage computers. And this will give us a 5-to-1 integration, 70% reduction in high-voltage cabling. But along with it, we'll also be able to offer a Vehicle-to-Home. Now, before I talk about the Vehicle-to-Home capability that comes along with R2, I should correct a misconception that exists about R1 today. Some people think that we're not able to offer a Vehicle-to-Home with R1, which is untrue. We have a DC capability. We have DC output, which can be exported out of the vehicle.
We will be later on, we'll be announcing this later, but we'll be offering a bi-directional charger that you can mount on the wall of your garage or wherever you park your car that you want to, that can pair up with the first-gen R1 or the new Gen 2 version and can actually be used as a vehicle-to-home charger. And we'll provide about 24 kW. Now, that's plenty for most homes. The capability we're going to bring forward with R2 is the ability to have AC output coming directly out of the charger and offer a lower power, but a cheaper 11 kW product that'll be available. And that'll be coming soon at R2 launch. So where do we go from here?
As we look forward to R2, which we're already designing, we talked about ECU consolidation, and we don't see that we're going down from seven ECUs to three or something like that. We think we're sort of in the good sweet spot already. What we'll focus on is really integration within the ECU. That's going to be our focus. We still think, actually, we have a lot of headroom in the harness. We think we can reduce a fair amount of the harness and further drive the wires down. We already have a draft version of our first harness. So this is not wishful thinking. It's real data. And how do we do that? There's a couple of different ways we'll do it. The R2 is a reduced feature, smaller vehicle.
So that size is smaller, and there are fewer features. That automatically leads to a smaller harness. But we're also working hard on optimizing it. And we've really been able to drive it down quite a bit. And then the big, big focus in R2, obviously, is cost. And we expect to drive the BOM costs down. We're actively in the midst of sourcing, and we have some good data that tells us we're on track. We're driving costs down by a variety of measures. Again, integration, as I showed you in the power electronics space, but integration within ECUs, feature simplification, also really supply chain leverage. We've come a long way. We've learned a lot. We also know that R2 has more capacity, and R2 has, there are more units that will be built. So automatically, our suppliers are offering us better pricing.
But what none of this envisioned is our agreement with the Volkswagen Group. And we expect, I'm already getting some phone calls from people, that it'll actively influence our ability to drive the BOM down further. So finally, where do we go from here with R2? We're going to continue to drive technology leadership, as Kjell showed earlier on. It's one of the top reasons why people buy our vehicles. We're going to offer much more compute, both general compute and AI compute. We're going to have the best sensors out there. We think that's very important. The best displays, Retina displays. You saw during the R2 reveal a really awesome, new, cool haptic scroll wheel. We'll have that. We'll have vehicle-to-home. So we're going to focus on features, absolutely. But as we do that, we're not going to reinvent the wheel.
The focus is we're going to leverage what we built for Gen 2. Gen 2 was not just for Gen 2. It was really a stepping stone to get us to R2. So we're going to have the same zonal architecture, not the same ECUs, but we'll leverage that concept. We're going to have very similar sensor constellation, in fact, almost the same. And why do we do that? It'll really help James in terms of autonomy, really leverage all the data we would have got from the R1 fleet. We're going to keep many of the things like our vehicle access system, ultra-wideband, the same. We did phenomenal work launching Dolby Atmos. We'll leverage all of that. So name of the game is leverage. We have a lot of fantastic foundational technology that we developed for Gen 2. We'll take it straight into R2.
And finally, we're going to drive cost. Cost, cost, cost is really our focus in R2. It comes from increasing integration, optimizing the BOM, working with the supply chain. And really, also, one area of cost, we don't talk as much. We've talked a lot about BOM. But as we design R2, an area of focus is really how do we keep assembly costs down. So as we've learned, R1 was our first vehicle. With Gen 2, we did a much better job of bringing assembly time down, which is one of the things that helped. I think, as RJ mentioned earlier and Claire earlier, we really have improved our efficiency at the factory. And we continue to do that when we think about building R2s, how do we put these vehicles together. So with that, thank you very much. And I'm going to hand it over to James.
We'll talk to you about how they build the phenomenal autonomy features on top of the hardware platform. Thank you.
OK, thanks, Vidya. Good morning, everyone. My name is James Philbin, and I run the autonomy and AI team here at Rivian. Very excited to tell you today about some of the new technology we built in the Rivian Autonomy Platform that actually launched with our Gen 2 vehicles. Maybe to think about what are some of the principles we thought about when designing this new platform. I think four key things come to mind. One is we wanted the platform to be scalable. So that means building headroom into the system. Everyone has the expectation that Rivian software gets better and better over time. We want the same to be true of our autonomy stack.
And so to do that, we've thought about three key things with scalability. One is building flexible software architecture in the autonomy platform. The second is really designing our ML models in the right way so that they're leveraging all of the rich data we get from the vehicle and can really understand that. The third is all the amazing sensor compute headroom that Vidya has been talking about. So really having the best sensors, headroom in the compute, and all of those things, yeah, fit together. The next key aspect is vertical integration. So in addition to some of the benefits to velocity, time to market, and costs, having a lower cost that we're seeing, as RJ has spoken about, there's sort of two additional technical reasons for autonomy why vertical integration is so important.
So the first is that when you think about these being a series of ML models, you really want to be able to train these things end-to-end. So if you have a supplier box, it's just a black box, you're not able to do that. You essentially have to take the interface that they give you, and there's no improvements you can make on that. The second is we want the system to be re-simulated in the cloud. So you basically need to take this black box, take the software that runs on it, be able to put it up in the cloud. Suppliers aren't providing that. So I think this concept of vertical integration is maybe even more important in the autonomy domain. The third is we want to be AI first.
So that really means using these rich, powerful, large ML models, using lots of data to train them, so really leveraging our fleet. And the third is, as AI is obviously rapidly expanding, rapidly improving, we want to take those benefits as they come out and be able to apply them within our platform. And the fourth is around extensibility. So computers getting better, sensors are getting better. We have actually new modalities even coming into the automotive domain. We really wanted to build a stack that can extend over time, can leverage those without having to rewrite everything. So yeah, all of these four really fed into our design for the new platform. So great autonomy starts with great sensors. I think Vidya has covered some of these facts already. On the Gen 2 platform, we have 11 cameras. That's 55 megapixels of imaging.
It's actually the highest number of megapixels of any market, any vehicle on sale in the U.S. today. It's a huge amount of resolution. As well as the LED flicker mitigation you just heard about, and these cameras actually have amazing dynamic range. So where that matters is things like driving into a tunnel, you have a sudden contrast change. The new cameras are actually able to expose all of that data correctly. So Rivian, we really believe that great autonomy needs multimodal perception. So that's not just from cameras. That includes other modalities. So cameras have this amazing angular resolution. They see a lot of detail, but they're not detecting depth. They're not detecting velocity. On the other hand, radars, they're very sparse, but they get that depth, that velocity measurement directly.
So we believe that the best perception actually comes from the fusion, from the combination of those sensors. And I'll show you in a second how we do that. These radars, obviously, they work in bad weather, low light. And independently, they're capable of detecting up to 1,000 ft ahead. That's around 10 seconds at highway speeds. In addition to the sensor capabilities, we also spend a lot of time thinking around the configuration of these sensors, making sure that we have 360-degree sensing and actually overlapping the sensors that provides redundancy, robustness, and great performance. OK, let's talk a bit about the perception and prediction stack. So this is really the core of the new platform. This is a large, deep ML model that takes all of the sensors together in a combination and generates this world model output. So how does it work?
On the left here, you have the sensors coming in. Note that not only do we have cameras, but we have radars. Radars are really a first-class citizen in this model. These go through these powerful neural net backbones that can really process and generate these latent features. All of that information is then projected into our geometric temporal latent space. Imagine that's like a big tensor of numbers, essentially. It has some temporal consistency as well. It's able to aggregate information over time. Then we use a transformer-based backbone. You may be familiar with those from some of the large language model innovations that have recently come out. It's a very similar approach. The transformers in them have this self-attention layer.
That's one of the ways that this model can actually leverage all these multiple modalities and multiple sensors to give the best possible output. What are some of the things, some of the heads that we get out of this network? Obviously, objects. So where are the agents around the vehicle? How are they moving? What is their speed and velocity? Map. So that's essentially the lane information, any static features about the world, such as traffic lights, stop signs, and other things. The predictions is how we think agents are going to react in the near future. So it's very important for planning, say, a lane change or a zipper merge that involves a lot of interactive prediction. Occupancy is essentially the free space around the vehicle. In future, we want to add into this same model, essentially, the behaviors.
Learned driving behaviors seen from real data. And that can give us a very naturalistic, human-like autonomy approach that we can integrate into the planning step. In the new platform, we also introduced a state-of-the-art motion planner. How this works is it can take all the world model information I just showed you. So those agents, those condition predictions. And in the near future, these learned behaviors. And we can also fuse it together with the rules of the road. And that's actually an important aspect. So if you're just learning driving behavior from watching humans, well, humans, we bend the rules in all sorts of ways, large and small. Humans are often rolling stop signs, maybe breaking the speed limit slightly. We don't want our autonomy stack to kind of learn those bad behaviors.
So by having these rules of the road and having a cost-minimal motion planner that can actually integrate that behavior, we can add these sort of guardrails onto that learned behavior and produce a really predictable, great experience for our customers. OK, so all of that amazing sensing and AI processing is a really amazing computer. This is the same autonomy compute module that Vidya has just shown. So as she was saying, this is 10x more powerful than the computer on Gen 1 and actually 250 TOPS. And so 250 TOPS, that's just a number. I wanted to make it a little bit more real for people. So brand new iPhone, I think iPhone 15 has 35 TOPS. So it gives you a comparison. The brand new M4 MacBook or similar lapTOPS have around 40 TOPS. And Rivian Gen 2, 250 TOPS.
So for the vast majority of our Gen 2 customers, the most powerful computer they own will be the Rivian sitting in their driveway, which is pretty amazing to think about. Extensibility is another key aspect of the new Rivian autonomy platform. In addition to some of the abstractions that Wassym was talking about, we also think about two additional abstractions specific to autonomy. So one is the sensor abstraction. So really having a single interface for different types of cameras, different types of radars, so that you don't have a strong tie to any particular sensor. You can upgrade and change them over time. The other is on the neural net processing side. So all of these chips have slightly different APIs that you use to run inference. And we wanted to abstract that away. So essentially, we have that independence from any particular compute platform.
And that sort of extensibility allows us to cover a wide range of vehicle platforms and more in the future. And then on the sensor side, I showed you the perception approach. This is another type of abstraction where using the same approach, we can project different types of sensors and new sensors into this same learning framework. And again, that allows us to integrate new sensors, upgrade new sensors, while still being able to leverage the old data, the prior learning that we did. Big ML models are only as good as the data you feed them. And so we spend a lot of time thinking about how do we build this data loop. So starting on the left, the data can come in from our different types of fleets. And there are sort of three types of fleets that we have access to.
So one is the engineering vehicles that we're constantly using for testing. The other is actually a ground truth fleet that has LiDARs mounted on it. And that gives us that 360-degree fine-grained coverage around vehicles. And then the third is our customer fleets. All of this information is anonymized. And maybe I'll just say a few words about that. So when we do the anonymization of the data, we kind of remove, we try to remove any data collected from homes, people's place of work. We also blur license plates, signs, and faces. And so the data that comes in is really kind of anonymized. It's just typically driving data from public roads. So we take that sensor data. Here you're actually seeing some data from one of our ground truth fleets. That's a LiDAR scan that's all been aligned there. And then we send that to the cloud.
That's all auto-labeled. So we have this sort of layering of bootstrap models that run in the cloud. And they're able to auto-label most of our data. So I think 99% of our data is actually run through these auto-labelers. That is generating the labels for the world model that we showed before. So it's very, very powerful. So this data. The model then improves itself. Then we train a new improved world model. And that can ship via OTA. And as the models are getting better, customers are able to use these features in more places. We see more edge cases. That data then comes in. And again, we get this sort of virtuous cycle, this virtuous upward spiral of improvement. The quality of your autonomy system is often only as good as the amount of long-tail scenarios that you've seen.
And so one of the new technologies we built with Gen 2 is something we call the autonomous data recorder. And so this is able to basically take triggers sent directly from the cloud. So we're not waiting for these triggers to be uploaded to the vehicle via OTA. We can actually push them live. And this is a really powerful way of when we identify a long-tail scenario, we can actually write a trigger that can detect that scenario and automatically bring that data back from our fleets. That data then could be used for validation, for eval. It could also be sent straight to our auto-labelers to improve the models. So it's a really powerful way of extracting data from the long tail. Another key aspect of the new platform is simulation.
So one of the principles we had is that every event that we encounter in our fleets, we can take it back to the cloud, replay it, and evaluate it through simulation. Since Gen 2 launched, I think it's about three weeks or so, we've actually run, our engineers have run more than 5.6 million scenarios through simulation. Typically, we can get a scenario back from the fleet. An engineer can look at it, and we can actually fix it the same day. So it's a massive improvement in velocity over what would have typically been done, where you'd have to actually, if you knew of a scenario you needed to improve, you'd actually have to get a test fleet, take a vehicle out, try and find that data, maybe try and stage the scenario. This is a much, much faster, much better way of developing.
OK, so I think three key takeaways for you today. So number one is that autonomy is a key competitive and strategic advantage for Rivian. We think that as these systems get better and better over time, customers will expect more and more from the autonomy systems available in their vehicles. And so it's very, very important for us that we really own this piece. The second is that we're taking a really state-of-the-art technology approach to the problem, supported by a lot of the great sensors and compute that Vidya's team have built for us. And the third is that this is like a key value driver. So we're able to ship features to customers that they love. And it actually can save our customers money. So customers who use Rivian Insurance, they actually get an 8% discount on their insurance on average from using these driver assistance features.
We think those value drivers will continue to expand and grow over time. With that, thank you. I'll hand over to RJ to talk about some of the new product pieces.
All right. There's a lot of technology in the vehicles. Hopefully, you see that from our teams. What I'd like to focus a little bit here is across the product, look at some of the cost reductions that you've heard Claire and I talk about in the context of Gen 2, and importantly, also help map how that translates to R2, R3, and beyond. From a product point of view, these decisions we've taken around cost, which for the last year or so I've been talking about, we want to make sure we really recognize these are also simultaneously driving improvements in the features. The experience in the vehicle is being elevated.
What it feels like to drive a Rivian, live with a Rivian every day, is getting better. The Gen 2 vehicles are fundamentally better, but also simultaneously much, much lower cost. And so that activity around driving cost focus into what we're doing without compromising the experience has been something that's embedded into the organization. We're going to see that. And I'm going to demonstrate some examples here really clearly with R2. So first, let's look at a walk of our Gen 1 cost structure. I'm going to walk that down to our R2 cost structure. And so we're looking at on the left of this chart, this is our cost in the first quarter of 2024. And with the introduction of Gen 2, by the end of this year, there's a 20% reduction in our material cost. And now we're showing that that's Q4 of this year.
It continues to get better following that. And as Claire goes through the financials later, we'll see further increases, further improvements, I should say, in R1, but just with the changeout we just made purely on material cost, a 20% reduction. Now we can look at what are the categories that falls into. Propulsion is a big one. The insourcing of completely insourcing all of the motor development design is a big driver here. We also simplified our body structure. And while that doesn't necessarily, it's not fully captured just in the material cost, it also dramatically improved how we run the line. Significant reduction in the number of amount of adhesive, the number of joints in the vehicle. So the actual conversion cost or the cost to build vehicles and plants is a lot lower.
You've heard a lot about our electrical systems being meaningfully lower cost, multi-thousand, more than thousands of dollars of savings in the shift to our Gen 2 platform. Then lastly, improvements in our chassis system. This gets us to a materially lower BOM cost at the end of this year. The question I get asked all the time when people take a look at R2, they say, well, R2 looks awesome. It's all these features. And I want one for myself. But how is it cheaper than R1? It looks like it has a lot of the same content. It's actually about 50%, just under 50% lower cost relative to what we did in R1. A lot of that comes out of the electrical work.
You've heard us hopefully talk again and again around the importance of consolidation of ECUs, the importance of simplification of the network architecture. Wassym, in a funny way, carried out just a piece of the harness that was removed from R1. We're going to continue to do that in R2. And that incredible focus and discipline around electronics in the vehicle will represent one of the biggest cost savings in R2 relative to R1. And as you heard from Vidya, this is not hypothetical. This is being built in our development process and built in our prototypes today. The next is propulsion. So we've got, and I'll talk about in a moment, updates to our in-house drive units that drive significant cost savings out. We also have made significant updates to how we've approached the battery pack. We're moving to a 46-millimeter diameter cell.
It's a 4695 cell from a 2170 cell that's in R1. Then the body structure. We've really simplified the body architecture. I'll show a couple of great examples of what that looks like. Use of high-pressure, large high-pressure die castings to accomplish a lot of part consolidation and part elimination. Then lastly, in the chassis, this is an area where both the much stronger sourcing leverage that we have today relative to where we were when we sourced R1 back in 2018, 2019, coupled with architectural decisions to simplify the chassis architecture. R1 has 150 millimeters of ride height adjustment, crazy dynamic capabilities, really exciting. R2, we've simplified that. I'll talk about that in some detail. But why don't we first start with a look at the battery pack in R1? I'm going to blow apart here, if you will, or explode the Gen 2 pack.
And really, the only component that's shared between Gen 1 and Gen 2 of a battery pack in R1 is the cell itself. Everything else has changed. The module design has changed. The pack enclosure has changed. We transitioned from an assemblage of aluminum components and aluminum extrusions to large high-pressure die castings at the front and back of the pack. And an example of that, just looking at the front of the pack, we remove 17 individual pieces and consolidate those down into one. That's a cost reduction of the front of the pack by about 47%. Another example, we look at the bottom protective shield on the bottom of the vehicle. When you get the off-road package in R1, it is incredible what you can hit the bottom of the vehicle with. It's think of it like a ballistic shield.
You can land on a sharp rock, and the battery and the vehicle are OK. We were able to take about 54% of the cost out of that lower plate while not increasing the mass and while simultaneously maintaining the exact same level of extreme performance with what we launched with. Now, those learnings that went into the Gen 2 pack, we've taken into the R2 pack as well. The R2 pack benefits from a much larger form factor cell, as I said, a 4695 cell. As you can see here, the pack consists of three long modules down the length of the vehicle. Importantly, the pack is highly structural. In fact, the top of the pack is also the floor of the vehicle. So we have, again, an example of focus on part consolidation or part elimination as a way to drive cost savings.
Now, in addition to the way we've looked at the pack design being a structural member of the vehicle, we've also designed it such that from a manufacturing flow point of view and how these modules are set up, those three long modules you see along the longitudinal axis of the vehicle, those are the same dimensions. We build three of the same module, and those go into this singular large structural pack. Now, I also want to talk about drive units. We launched, as I said earlier today, with what we call internally our Origin drive unit. It was a four-motor system, two motors per axle. We developed the power electronics, so the inverter in-house. We developed the gearbox in-house. We sourced the rotor-stator assembly from a Tier 1 supplier.
About a year and a half ago, we introduced our Enduro drive unit, which is what's used in both the commercial vans and the single motor application, a front-wheel drive application, and then our consumer vehicles in a dual motor application, one motor in the front, one motor in the back. That's 100% in-house. It was our first completely in-house motor. We're beyond just the inverter and the gearbox and the assembly. We also build the rotor and stator in-house. Now, with the Gen 2 vehicle, we've also brought the quad motor, or I should say the two motor per axle system, which enables both a tri-motor and a quad motor version of the vehicle. We call this internally Ascent. This is 100% in-house. It's an extremely high-performance motor, but it's also much lower cost.
And so what I'd like to do is use Ascent, because it's our most recently launched motor, as a chance to just explain some of the innovations that have gone into this. This is the beginning of what we'll see even further improvements as we look at R2. But first is the inverter. Vidya talked about our strength in designing power electronics in the context of the DCDC, the ACDC, the work that goes into all the onboard electronics and consolidating that for R2. There's also power electronics in the motor. We have to take the DC from the battery and put it into the motor. On the Ascent drive unit relative to the origin drive unit, it's a 75% more dense package. We've increased the volumetric efficiency by quite a bit. We've really eliminated the vast majority of any fasteners in the inverter itself.
We now weld those assemblies together. It's both more robust, but it also takes fasteners out of the platform. And all that allows us to build that inverter for 25% less than our Gen 1 origin platform, with actually higher levels of performance, as we talked about a moment ago. Moving from that into looking at the housings themselves, we've really simplified and minimized the number of structural components. And we've also by consolidating the parts. But that also provides higher levels of stiffness. And we've isolated a number of the components to improve the NVH of the vehicle. And later today, those that are here will get a chance to drive the vehicles. And you'll be able to drive a quad motor.
The performance levels are not only remarkable, but the refinement of the motor is a material step change forward relative to what we had in our origin. We had a few folks that are here today drive some vehicles yesterday. And their owners as well. And they said to me, now I'm really torn because I have to turn in, I have to trade in my Gen 1 and buy a Gen 2. And I said, well, that's great. That's the goal. So moving from the housings and the structural integration that we've accomplished there and part consolidations come with that, we've also really optimized how we've approached the hairpin windings in the stator. And so in the Enduro drive unit, which was the first completely in-house rotor stator assembly, there were 15 different hairpin winding shapes. In what we've launched here with Ascent, there's 3.
So really significant simplification of the hairpin winding process. For those that go on the full tour, I invite you to look at that. It's a really cool process. We take big coils of wire, and we bend them into the right shapes. We're not bending them into three shapes with Ascent as opposed to 15 with Enduro. And that really simplifies the equipment, but also allows the equipment to run a lot faster. Then the next area, which I want to call out, which is important from a performance point of view, is these have direct oil cooling. Our origin platform, and I feel bad saying this for all of our origin owners out there because I'm making you really want to get a Gen 2 vehicle.
But our original drive unit used a water-cooled jacket around the stator and therefore had more limited cooling efficiency, particularly in high-performance applications. What we've done with Ascent and will do going forward is we have an oil-cooled platform, but we actually oil-cool directly into the rotor. And so a lot of work went into modeling oil flow, managing things like windage losses, so the losses of the rotating elements of the motor moving through oil. And when you drive the vehicle, particularly if you're driving it in an aggressive way, like for example, this past weekend we competed at Pikes Peak and beat our own record by 30 seconds, which is significant, but set the all-time course record for an electric truck. Cooling is very important. Now, that's just a light touch of all the things we've done from motor development.
What I'd like to look at is in totality, what does that look like from a cost and performance point of view. So looking at our quad motor here, that was in origin. It was a high-performance vehicle. It's a great vehicle. It's a 3-second 0 to 60, 11.7-second quarter mi, delivered around 328 mi of range on the EPA drive cycle. If we compare that with our completely in-house quad, the performance is phenomenal. 2.5 seconds 0 to 60, range is up to 350 mi. Cost is reduced by 24%. But importantly, it's more exciting. So when we say we've taken cost out, it's not as if the vehicle has become less capable. It is really capable. And when I say 10.5-second quarter mi, for those that aren't real performance enthusiasts, this is remarkable.
Some of the fastest hypercars in the world are not able to accomplish this. So this is faster than almost every extreme sports car in the world. A McLaren F1 does that in just under 10.9 seconds. So we're faster than what often many people think of as like the ultimate supercar. Moving that to the Tri, where we have a two-motor system in the back and a single motor up front, this is very high performance, actually higher performance than what we did in our quad. But it takes the cost savings even further. So it's a 32% reduction relative to what we launched with.
Then looking at what represents really the meat of volume for the R1 platform, what we call our Dual Performance, that's a 47% reduction relative to what we launched with, with significant range, 420 mi of range, and performance that's very similar actually to our original quad drive unit. This is the R1 portfolio motors, the Ascent Quad, the Ascent Tri, and our dual. Going into R2, we're taking those learnings and we're pushing it even further. The R2 drive unit, its internal code name is Maximus, which I love that name. It's also the name of one of my sons. But the Maximus drive unit reduces the overall piece cost or material cost in the motor by 30%. We further consolidate the number of components into it. We further simplify how it's assembled.
A big focus is on the amount of labor content in the motor. We've reduced the number of parts in the motor. Driving the reduction in cost is a reduction in the number of fasteners. 32% less fasteners than what was in Enduro. We really think of this as an Enduro Gen 2. From this image, it may look a lot different. The big difference here is we've moved the inverter to the side of the motor, which allows it to package really tightly. I wish I had one sitting right here. The power density of this and the size of this is just beautiful. It fits really nice into the front of the vehicle and into the back of the vehicle for both R2 and R3.
Now, if we look broadly at the high-voltage system, of course, everything that I've just talked about requires a lot of cooling. In the Gen 1 vehicle, we had a very sophisticated cooling system that allowed us to cool the battery, the drive units, and of course, the cabin with a single vapor-compression refrigeration cycle. Heating was accomplished with essentially resistance heating. So we weren't achieving the type of efficiency we needed from a heating point of view. What we've done is we've replaced that vapor-compression refrigeration cycle and resistance heating system with an integrated heat pump solution. We developed this heat pump solution to really drive a lot more efficiency into how the vehicle performs in cold weather environments. It's a 3x improvement in efficiency in heating and without adding any cost into the system.
It's slightly more cost advantage than what we had in Gen 1, but dramatically higher performance. A variation of this will be used in R2 as well. Now, a lot of time and effort also goes into the body structure. In the Gen 2 body, of course, we had constraints that were set up around how the body was originally architected. But even with those constraints in place, we were able to really significantly pull a lot of cost out of the vehicle. We took 65 parts out of the vehicle, out of the body structure. A number of parts were converted to steel. What you see here animated on this slide, the blue parts being pulled out, those are all the parts that changed as part of the Gen 2.
That helped us drive a 30% improvement in the production line rate, the number of jobs per hour. Overall, we were able to reduce the number of connections, joints, and welds and adhesive joints by close to 1,500 joints. Really heavy emphasis on pulling processes, parts, and steps out of what it takes to build the vehicle. Ultimately, the vehicle achieves the same level of safety, which is it's the safest SUV and the safest truck you can buy today. It's five-star safety, IIHS Top Safety Pick+ . The truck is the only electric vehicle, the only truck that accomplishes this. It is truly the safest vehicle on the road. We've increased the stiffness of the body, which you can really feel from a dynamics point of view when you're driving it while driving out a considerable amount of cost.
Next, I want to show an example of how we're driving cost into R2. A question that all of you should be asking and is absolutely foundational for us as a business is R2, hopefully see the level of excitement for it in terms of how consumers reacted. We're offering at a much lower price, which means the cost to build the vehicle needs to be a lot lower. It's a lot of content in the vehicle. You see it, and it doesn't feel like it's a cheapened Rivian. I'm going to show one example here with the doors. These are the rear doors in the vehicle. I show the R1S on the left. This is a Gen 2 door system on R1S and the R2 door. It's a 54% cost reduction relative to R1. You say, well, how is that possible?
It looks so similar. The door is slightly smaller. But a lot of it's through a number of part reductions. So we've reduced the number of parts in the door by 65%. And some of it's architectural. So you'll notice in the R1S door, there's a division bar. So there's a piece of fixed glass in the back side of the door. And whereas on the R2 door, the entirety of the glass drops. And this seems like it's such a simple thing, but the whole vehicle packaging was optimized in order to get a full glass drop on the rear door without a div bar. And it's cool when you sit in it because you're sort of trained to expect this fixed piece of glass or the glass to not drop all the way because of the rear tire.
We've set up the proportions of the vehicle and iterated more times than Jeff and our design team would have liked to get to this place where we remove the div bar, have full glass drop, and the vehicle proportion looks great. That was one of the many optimizations that have been driven into the vehicle to take cost out without taking away the performance or without taking away the experience for the customer. I'll use another example. Let's look at the underbody structure in the rear. On the left, we see the R1S structure. This is an assemblage of components. On the right, we see R2, where it's a large high-pressure die casting. It's a 96% reduction in the number of parts. It's 32% lower cost. The structural performance of it is outstanding.
And so you saw in R1 with the battery pack, we've introduced large high-pressure die castings into the vehicle to consolidate parts. We're doing this throughout the body structure in a number of really key places in R2. And the rear of the vehicle is a really good example of that type of part consolidation. Last, let's look at the front of the vehicle. Let's look at the suspension. I referenced this earlier. One of the core elements of Rivian as a product and as a brand is the vehicles are fun to drive. And in R1, it's silly levels of performance, as I often say. And so what we have is massive ride height adjustments, 150 millimeters of adjustment, that's 6 inches of ride height adjustment. We have fully adjustable roll control accomplished with an electrohydraulic damping system. So we can control the roll resistance to an infinite degree.
So it allows you to go really extreme off-road and really extreme on-road. We have active damping. And it's set up around a double-wishbone suspension in the front and a multi-link in the rear. On R2, we've gone to a MacPherson strut. So much simpler from an assembly point of view, much fewer parts. We have a passive coil. We have passive roll control. And we have a semi-active damper. So there's some ability to adjust the damping frequency, but nothing to the degree of what we have in R1. And that allows us to achieve 72% BOM savings. So while the performance will not be as extreme as R1, it's still a really nice, but it's still a really nice to drive vehicle, there's a significant level of cost reduction that we've achieved. And I'm just showing the front axle here. Of course, similar exists in rear axle.
So while the cars look and feel very much as a Rivian vehicle, what I want to make sure is clear is we are hyper-focused on making sure out of the gate, when we launch R2, the bill of materials cost structure is set up for us to have strong positive cash flow in that vehicle for it to fund and the continued flywheel of growth for us as a business. Let's look at R2 a little bit further. Let's look at the body structure. R2, when we unveiled it a couple of months ago, I think folks expected it. What was unexpected was the other thing we were working on in this platform, which is R3.
The excitement of having those two vehicles represent really the, or demonstrate, I should say, the elastic nature of our brand, how our brand can fit in a five-passenger two-row SUV. The brand works really well in a crossover with a form factor that's not expected. It's different than what is out there. Something that's very unique with R3. The architectures of those two vehicles are shared. So that means the battery pack is, we have a shared battery pack. Both use the structural pack you can see here. The floor or the top of the pack is the floor of the vehicle. The mounting points, the way it fits into our plant, the mating features are the same. The crash strategy for R2 and R3 is the same. The motor orientation and mounting is the same. The steering and braking systems are the same.
A lot of focus has gone into making sure that despite the fact that these two vehicles look different, feel different, have different form factors, and appeal to different types of customers and customer needs, they're very common under the surface. I'm showing here the physical things, which may be harder to imagine how those are the same, but they absolutely are. Of course, the software and network architecture and electronics that you've heard a lot about today, that's very common between R2 and R3. Of course, as we now can talk about products well beyond R2 and R3, even outside of the Rivian family, but into Volkswagen Group's portfolio products as well. Now, across all the systems I've talked about, I've really emphasized here the technical changes we've made. Simplification of design and consolidation of number of parts, removal of fasteners, elimination of parts.
I also want to be very clear that the vantage point or the position from which we're negotiating with suppliers for R2 is fundamentally different than how we sourced R1. R1 was largely sourced in 2018, 2019, and a small little tail into 2020. The auto industry was at peak volume at that time. It was a very hard time for a new brand with unproven product in 2019 to be sourcing contracts. A lot of the cost savings we see with Gen 2 are a result of us also renegotiating with suppliers. What now exists as we're sourcing R2 is a level of supplier excitement because they recognize how successful the brand has been. They recognize that we projected volumes with them and we're achieving those volumes.
As suppliers think about who the customers are going to support, who the customers are going to lean into, they recognize R1 has been the best-selling large EV in the United States. It's the best-selling large premium vehicle in California, EV or non-EV, as I said before. And so it's introduced a level of confidence within the supply base for R2, which if it achieves even close to the level of market share we've accomplished in the premium segment with R1, this is going to be a lot of volume. And so just to illustrate how that sourcing leverage translates into bill of materials pricing, we've selected some components here that are very similar between R1 and R2. And I'm showing the difference in cost between what we sourced in R1 versus what we're now sourcing in R2. First, looking at front displays. It's a 47% reduction.
If you're sitting in R2 later today, for those that are here, you can sit in an R2 or at least look in an R2. Very similar displays to what is in R1, significant cost reduction. Cross-car beam, that's the structural element that sits underneath the instrument panel, 58% reduction. The center console, there's some simplifications to the design as well, a 71% reduction. These are sourced numbers. The front brake calipers, 40% reduction. Vidya talked about the cameras that we've developed, which also gives us a way of removing some of the margin stacking, 20%, 28% reduction. Steering rack, very similar set of requirements. In fact, R2 steering rack has even more aggressive requirements in terms of some of our objectives for redundancy, 22% reduction. So I'm giving six examples here. There's thousands of parts in the vehicle.
This is intended to demonstrate the success that we're having in sourcing R2 and the success that we're having in designing these components across the R2 bill of materials to achieve the cost levels that are necessary to support the price points that we've talked about, the vehicle being available in the $40,000-$45,000 price point range. Now, the other thing we have now publicly said, and we said it when we unveiled R2, is that we're going to be building R2 first here. So this was a decision we spent a lot of time on. And it allows us to save $2.25 billion relative to launching it first in Georgia. It also allows us to remove launch timing risk and allows us to make sure that we get this product in the market as quickly as possible. Georgia is still a very important part of our strategy.
It's an important part of scaling our mid-size platform. But to have it launch here first with the Normal facility having 215,000 units of capacity per year, that's a really important element. And so for those that are driving in, you can probably see some construction happening in the building. There's a new structure getting added on top of our paint shop, which is to facilitate what you see here in this diagram, which is the general assembly body and end of line being this new building that's getting created. But we're going to leverage our paint shop. We'll leverage our stamping operation. We'll leverage our battery and driving areas of the plant. But some of the construction you already see underway is to facilitate adding those orange parts you see here on this graphic representation.
So everything that you're hearing from us around our product and around how we're running the business and around how we're driving towards profitability, my hope is that you're seeing really an extreme sense of urgency we have as a company. This is the badge that goes on the back of our new quad motor. It's a fun way of communicating. This is a 1,025 horsepower vehicle, but it's also friendly. So when you see that in the back of Rivian, it means it is a very fast Rivian. But I think it also embodies how we operate as a company. We are moving very, very fast. We are driving towards the improvements necessary to get to positive free cash flow, certainly before that, positive gross margins this year.
With Q4, you've heard Claire and I continue to guide on this, that we have confidence in our ability to demonstrate positive gross margins on our path to healthy gross margins for R1 and very healthy margins for R2. But that's built around an organization where the spirit, the energy, and the passion is embodied in this badge that goes on the back of our quad motor. So with that, we're going to now have a Q&A to go deeper on product. We'll pull some chairs up here. We'll spend some more time talking about that. Then we'll break and come back and focus on the go-to-market side of the business. Then, as I said this morning, Claire will walk us through some of the more detailed elements of the financials. So Wassym, James, Vidya, you guys want to come on?
Great. Well, thanks, everyone. As RJ mentioned, we're going to host the technical Q&A now. As a reminder, there will be a separate financial Q&A after Claire's presentation. If you do have a question for the technical Q&A portion, please raise your hand, and Tim and I will try to make it your way.
Hi, good morning.
Good morning. This is George Gianarikas from Canaccord Genuity. I had a question around autonomy. I mean, there's a rampant debate in the marketplace around end-to-end neural networks versus hybrid, also around how much perception you need, if you need maps, and how much data you need. It sounds like you're sticking with the hybrid approach. If you could just sort of articulate to us why you think that's superior? And second, how you expect to, it sounds like you're also leveraging a lot of data, how you expect to increase your data advantage relative to the amount of vehicles you have in a fleet. Thank you.
Yeah, oh, Amy?
Yeah, you're good.
Yeah. Yeah, I'm not sure that I would almost call that a false dichotomy. So I think the architecture that we've shown actually will scale to including fleet learning and driver behaviors. So that's that learned behaviors piece. I think where we would disagree is that that's not everything, right? Being able to put in controls, being able to sort of curate the experience to customers is really an important part of actually shipping a product. Just learning from human-driven driving behavior actually doesn't allow you to do that in many, many ways. So we think that that end-to-end piece is a very important part of the stack, but it's not everything in the stack. And so our motion planning architecture actually allows us to put those guardrails on that allow us to kind of curate that autonomy experience.
I think, yeah, that's personally what I feel is the correct way to build these systems.
But to be clear, I mean, the platform is absolutely built around an end-to-end architecture. I think the subtlety, and I think this is often just left out, is the necessity for some rules of the road for all the reasons James talked about when he was presenting earlier. If you were just simply to watch human behavior and map from that, you start to learn some bad behaviors.
Morning. Thanks for all the details. Jordan Levy from Truist Securities. Just wanted to see if you could expand a little on the structural battery pack and some of the, if there's any complexities associated with transitioning to that. You talked a bit about the cost savings, but just interested in learning a bit more about that.
Yeah, if we were to look at the battery pack that's in R1, it provides structure to the vehicle as well, but not to the degree it's hard mounted. It adds torsional stiffness, bending stiffness advantages, but not to the degree that we've done it on R2. In R1, there's a separate floor and there's a separate top of the pack. In R2, the floor of the vehicle is the pack. Now, what that means is it also creates requirements around how the pack seals to the vehicle. Because if the pack to vehicle seal leaks, the vehicle has the risk of leaking as well. So it does set up a set of requirements around how, and you probably could see it in the image, there's actually a seal around the periphery of the pack that when it goes in, it presses up against the upper body.
It sets up a set of requirements on what that seal plane looks like. But because it's been designed in from the very earliest days of the architecture, that doesn't introduce any complexities, and the cost savings from it are meaningful. It also allowed us to reduce the amount of material across the rest of the vehicle to accomplish our torsional stiffness requirements and bending stiffness requirements at a vehicle level.
Great. Thank you, Dan Levy, Barclays. I think you've gone through a bit of a journey in terms of figuring out where you want to vertically integrate versus where it's maybe better to use other partners. I know in the past you've talked about you had previously discussed doing your own cells in-house, and I think you've abandoned that. So maybe you could just talk to two areas specifically now. How do you view the in-house versus supplier split? One is on the compute electrical architecture side, where it seems like you're still extremely committed to doing that in-house. And the second is on the battery side. To what extent are you looking at in-house versus using partners?
Sure. Well, Vidya, I tagged you in this. I guess just speak generally around the electronics. I think you all understand the complexities of a tier-one-based ECU topology. And ironically, some of the history of that dating all the way back to electronic fuel injection systems in the 1960s that then led to this proliferation of small little ECUs numbered in the tens of ECUs, so 50, 60 ECUs, as Vidya said, some vehicles up to over 100 ECUs in a vehicle. That's proliferated. And along with that, the organizational structures that exist across legacy manufacturers that make it really hard to move off that system. And it adds a tremendous amount of cost and adds a mountain of complexity to the development of the vehicles.
For that reason, early on, from a decade ago, we knew that we weren't going to go down the path of building any dependence whatsoever on tier ones. The Gen 1 vehicle, as Vidya outlined, clearly the 17 ECUs that were in the vehicle were developed in-house with the plan to consolidate those at a future date with. Now, we've seen that with Gen 2. But this, we believe, is a significant structural advantage over time.
It's a structural advantage both in terms of cost, thousands of dollars of cost savings per vehicle, but also, as we heard from Wassym, it allows us to create really rich features that leverage multiple aspects of the vehicles, whether that's you walking up to the vehicle and the vehicle unlocking, the lights doing an animation, the climate control activating, the seat moving to the right place, the battery management system preparing the battery for the drive. All those things we can do very easily because we control the stack.
The strength of that platform has really been validated through the partnership we've announced and the $5 billion deal that we announced with Volkswagen Group, where our platform, so that's our network architecture, our topology of ECUs, and then, of course, the software stack that sits on top of them forms the basis of what will be going into a wide range of products across Volkswagen Group. That's a really exciting validation. It's not as if this was something that Wassym, Vidya, and I talked about and thought about two weeks ago. This is after a lot of detailed work and analysis with the Volkswagen Group technical teams. To have those teams so excited to be able to leverage this platform that we built to really drive their products forward is remarkable. Now, with that said, Vidya said it a few times.
We talk about this as if it's sort of simple, but that requires us to build an organization capable of doing that. And so, I mean, I remember that when we were private before we went public, when I was talking about this with some of our investors, I said, "Hey, we're going to build all of our electronics in-house." And they said, "Really? That seems like a lot of work." And I said, "Yeah, but it's going to be the right call in the end." But that decision happened a long time ago to build the computer design team capable of designing these computers, testing, validation, requirements management of a really full end-to-end development process that can't be created overnight. There's a lot of infrastructure to do it. To do this, we have to build our own testers. We design those testers.
So we have, I believe, a significant structural advantage in what we've created there. And before we jump to battery, Vidya, it'd be good to just have you comment on this because you've been so your team's doing just such an amazing job of building this platform out.
Absolutely. I think the one key differentiator I think I look at when we decide whether we build in-house or not is it going to run differentiated value-add software. And if that's the case, you want to build it in-house because you want it to evolve. Our product evolves over time. We get not only the advantages of feature introduction over time, but also just speed and velocity. Now, as I said, so that's why we build our ECUs, but we're not off building speakers or displays in-house. We spec them, and we will drive what we want, but we won't necessarily build them. And then I'll give you another example about something that runs software, but we don't feel a pressing need to develop in-house, like the airbag control module. It's not differentiated value-add that you change over time.
You make sure it's very robust, it's safe, you test it, you put it there, and you don't really touch it during the lifetime of the vehicle. So the key differentiator in my mind is, will it run differentiated value-add software? Then absolutely, you want control over it.
It's interesting. You know, with the exception of the airbag control module, this also changes the way we approach sourcing. So typically, when you're sourcing a, we'll pick any system, a seat, a headlight, a door system from a supplier. And a number of you as investors, I'm sure, talk to some of these large Tier 1s. They want to sell the system. They want to sell functions, and they want to sell software and hardware with it. We want to have essentially no computers in anything. We want to bring your own computer kind of company where we'd rather every system be as dumb as possible because we do all the compute and all the thinking within our platform.
So it really changes the way we engage with suppliers where we no longer need a component to come with or a feature to come with an ECU, but rather it just needs the light or the actuation or whatever the device is. A seat is a good example. We would not want a seat with an ECU. We would want a seat with wires coming out of it that we then control with our platform. And so it's a mindset shift in how the auto industry supply chain is also set up where for us, we need actuators and sensors, but we don't need the compute platforms to come with them. Now, to talk a little bit about battery, you asked, we've in the past looked at whether bringing battery cells in-house would be necessary.
This space has evolved so many times and in so many different ways, heck, in the last 10 years, but even in the last couple of years. The way that we think about it today, which is an evolution of how we originally thought about it, is it's really important that we can control our destiny around battery sourcing. Now, that doesn't mean we need to make the cells ourselves, but it does mean that we need to think about the upstream supply of materials going into the cells. Case in point, which we all experienced over the last couple of years, was what happened with lithium pricing. So lithium hydroxide in 2021 was under $20. It was, I don't know, $15 a kg. In 2020, early 2023, it was like $80 a kg. And today it's $13, $14.
I should have looked this morning, but I think it's $13-$14 a kg. The cost didn't change during that timeframe. So we didn't see costs suddenly increased by 7x between 2021 and 2023. But the market dynamics and the commodity pricing changed dramatically. And so the instability of that as a commodities market means that we want to build direct relationships with key sources of, in this case, lithium hydroxide, and direct those sources to cell assemblers, cell manufacturers. But the ability for us to be sourcing the upstream raw materials, directing that to a battery cell partner, specifying and collaborating with that battery cell partner, what are the chemistry trade-offs, let's say, between fast charge and so power cell versus energy cell, how much silicon do you put in the anode? These types of changes are really important.
We've built a team that really understands how battery cells are built, deeply understands the trade-offs that need to be made from an electrochemistry point of view, and then deeply understands the supply chain such that while we may not run the cell manufacturing facility, we deeply understand what the cost structure looks like there. Now, beyond the cell, so as you think about the module design, the pack design, as Vidya talked about earlier, the battery management system, the BMS that's controlling it, those are areas that we deeply feel we need to own. It allows us the flexibility of deciding not only what we're getting for upstream raw materials, but also deciding who's going to ultimately assemble and manufacture the cells that we then put into our packs.
Yeah, hi. Kevin Godwin from CPP Investments. Great presentation. It's been really helpful to understand where you're going. I have so many questions, but I'll just kind of give you one. Just in terms of autonomy and your roadmap from I'm not entirely sure what your roadmap is in terms of level two, level three, level three plus, level four full autonomy. And then can you talk about the economics around how you're going to monetize that? Can I sell it as part of the price of the car, subscriptions? Just how should we think about that?
Yeah. James, let me cover the business model piece. It's really interesting. We've had, and I'm sure all of you have, a lot of debates around what's the willingness to pay for a highly capable level two, a highly capable level three as we start to then think about new business models around things like robotaxi. What does that business model look like? This, of course, has had a lot of coverage recently. And I'll speak to it first in the context of a purchased or owned vehicle, how we think about this.
We intentionally haven't put, and even in Claire's slides, where she goes through materials, Claire's materials, we're not going to break out how we think about the vehicle purchase price relative to the value that we can potentially capture through selling our autonomy platform on a vehicle level because it's a very dynamic environment. What I said differently is this something where customers want to pay for it upfront in the purchase price of the vehicle, or do customers want to pay for it as a monthly subscription? It's early days today, too. It's a human psychology question, but it's early days to say how that will ultimately evolve.
I think part of this also links to does it become embedded with the brand such that when you buy the vehicle, you'd rather just have it all paid upfront, or are you comfortable paying for it on a variable basis? Now, we can do both, and our current view is that it's probably going to be both. There's going to be a way you can buy it upfront, and you can finance it with the vehicle. And then there's a way you can pay for it on a variabilized basis. But the amount of effort that goes into developing the systems warrants it generating revenue and value incremental to the vehicle purchase price. We absolutely do believe that. And in terms of the ceiling, in terms of what can be accomplished, even the labeling of level two, level three, level four, it's very murky and confusing to customers.
And one of the things that makes it immensely confusing to customers is when you're on the road and the vehicle's operating itself in level two, or if it's operating itself in level four, they feel the same. Unless the car crashes, they should feel the same. The difference is the robustness of the solution at level four allows you as the occupant of the vehicle to be less involved. So level two, your hands are on the wheel, eyes are on the road. Level three, your hands are off the wheel, eyes are off the road. Level four, you can be completely removed as a redundant layer. But while the vehicle's operating, it feels the same. And so the murkiness around how we call out those levels and in some ways the murkiness in how we enforce those levels has, I think, created a lot of confusion.
This comes back to one of the earlier questions, which is how do we think about safety in the vehicle? So what are the rules of the road? What are the types of roads we enable the features on? What's our confidence level of an incident or disengagement happening? We've tried, and I hope the industry moves this way, to avoid level two, level three, level four naming because it just is so confusing to consumers. What we're driving towards with both the sensor set and then the perception stack is something that gets better and better and better and builds confidence such that you as a customer feel more and more comfortable playing less and less of an active role in driving the vehicle. For our platform, that starts on highways.
And so the features that are in the Gen 2 vehicle represent, call it the first release of our Gen 2 architecture, but the ceiling is so high. So it'll get every OTA that comes, James references, it will get better and better. And the training, the offline training that's happening is immensely powerful. And it's architecturally so different than what was in our Gen 1 vehicle, where it relied on a third party for a lot of the function, and it was the fusion of the perception stack was much later. This is an early fusion system built and designed around end-to-end AI that allows it to get better and better through the accumulation of miles, which James showed that really cool image of zooming out where you see just a subset of a fraction of the amount of data we're pulling off the vehicles.
But James, we talk about this all the time. Some of it's asymptotic. It'd be good to just have you share your perspective on, we're going to be asymptotically getting better and better. And at what point in that asymptotic improvement do we say it's level four? But James, maybe let us include us here in the debates we have internally.
So yeah, I mean, I think the core pieces of technology are actually not so different between these different levels. If you were to go and look at a level four, like a ride-hailing company building a full driverless vehicle, the pieces of the stack would actually look quite similar. So we're really building that scalable technology that can take us much, much further than today and can continue to improve. I think the distinction between these levels, as RJ was saying, is maybe not so important for customers.
I mean, when we look at the voice of customer feedback we get, it's things like, "I want to use it in more places. I want it to have more hands-free experiences. I want to get some of my time back. I want help with parking. I want help with seeing pedestrians in a dark night." All of these things are not necessarily in that level two, level three framework, but they're still very useful features for customers. So I think for us, you think about the stack of technologies you build, and then that supports a layer of features, a set of ODDs, a set of capabilities that really then customers love. And I think that, yeah, that kind of goes beyond that sort of cuts across this level distinction.
One of the things within self-driving that's also important to note is there's a lot of human psychology at play here. And this is still early days for customers having comfort and confidence to let the vehicle drive itself. And it's going to take some time for customers to transition to being really, truly comfortable. And part of what we do, even in terms of the user interface, is we want to show customers the vehicle can see everything. It can see it really well. So the visualization you see in the center of the driver cluster is it's not functionally doing anything other than saying to the customer, "Look, we got it." The car sees all these things, and it can react accordingly.
And so it's almost hard to imagine going from where we are today, where most of the miles driven on our roads today are by humans, to a world where, in the, call it, the future, we can debate when that is, where most of the miles driven will be driven by the vehicle. But it will be slow, and then it'll be done. But part of that is building confidence. And so for us, the safety element is really, really important.
Thanks. Joe Spak from UBS. RJ, you really lit up. I could tell you're really proud when you started talking about the motors. You showed the great cost and performance improvement within Rivian. I'm wondering if you could share any data or information you have on how it compares relative to maybe some other competitive motors out there, either on a—I don't know if it's a dollar per kilowatt hour or whatever the right metric is. And then also, it seems like maybe you've been using that to, again, more for performance and range versus cost, meaning there could be a trade-off versus sort of the battery pack size, but maybe I don't know if that's correct or not. So just how do you think about that trade-off in the motor improvement?
Yeah, the performance cost range trade-off. So first, on our one, we've made intentional decisions around some of the components that go into the drive units to play sort of across that spectrum of performance, cost, and range. And perhaps the most flexible area that we've done that is around the power conversion and what we're using for the inverter. And so in the drive units, we have a silicon carbide power module. We also have a silicon IGBT. And so the Max Pack uses silicon carbide for the primary drive motor, whereas some of the other variants are using less capable power or less efficient power conversion, which gives us less range, but also less cost, and we build into the price.
The way we've approached the configurations of battery plus drive unit, so max pack variants are more expensive to build, but they have the highest efficiency power conversion with silicon carbide, whereas some of the other pack configurations are using IGBT. That approach is taken even further on R2, where we'll have 3three different powertrain configurations: a 1-motor rear-wheel drive variant, a 2-motor all-wheel drive variant, and then a 3-motor tri all-wheel drive variant. It allows us to still deliver great performance of the rear-wheel drive. It's quick. It's quicker than most sports cars still, but it's not silly quick like the tri motor on R2. That's been very much something that we actively looked at on R2, recognizing that the base version needs to make a trade for cost in a way that we didn't on our flagship product with R1.
R1 doesn't have a rear-wheel drive variant. The slowest version of an R1 is still really, really quick, really high performance. But again, it's a flagship product, so that was intentional.
We've got time for one more question before the break.
Hey, you guys. I'm Garvin Jabusch from Green Alpha Investments in Boulder Thanks for the specs. First of all, thanks for a great morning's firehose of information. Secondly, great specs on the Ascent motor, also a motor question. Can you share any equivalent specs on Maximus yet? And secondly, I just can't resist as a follow-up. I love the quad motor badge. Can't wait to drive around Boulder with one of those on my truck. But what is that? Is that a Wookiee? What are you guys going for there? I do love it, though. Thank you.
I love that. My favorite question of the day. So the Maximus drive unit, actually jumping off Joe's previous question around cost, Maximus is really heavily optimized around cost. So it's a little bit smaller than you can think of it as like Enduro 2.0. It's smaller than the Enduro 1.0. We're using some new tech that's using a different technology for winding the stator. It's even more cost-effective, reduces the number of welds through continuous winding. Some really clever manufacturing approaches there. But it's still going to be high performance. And the unique thing about Maximus, the rear drive unit for Maximus will need to have enough power and torque to exist as a single motor application. And so we've been optimizing the combination of the rear drive unit and the front drive unit to allow for a rear drive-only scenario that's still sub-5.5 seconds 0-60.
But some number of trade-offs around how to get to the cost numbers I talked about. Now, in terms of the character on the back of the vehicle, this goes back a long ways, but it was born out of us thinking about how do we start to communicate some of the security features in the vehicles. This was like probably seven, eight years ago. And so the cameras can all turn on into a mode that we call Gear Guard mode, where it watches the vehicle, watches your gear. And so if you walk up to Rivian and the vehicle's locked, this friendly character comes up on screen with a camera and says, "Hey, I'm watching." So it's our way of being as friendly and Rivian-like as possible as we communicate that we're seeing everything around the vehicle.
But it's led to this character in our ecosystem that's a lot of fun. There's a print out of it in the back of the room. And so when we're designing the badges on quad, I've been driving a quad prototype and get in it, and it's just mind-blowingly quick. But our brand is so inviting, and you can drive a vehicle that's faster than, I mean, significantly faster than a Ferrari Enzo, significantly faster than a Bentley, McLaren built. And I said, "Well, how do we communicate that this is a little bit silly?" It's like a truck that's so quick and an SUV that's so quick. So we have this Gear Guard on the back that just says, "Hey, I'm going fast. I'm silly fast.
The only way to do it from the UI and the vehicle to activate launch mode is you need to swipe Gear Guard. Otherwise, you cannot do it. So you really need support from Gear Guard to go really fast.
Yeah, I think this is an example, though, of something. And Wassym is our chief software officer. We joke sometimes he's also our chief Reddit officer. So we get a lot of feedback from customers. And we can try things. We can explore things. But as we see things that are connecting with our customers, with our enthusiasts, Wassym's team does a lot of work to say, "Hey, let's go create features for them." And Gear Guard's an example. But Wassym, what are some of the other things that you've created that are from you and your team scraping of Reddit?
Oh, many features. I mean, the roadmap evolves constantly with inputs from RJ and some of our customers, and then a lot of feedback that we get from customers on the different social forums. I mean, whether it's small additional features from a convenience standpoint, or actually it goes back to even creating new drive modes, new experiences for our customers. So really, the technology that we have, the fact that we can deeply update the entire vehicle, it creates a whole emotional connection with our customers where the vehicle evolves over time. Customers know it, and they are part of that experience. Customers are really an important part of the ecosystem. And their feedback helps us make the vehicle better. And they're embracing. Customers, they ask also for lots of features for Gear Guard. And we added Gear Guard in our Pet Mode.
We're adding Gear Guard to our incident cam. It's really part of that connection and that story that now we have with our customers.
Great. We're now going to take a 15-minute break before coming back for the commercial and finance portions. Thank you. Thanks, everyone. We're about to get started. If everybody could please take their seats, and we can start back up. Thank you. Okay, we're going to get started. So if everybody could take a seat, that would be great. Thank you
All right, we're going to have the second half of our session now, and we're going to start that off with Kjell walking us through some of the details around our commercial business, or I should say go-to-market business, side of the business. And then Claire will spend time going deep on financials. I think a lot of the questions that came up this morning we'll be able to go even deeper on in the second half of the session. And then we'll have a Q&A to close out any remaining open questions. And then I'm excited we're going to plant tour of drives at our delivery center. So there's a lot to see there as well. With that, let me invite Kjell to come on and join us.
Thanks, Steve. Hello, good morning, everyone. Kjell Gruner, I lead the commercial team at Rivian. So that is everything customer. So I'm going to talk about customers and our very unique Rivian go-to-market approach. And let me start with customers. When you look at customers, where are they coming from? Our customers are coming to 76% from an ICE vehicle. So they're making that switch. It is happening what RJ already described as the core of our mission. So we see it is working. From a commercial perspective, the good news is we are addressing the entire market. Customers for ICE engines, as well as BEV engines, are interested in Rivian and coming to Rivian.
When they purchase or lease an R1, what are the reasons to do that? According to our market research, these are the top three reasons to get into an R1. Reason number 1 is the beautiful and very distinct Rivian design. Kudos to Jeff Hammoud. He's here. Due to your work, people are coming to the brand. Looking at reason number two and three, that's driving performance and innovative technology. I would cluster that reason as essentially technical excellence. Everything that this awesome team of engineers is developing, everything that Wassym does, Vidya does, James does, the other colleagues are developing in Palo Alto and Irvine is appreciated by our customers as a reason to get in one of these vehicles. So that investment is clearly paying off. Then the next question is, after that purchase, do they love the experience?
To me, the core question you can ask a customer to know if they're happy with the product they bought is, would you buy it again? There is an independent study out there that measures in the automotive industry in North America exactly how happy customers are. Here are the results from that study. Rivian is the number one in the industry. 86% of our customers say, yes, I would buy again. We're even leading by a wide margin. Number two, three, four are 77%, 76%. We're leading by a wide margin. This is great from many perspectives. It's something to be proud of, but it's also great for loyalty. Yes, they would come back. They would buy another R1, an R2, R3. When people are happy about something, they would also spread the news. They would tell their friends.
They would share on social media. So this also helps us to expand the brand and build brand awareness. And we have to do that. That's a huge job to do. Maybe job number one to really expand the tent of the brand and who we are reaching. Because when you look at those numbers, brand awareness, there are a couple of studies out there that show brand awareness. We are at 40%. Leading automotive legacy brands, they are north of 90%. But many of them had more than 100 years to get there. So you could also look at it from the perspective for a young brand, 40% is pretty awesome. I look at it from that perspective. That is growth potential. Growing from that 40% to north of 90%, that is our target. So a lot of growth potential there.
And that's what we have our unique Rivian go-to-market approach for. And I would like to double-click really on that go-to-market approach. And I think you can really appreciate it if you compare it to a legacy go-to-market approach as common in the industry today. And let me show you what I mean. And I would like to walk through a customer journey. Let's start with a legacy approach to go-to-market. Classically, you have from awareness, shopping, delivery, service, ownership. So that's the customer journey. In a legacy approach, you have different players. You have the OEM layer, and then you have multiple dealerships in the market. They're all working there. And I would really like to focus on the digital side of our business, the digital customer journey. And let me use one case study to show how this works from an ecosystem perspective.
You see a social media ad. You would click on it. You would land on the OEM's website. That's where you end. Then you go to the configurator, play around, think, this is an interesting vehicle. Do they have some of those vehicles available? Now, here it becomes interesting because inventory is owned by the dealers. So you would enter your ZIP code and then end on a dealership website. That's a whole different ecosystem, different data ecosystem. They will look very similar because, of course, it is within the brand guidelines of that brand. Going to have the same colors, fonts, et cetera. It's a different data ecosystem. And then you are on that dealer website seeing only their inventory. If there's another dealer in that town, you would need to go back to the OEM site, go to that dealer.
Of course, they want to keep you in their ecosystem. It's kind of their customer. Let's compare that to the Rivian business model. This is what our go-to-market approach looks like. It's one consistent omnichannel seamless customer journey. It's one data ecosystem, very seamless. I would like to use the exact same example I used for the legacy business model for our business model. You see the social media ad? You click on it, you land on rivian.com. By the way, just redesigned three weeks ago. It's beautiful. You should check it out. You will be in good company. There are 2.9 million unique visitors per month we have on our website. You check out all the great technical features. You go to the configurator, configure your vehicle, and then you click on shop. At that moment, remember, typically you would end up at the dealer site.
You stay within the Rivian business model. It's the same data ecosystem. Click on it, you see the inventory that is available. You can filter by powertrain, by color, by wheel, and do the entire checkout process online. By the way, also pricing is clear. There's no negotiation, no intra-brand competition with that other player in town who might give you a different discount. The price is there. You can check out online and do the entire transaction from your home. Very convenient. So this is all in place today. What we are adding is the pre-owned side. Currently, it's new inventory with Rivian now appearing on the used car place. We're adding used car inventory. And later in the year, we will also add the Certified Pre-Owned business. So vehicles that are used, but they are certified.
This is a very important downstream business for us, and it will help us on our path towards profitability. By the way, it also helps to increase residual values. That's always our target to have high residual values. Plus, we can address new target groups. A used Rivian is going to be more affordable than a new one, so we can address new customers as well. So that's the very important digital side of our customer journey. At the same time, real life is also physical. And especially for us as Rivian, we are a very experiential, a very physical brand. So let's look at the physical side of the customer journey. Our advantage is typically a dealer is both showroom as well as service. We can separate this. So we're doing this and we're creating showrooms. We call them spaces. Let's have a look at some of them.
You will see them here. All very unique, very Rivian. We have 12 at the moment. Enjoy the view. So this is the physical space where you can meet also our team, where you can see the products. And this is meant to drive awareness. So they are in high foot traffic locations. They really also help us to drive awareness, to share two numbers here. At the moment, we have more than 85,000 visitors per month on average in our 12 spaces. So generating a lot of awareness here. And they're also very effective in what they're doing. So we benchmarked at this place where they are, for instance, in a mall, what other lifestyle brands are in that area, and how is traffic for Rivian compared to the other brands. And we are 60% higher than brands in the vicinity of where we are.
So it's great to drive awareness. Next step in the customer journey then is orders, demand generation. We're monitoring that very, very closely. And I would like to share one case study with you. Last year in October, we opened a space in Atlanta. And looking at the orders in Atlanta, we saw an uptick of 30% in orders in Atlanta after opening the Rivian Space there. So those spaces are very effective along the entire customer journey. So we're thinking, how could we do more on the physical side while being very, very lean and don't spend additional money? So we looked at the other physical footprint that Rivian already has and thought about how can we use that physical footprint to drive more demand. So we began leveraging the service centers that we have more for demand generation. Here you see some pictures from our existing service centers.
And they're awesome for Demo Drives because I mentioned the spaces are in high foot traffic areas, often not well suited for Demo Drives. The service centers are. You can really do Demo Drives in the vicinity and neighborhood there. And we've begun adding salespeople then to the service centers. Because after that demo drive, that's the moment of truth. That customer gets out of that vehicle with a smile from ear to ear. So then we need to have a salesperson there that can essentially close the deal. And this is what we are implementing. We will have late summer in 32 of our 59 service centers, salespeople on ground to get when people get out of that demo drive, make them happy. And I can't emphasize enough how important those Demo Drives are.
Because our vehicles are so unique, getting people into those cars, let them experience what they are really capable of doing is a game changer. I also would like to encourage you to take advantage of the Demo Drives that we offer here. Please do that. If you can't do it today, go to Rivian.com on the upper right-hand side. There is book demo drive. It's just one mouse click away. You won't regret it. Let me conclude the physical build-out of our footprint. Here you see a map. This is our vision for end of next year. We are expanding our physical footprint. A year ago, we just had two spaces. Currently, we are 12, and we will stay very lean, very focused. Even premium brands have 200, 300 dealers in the United States.
Our target for end of next year, so in the next 18 months, is going a bit north of 40 spaces where we have white spots. We're also building out service centers while massively driving efficiency in service. We still need to build out those service centers because also our fleet is increasing. Units in operations is increasing. So we're going from 59 that we have at the moment to more than 100 by the end of next year. So I talked a lot about service centers a minute ago, and I didn't really talk about service itself. Let's change that. Let's talk about service. We have a job to do. We need to ramp up service, and we're doing that. I would like to explain our business model a bit more in service. And let me also compare here to a legacy business model.
That, by the way, is not only true for the past, but also for the future because of how vehicles are still being built. We have all been there. How does it work? We would drive to a dealership. They would physically connect their diagnostic scan tool into the OBD and start diagnostics at this very moment and find service issues, need to identify work that needs to be done. That all cannot be prepared because you don't have the information until the moment the customer is at that service bay. You embark on a process to diagnose, order parts, repair the issues. Our vehicle architecture, as also was seen showed both from the software side as well as video showed from the hardware side, is a game changer because we can remotely diagnose the vehicle entirely. So we know exactly what's going on.
Often, you won't even need to go to a service because there's a software fix. Even if you need a service, we can plan for it. We can have the parts there. It's much more efficient. There's one game changer because we know exactly what needs to be done. Typically, you don't even need to go to a service center. This platform enables mobile service. That has two advantages. A, customers love it. These are the service vans that would appear then at your driveway at home, the beautiful service van, our own vans built out completely for service. If a customer lives more remotely, we also have R1Ts built out. By the way, we have vans here, R1Ts here. Please check them out. Made for mobile service, leveraging those industry-leading 420 mi range.
We would come to the customer's home, do the service there, and go back. They can continue to work from home. They absolutely love that. We love it as well. You're going to love it because it is highly efficient. Because we have so many mobile work orders currently for Rivian, overall, we have 70% of service is mobile. That enables us to just have a smaller physical footprint. We don't need to build as many service centers, and we can build them smaller. Actually, for those 70%, this enables us to reduce the physical footprint by 49% versus in the legacy business model. It's a huge unlock. Helps customers. It helps us to be super lean. That's an awesome win-win. That's service. Let me conclude the Rivian business model, our unique go-to-market approach. I think we've built a very, very strong brand.
We're expanding it. We're taking full advantage of the direct-to-consumer business model that is built on digital first and with a scalable and exciting physical infrastructure. With that, I would like to hand over to Claire, and she's going to lead us through the financial roadmap. Thank you, Claire.
As part of this presentation, I want to spend time taking you through where we've been as a company, the scale opportunity that the mid-size platform provides us, our capital roadmap, as well as our path to profitability. RJ closed out his last section on our product development roadmap with the image of our quad motor badge of Gear Guard going 0 to 60 in under two and a half seconds. And we can take that same analogy over to our organization as a whole.
I've been here at Rivian for three and a half years now, and we've in that timeframe gone from zero revenue to $5 billion of revenue over the last 12 months in that timeframe. Importantly, over the course of those last three and a half years, we've built a scalable foundation in the organization. And so, as you heard this morning, we built our scalable technology foundation, especially as we're now building and producing our Gen 2 product here in Normal. And that same architecture that goes into Gen 2 will create the foundation for R2 and Rivian's future roadmap on a go-forward basis. Secondly, as you think about the operational growth and continuous improvement we've seen here in Normal. To give it perspective, when I came first to Normal three and a half years ago, Normal was a concrete floor. We had stamping presses in the back.
The mezzanine was being installed in our paint shop, but it was really the promise of what was to come here in this facility. So to have seen from that state to now producing, we produced over 100,000 vehicles to date. And we've now gone through all of our retooling efforts here within the Normal facility during the month of April, getting the plant ready for continuous improvement and an increase of 30% in our line rate that sets us up as well for the future integration of R2 within this facility. And as you just heard from Kjell, over the course of that last 3.5 years, we built all of the foundations for our go-to-market strategy, our digital-first infrastructure in our website design and approach to connect with consumers.
We built out nearly 60 service centers, 12 retail spaces, and as RJ mentioned earlier today, 500 charging locations as well. So to say that this is a team that has speed and agility would certainly be almost an understatement as we think about the journey that we've been on over those last three and a half years and how that foundation sets us up for what's coming next within the business. But before we get into Rivian's future, I wanted to spend a couple of minutes because I'm sure for many of you in the room, what's happening in Q2 is also top of mind. As you see here reflected, as I mentioned, we had our shutdown over the course of April within the plant and are now ramping up our supply chain for our Gen 2 product and ramping up production of that product as well.
We anticipate producing between 9,100 and 9,300 vehicles over the course of Q2, which is consistent and in line with our expectations. We continue to be encouraged by the progress that we're seeing here within the plant. Importantly, from a delivery standpoint, we anticipate delivering between 13,000 and 13,300 units. This is important as you look at the 4,000-unit gap between deliveries being greater than our production in the course of this quarter. Essentially, that reflects the fact that we've been able to sell through a large amount of our Gen 1 inventory balance as an organization with the vast majority of Q2 sales and deliveries coming from sales of our Gen 1 product.
That sets us up nicely as we ramp up production in the back half of the year with the Gen 2 product and get more and more of those Gen 2s on the road as well. We're also in a position to reaffirm our guidance from a production standpoint of 57,000 units of total production for 2024 and the expectation that we continue to have that we'll see low single-digit growth on a year-over-year delivery basis for 2024 as well. Today, we're also in a position to reaffirm our Adjusted EBITDA guidance of negative $2.7 billion as well, given all of the work that we've done to get our plant retooled products in place and the ramp that we see ahead of ourselves over the second half of the year.
RJ walked you through the composition of this slide earlier, but I wanted to come back to it briefly to touch on just a couple of key points. The first key point is that our material cost and our cost of goods sold within that is our largest cost within the organization. At our steady state, we expect that material costs will be roughly 75% of that cost of goods sold. So the critical importance of this material cost trajectory in Rivian's path to profitability is essential. I also wanted to provide a couple of callouts to make sure everyone understands how to read the chart as well. So as you look at each of the percentages for the key systems, this adds up to 100%.
So the way to read the chart is that 100% is reflective of how are you going to calculate each impact relative to the 20% overall material cost savings that we expect to see. It's not that each of those systems is changing by that specific percentage. So I just wanted to provide that point of clarity, as I'm sure this is a chart that many of you will come back to, and I'm sure we'll have many questions on in future conversations as well. As we look beyond Q4 2024, our profitability trajectory certainly doesn't stop here. We continue to see additional commodity tailwinds as we go into 2025 for our products as well.
What's not reflected in any of these charts are some of the financial impacts from material cost savings that we anticipate receiving as a result of our joint venture with the Volkswagen Group that will certainly enable us to reduce many of our material costs associated with our electrical ECUs that NVIDIA talked about earlier today as well. Beyond that, as you heard from RJ, we have seen significant opportunities for cost down in the sourcing of our R2 product as well.
Now, that gives us confidence that there's still more meat on the bone here with suppliers as we think about the further negotiations that we can have with R1 product and also the carrot that we've seen with many of those supplier partners that have been strategic partners of Rivian that are able to source parts for R2, but importantly, also have content on our R1 as well as our commercial vans. And so we've seen carryover savings and benefits to the business through those relationships as well. There are three key drivers in our path to positive gross profit in Q4 of this year, the first of which is our material cost, which we just walked through the details of. The second key for us is fixed cost improvement.
Here, given the increase of 30% in our line rate, and as you heard about earlier today, all of the design for manufacturability changes that we've made within the product itself, we anticipate seeing significant reductions in our labor and overhead costs to produce our units. Additionally, we'll see a reduction in our depreciation expense. We also expect to see increases in our revenue per delivered unit driven by sales of regulatory credits. As we stand here today, we have under contract over $200 million of regulatory credit sales for 2024. That gives us significant confidence in meaningful profit opportunity that will come through the sale of regulatory credits.
We've also seen many of our OEM partners leaning out away from their electrification plans, and this creates additional profit opportunity for Rivian through more competition for the sales of our regulatory credits that we've seen increase in value over the past few quarters as well. Kel also referenced the remarketing opportunity through the sale of used Rivians that will be launching later this year. So that becomes a new business opportunity for Rivian to continue to grow our top line and profitability potential as well. Next, our focus on extending our cash runway has also enabled us to look at opportunities to scale the business while driving cost efficiencies. Last year, to put it into perspective, we increased our revenue by $2.8 billion, yet our operating expenses increased by only $70 million.
As you heard from Wassym, we've looked for system and technical solutions to drive automation and efficiency across both our SG&A teams as well as our R&D teams. On the R&D side, it's important to note that while we've been going through a process of renegotiating many of our piece prices with our partners, we've also found that we've been able to be much more effective at negotiating our engineering design and development costs with those same suppliers as well. That is a significant portion of our R&D spend as an organization and has been an enabler of significant savings for us, especially as we approach R2 development in the future.
As we look ahead to 2024, again, we see the opportunity to continue to increase our top line revenue while also in this 2024 timeframe, bringing our operating expense on a cash basis, adjusting for our stock-based compensation and depreciation and amortization expense down in 2024 relative to our 2023 levels. This just comes back to the efficiencies that we're continuing to reflect across our operations. As you can imagine, following Kel's conversation, we need to continue to still invest and grow all of our sales and service opportunities ahead of us while digesting more efficiency across all of the other areas of the business. In addition to the OpEx savings, we've positioned ourselves to cut $2.5 billion out of our capital expenditure roadmap between 2023 and 2025.
We started last year with a CapEx outlook of roughly $2 billion and were able to bring that down to just under $1 billion. Even this year alone, we initially started with a CapEx forecast of $1.7 billion and were able to bring that down to $1.2 billion. The biggest contributor in our 2024 and 2025 CapEx savings has been the decision that we took to build R2 first in our Normal plant. That collectively, it's not just CapEx, but we've talked about that providing $2.25 billion of total savings to Rivian, driving efficiency.
And also, as you just heard from RJ on his description of the line layout, being able to leverage the core strength of our team here within this facility, all of our logistics infrastructure that's already set up and established, and importantly, the strength of the team that we have on the ground here in Normal to build and ramp that product for us. Over the long term, our impact as an organization really relies on the efficiency through which we can bring to market new products. So how we're able to more efficiently launch new capacity for our mid-size platform, given the large potential that it provides us, it enables us to continue to enhance the return on invested capital that we anticipate achieving.
And so while we've focused on material cost and really reducing the overall cost of goods sold as well as our operating expenses, CapEx is another key lever for us that we focused on as well over the course of the last few years and will be central to our long-term success in bringing and scaling R2 in the market. Next up, I wanted to provide a little bit of additional visibility into our Volkswagen Group joint venture that we announced earlier this week. In addition to the $5 billion of capital proceeds to Rivian, there are a number of key benefits. I spoke briefly about the material cost opportunity that this relationship has. In addition to that, we have the opportunity to leverage our operating expenses.
So as you think about it today, as you heard from both NVIDIA and Wassym, we've developed our full stack electronics, full stack software in-house. Now we have the opportunity to leverage those core teams to build core stack functionality and capabilities for not just Rivian vehicles, but for Volkswagen Group vehicles as well. And both sides will share in the expense of those teams building core technology for both organizations' respective vehicles as well. As we look at the cash contributions that are anticipated here, first, we announced the convertible note. We have received the $1 billion this week from that convertible note. And the next key item on the list are the $2 billion of consideration payable by Volkswagen Group to Rivian associated with the joint venture.
$1 billion of this payment will be paid at the closing of the joint venture and $1 billion of debt that's estimated to be available to Rivian in 2026. It's important to note that this $2 billion is intended to flow to Rivian's benefit. It includes funds to secure certain IP licenses for Volkswagen Group and to acquire Volkswagen Group's interest in the JV. We'll have many more details on all of this, and I'm sure there's lots of questions on this topic in particular. Once we come to definitive agreements later this year, and as we talked about earlier this week, we anticipate this being able to be closed in Q4 of this year. So we'll stay tuned as we're able to share more details.
Beyond the $2 billion of JV consideration, there's also an additional $2 billion of equity investment into Rivian that will happen throughout the course of 2025 and 2026. The 2025 investment will be based off of a financial milestone for Rivian, and the 2026 investment will be based on a technological milestone as well. Importantly, as we think about the efficiency in structuring this transaction, we have $2.5 billion of the $3 billion of equity investment in Rivian that is based off of future share prices for us to go and deliver against. As you can imagine, and you heard about us earlier today, we continue to be laser-focused on driving to positive gross profit this year and ensuring the start of production of our R2 platform in the first half of 2026 as well. The other key factor here is what does this capital provide for our business?
So the $5 billion of proceeds combined with our $7.9 billion of cash equivalents and short-term investments provides true foundation for scale. This allows us to ramp R2 here in Normal and, importantly, to build our mid-size platform in Georgia, ramping up over 400,000 units of future production volume for us, bridging Rivian to free cash flow positive and the opportunity to self-fund our future growth as well. Now let's take a little bit more of a closer look at our long-term production capacity. As you heard from RJ, we've designed the integration of R2 in Normal to share a number of key shared shops here, including our paint shop and stamping presses within the organization with a separate body shop and general assembly building that will be just behind our existing facility.
We'll be investing in 155,000 units of total production volume and we'll be increasing Normal's total production capacity to 215,000 units in aggregate. Now, if you add the bars under this Normal R2 launch, it does not add up to 215,000 units. And the why there is the operational flexibility that we'll have to flex up or flex down volumes between the respective lines to optimize around Rivian's total gross profit contribution out of the facility itself. We've had a lot of questions on how to think about the different volumes. So the directional guidance I would provide is that we would anticipate producing R2 on three shifts in Normal, R1 on two shifts in Normal, and our commercial vans on a single shift in Normal as you think about the relative volumes that we anticipate in our roadmap for the future.
Secondly, as we look at the opportunity ahead, R2 in Normal is truly just the beginning. We see R2 as being in addressable market across North America and EMEA representing 14 million units. That means for us that we see this mid-size platform as being a million-plus unit potential vehicle and platform for us over the long term. So while here we've outlined our path in Normal and Georgia, as I spoke about, there certainly are. This is not the extent of our mid-size platform ambitions as an organization as we see opportunity for international growth and additional scaling, knowing that R2 and R3 and R3X are not the only potential vehicles that we can also build on this platform for future growth as well.
Next up, as you heard from RJ earlier today, we wanted to provide more visibility into how we can bridge from our R1 Gen 2 product to R2. I hope each of you, through the detail that we provided in our roadmap, all of the efficiencies that we've seen, the sourcing efficiencies that we've started to see as well, have the ability to understand and build in this confidence roadmap for Rivian's future, knowing what a central role R2 will play in driving to scale volumes across the board and reaching and meeting Rivian's long-term targets as we think about the profitability potential for this product. The targets that we've shown here are also reflective of our normal R2 ambitions.
As you can imagine, being able to bring online additional capacity in Georgia or global capacity as well will only strengthen our position as we think about further cost sourcing efficiencies over the long term as we realize the full potential of this mid-size platform. Next up, while R2, as I spoke at length about, is certainly core and central to Rivian's future scaling and potential, R1 is also very important to Rivian's near-term path to positive EBITDA. As we look at the key drivers of this R1 trajectory from Q4 2024 - 2027, you see a slight increase in revenue per unit. A lot of that is driven by the fact that in Q4 of 2024, we won't yet have launched our quad motor configuration. There's more premium configurations from a mixed standpoint that will be coming in the future.
Equally, as we look at the fixed cost per unit, this is where you see the scaled advantages of introducing R2 into the Normal plant itself. In the Q4 2024 timeframe, as we provided our 2024 guidance, that anticipates 57,000 units of annualized production in the facility. As you look at 2027, you can estimate we won't necessarily have ramped up completely to the full 215, but you can imagine that this represents more about 200,000 units of overall volume that we'll be producing at that time here in the Normal facility, driving meaningful fixed cost per unit efficiency for not just R2, but importantly for R1 as well as our commercial vans.
As I also spoke about, we're not done yet as we think about the variable cost improvements that we'll continue to see as we source R2 and use the carrot of that additional business to take R1 costs lower as we fully realize more of the commodity tailwinds we've experienced in the product as well. Again, importantly, both the last slide and this slide don't reflect some of the additional electrical BOM savings that we'll continue to see through our joint venture. Where does that bring us? That brings us to a position where in 2027, we expect to be EBITDA positive as an organization based off of the production volumes here in our Normal plant.
On the gross profit side, I gave you a little bit of a look into the trajectory for both of our consumer platforms in R1 as well as R2 and how that contributes to a gross profit contribution pool that can more than offset our operating expenses within the business. From an SG&A standpoint, as we look to 2027, we'll still be in growth mode as we're introducing and ramping up production of R2. And so we'll be continuing to invest in all of our go-to-market functions, whether that's our build-out of our service centers, our sales organization as a whole, and that will allow that SG&A to continue to grow.
But importantly, we'll also see R&D savings given the fact that we're now in a position to leverage all of the core technologies that we've built as an organization as we scale out R2 production in Normal, and we'll be in a position to look ahead to Georgia and think about the continued growth from a capacity standpoint that will leverage the same design and platforms that we've built as an organization. This roadmap also doesn't contemplate the potential operating expense savings that we would anticipate through the joint venture with Volkswagen Group. Once we're at definitive agreements with Volkswagen, we can provide additional visibility into all of the financial implications into our model and future roadmap as well. But as we look at SG&A and R&D in aggregate, you can imagine that these collectively enable Rivian to drive additional scale and efficiency into the business.
Together combined, we expect to see low single-digit growth on an annualized basis in our operating expenses as we scale for the future. In summary, we have a strong balance sheet and visibility into our capital roadmap enabled by the announcement that we had earlier this week with our joint venture with the Volkswagen Group. We're poised at the precipice of R2's launch in the first half of 2026, being the true mass-market high-volume program that we're all excited to drive and experience in the future. Finally, I hope all of you walk away with more visibility into Rivian's path to long-term profitability. With that, I wanted to leave you with our long-term financial targets. We see a clear path to driving to 25% gross profit margins, high teens adjusted EBITDA margins, and approximately 10% free cash flow margin.
With that, I want to invite RJ up, and we'll do some Q&A on Rivian's financial forecast, business outlook as well.
Great. Thanks, Claire. In similar format, we'll take questions from the room.
Hi, Claire. Hi, RJ. Doug Dutton here from Evercore ISI. First, just a question on the waterfall chart. Are the battery costs in propulsion or electrical and other in that chart?
The costs are in a few different areas, but the one that you saw captured is propulsion. That's drive units, that's the power electronics, high-voltage power electronics, and the battery systems.
Sure. Okay. And then just on your LFP packs, is there a worry there that you're going to get hit with some of the 25% tariff from Biden on the import from CATL on the LFP packs? Is that worked into these numbers?
The tariff structure that's gone in place with the way we've sourced the battery today, we're already paying the tariff on what's going into vehicles now. I think the bigger question is what happens with trade with CATL and others in China and whether or not those batteries are even allowed into the United States. Fortunately, because of the way that we've laid out our platform and our architecture, we do have flexibility that if we wanted to use our high nickel cells for a smaller pack, we could do that. As is the case with battery, we've built a lot of flexibility into the organization and into the architecture such that if we needed to make those changes, we could.
Thanks, guys. A question just about, and congratulations on the funding. As you think about the tranches and how it's laid out, the equity tranches for 2025 and 2026, could you just talk to us how you think about dilution? Initially, you think the stock will be higher because you're going to hit milestones and there'll be less dilution, but just how you guys ran through that in your process of structuring it like that. Thank you.
Sure. As we thought about the advantages or opportunity that we had of structuring our roadmap, we wanted to create the opportunity for us to continue to execute against our plans. So certainly, there are things outside of our control, but as you heard from us, hopefully today, we're laser-focused on managing what's within our control as we're driving costs down throughout the business, creating and improving the product itself, and driving up demand to support our roadmap for growth.
Oh, great. Colin Langan from Wells Fargo. When you look at the waterfall from going from Gen 1 to Gen 2, 45% sounds amazing, but then isn't the price also falling by a similar amount? So don't you need to take more out? I mean, so how are you going to get to that being very profitable if the price is coming down with the cost? I guess I think I'm misunderstanding something.
Well, you rightly point out the price is a lot lower to achieve that price with the margin structure. Claire just talked about the 25% gross margin. We also have to take a lot of cost out. So the 45% is building off of the Gen2 R1, and that represents the material cost. What we didn't talk about, there's also meaningful cost savings in the non-bill-of-materials COGS. So that's your conversion cost, that's our logistics cost, and importantly, and often not thought about, that's also our warranty cost. And so the combination of those non-bill-of-materials COGS, cost-of-goods-sold items plus the bill of materials allow us to get to a cost level that supports the price point we've talked about with R2 starting at $45,000 and the margin structure that we've talked about.
Is there any rough split between the two? How much is the one buckets? Is there any split between the two that you had mentioned, the material cost and then the non-COGS? Is that like a separate?
We haven't provided it, but I think you can sort of back out with pricing and then what we've built in terms of the cost down roughly where we expect those to be. But there's big, I guess, one of the things that's harder to capture, but hopefully is being seen. R2 has a lower bill of materials, but it's also a lot easier to build. And so the amount of labor hours embedded in each vehicle will be lower. And we see that show up across every aspect. The body, the general assembly, drive units, chassis system was a really good example. And you look at the front suspension, the MacPherson strut goes in with a couple of fasteners of double wishbone, electrohydraulic damping, air spring suspension is a lot more faster, has a lot more complexity to assemble.
That plays out across the rest of the vehicle as well.
Hey, Jason Getz, Mizuho Securities. I wanted to ask a little bit more on the VW agreement, again, on the 2025 and 2026 equity investments. It looks like those require certain financial and technology targets. Can you talk a little bit about those targets and what they entail? Are there specific timeframes within those years, or are they flexible? Just trying to understand the dynamics of the agreement a little bit more. Thanks.
Sure. As we think about the milestones themselves, on the financial milestone, the way to think about it is essentially that we need to achieve that milestone as the gating factor for that 2025 payment, but we would have the opportunity for that to go later should that milestone not be met. But as we think about our roadmap, as you can imagine, we ensure that we set milestones based off of management's target levels as well. And so this is no different than what we've talked about today as we think about our Rivian's continued roadmap to profitability as an organization to provide a little bit of a lens into how to think about the achievability or feasibility of that target itself.
As we think about the technical milestone as well, it's important to note that Rivian will have technical leadership over the joint venture with Rivian appointing the Chief Technology Officer of that organization. So while the milestone will be based off of a JV-related technical milestone, we certainly have control over the execution. And again, similar to the financial target, I believe it's something that we can clearly have line of sight into achieving as well in the timeframe.
Mark Delaney from Goldman Sachs, thanks so much for hosting this Investor Day. A question on the demand environment. You spoke a lot around the new features that you've introduced with the refreshed version of the R1 platform, better range, lots of power, the exciting motors. Kel talked about some of the demand response you've seen with the new stores and experience centers that you've opened up. You spoke around rolling out leasing into more states on your last earnings call. So maybe help us better understand what you've been seeing in the demand environment of some of the recent things like the refreshed R1 is helping with bookings.
Sure. Yeah. On the Gen 2, we'd been talking about this in, call it the last four earnings calls, and we'd always focused on cost. It was so hard for me to resist talking about all the features and improvements that were coming as well. We were really excited to be able to unveil the product a few weeks ago. We had a lot of media experience of the vehicle. It had an overwhelming positive response. There was an overwhelming positive response to what we've done. Accordingly, that's led to a lot of excitement from customers. I would say we're really encouraged by what we've seen in terms of customer feedback. Of course, preceding that or precipitating that is the positive feedback we've had from the journalists and media that have been through the vehicle.
We're now at that point where these are starting to be delivered to customers as well. Just as we've seen all along with the Gen 1 vehicles and the full fleet of R1s that are in the world, word of mouth is incredibly powerful. So we're excited about what's in front of us with Gen 2. We're excited about the opportunities to continue driving improvements to the product with the software and technology that we talked about. It underpins us continuing to maintain the guidance that we've had for the year with confidence.
Hi, George Gianarikas from Canaccord Genuity. I just wanted to, I know you're still working through the details, but understand the potential for OPEX transfer to the joint venture. I mean, will Wassym potentially work for the joint venture or for Rivian? And to the extent you're still negotiating, is there the potential for OPEX to transfer to the point where you become EBITDA positive maybe in 2026 before 2027? Thank you.
You want me to start? Okay.
We can tag Dima.
We'll tag Demo on this one. So as we think about the OpEx potential, as I mentioned, today, Rivian is paying for all of the investment that we have across our software and electronics spaces as a whole. And so the foundational work that those teams have done then creates the platform for the work that will happen within the joint venture for R2, for Volkswagen Group's platform of products as well, and will be leveraged on a go-forward basis. And so as you think about it, we'll now be in a position to share, and we'll get into more details when we get into the definitive docs, but just want to give you a little bit of a framework to think about that shared operating expense that creates that core stack for the JV itself.
Then any expense related to if there's a work that's just dedicated to Volkswagen Group products or if there's work that's just dedicated to Rivian products, each of the respective parents would be paying for that work within the joint venture. The operating efficiency is really driven by the fact that we'll now have the ability to share in that core stack work, but also we'll be in a position to add incremental features to our products that one side or the other may be developing as well within the joint venture. That's a key enabler for us as we think about all of the joy and innovation we can bring to our customers. It's also going to really help us as we go international, given that our products and teams will be establishing that foundational work for their global platform as well.
It's also worth noting, and we've all alluded to it, but I want to be explicit here, the continued emphasis on driving cost efficiency into the platform and the way that the joint venture is set up is such that the cost associated, the development cost associated with further optimizing for cost, further optimizing around component set or chipset is something that will be shared amongst not only Rivian, but with Volkswagen Group. The beauty of this is it applies a lot of volume onto a platform that we've developed. So the volume benefits implicitly will lead to lower costs in negotiation with suppliers, but we'll also continue engineering technical solutions that drive even more efficiency into the system.
Yeah, Tom Narayan, RBC, thanks for taking the question. So right before your guys' conference call you had after the VW deal, VW had their own conference call. And one of the big questions that came up is, what about their Scout brand, which would feature competing directly head-on against you guys? They seem to be very committed to that brand. Just curious how you think about how that plays out.
Yeah, part of the foundation of the joint venture with Volkswagen Group is to take, as Claire just said, and as you've heard from us, take our platform, take our software stack, and apply it across the breadth of Volkswagen Group brands. So Porsche, Audi, Lamborghini, Bentley, of course, Volkswagen as a brand, and eventually the Scout brand. We really deeply believe that it's important that customers start to have choices. And today, particularly as you look at lower price points, as you look at if you're going to buy an EV in the sort of $40,000-$50,000 price range, there's a remarkably small number of choices that exist there. And that's evidenced by the extreme market share that you see Tesla still continues to carry.
In order for us to go from, take the U.S. market, for example, to go from 8% EV adoption to 100% EV adoption, which needs to happen over the next decade, we need to have a lot more choices. The number of choices that exist in terms of form factor, brand presentation, product attribute trade-offs needs to look a lot more like what we know in the ICE world, where you have hundreds of choices. The way we think about the opportunity is we're competing against the 92% of customers who haven't bought an EV. In that price point, in the $45,000 price point, there's really specific form factor needs. There's very specific utility and functional needs. There's brand and design characteristics that customers are looking for. It's one of the reasons we're so bullish on R2.
I think hopefully you've seen the market reaction to R2 is because there's a need for a product that fits within that brand portfolio. The other thing I'd say is the way that the vehicle feels, the look, the feel, the brand experience is independent from the computers that are running beneath the surface and the operating systems that are running on top of those. The user experience, the user interface in a Porsche will feel very different than a Rivian, or in an Audi will feel very different than a Rivian. And that not just in the pixels, but actually in the overall UX, the number of screens, the proportion, size, shape, form factor of the screens, even the interface, our inputs entirely through multi-touch or their physical controls. Those UI/UX elements are on the top at the applications layer.
That's not part of the core shared platform. The reason I say that is I want to be really clear that this is not a situation in which all these brands will have the exact same feel. The feel unique to the brand. It's a decision from the design team and the product teams that's going to drive ultimately how they feel in terms of what the customer interacts with.
Hi there. Thanks a lot for this great event. I was just at a couple of questions about the VW topic. One was we talked about the leverage on the cost side. Could you just talk a little bit more on how the output of the JV will be monetized? Let's say VW suddenly decides to adopt the architecture across a huge volume. They have huge volumes. So how would that affect the JV and how would that be recognized in our books? And then the second question was, this is about electrical architecture, but there's also powertrain technology that we spent a lot of time talking about. And there's obviously a lot of work done on the autonomy side. Were those topics that were on the table or you didn't want to talk to them about, or were they not requested?
Can you just help us understand that, two things?
Yeah, we as a company have looked at partnerships through a variety of different lenses. A larger shareholder today, Amazon, is a big strategic partner. And that's manifested in lots of ways, but the most visible is we designed and developed a van that we believe is by far the most optimal solution for commercial applications. We've looked at ways to share platforms, to share drive units and batteries. But one of the things that's unique as we start to think about partnership in the context of automotive is a gating element to collaboration is often the communication of electrical systems. So that's platform sharing. This is like the first topic of beyond, like do the vehicles package similarly? Can the vehicles' electrical systems integrate?
For that reason, or for many reasons, along with that, we've focused our partnership with Volkswagen Group around the network architecture, the ECU topology, and software. That is the scope of what we're doing. It doesn't include the drive units, doesn't include battery, doesn't include autonomy, doesn't include vehicle platform opportunities, doesn't include anything around the go-to-market side of the business. But that's the most foundational element of relationship and partnership. As we spent a lot of time with the senior team at Volkswagen, it also represents, we believe, hopefully it comes through here, one of the most important and hard to replicate aspects of moving to, call it newer technology vehicles and moving to a software-defined vehicle. So that's the scope of the joint venture.
As a company, there's lots of other things we could look to partner with or to leverage our drive units, as you called out, battery systems, vehicle platforms. But the beauty of this joint venture is it really leverages what we think of as the hardest to replicate or the hardest to copy aspect of what we've built. Now, in terms of how it's recognized, I think there's a question that you embedded in there, which I want to call out is this partnership and the scale of it hopefully demonstrates this is to deploy our platform widely within Volkswagen Group. And as we said in their investor call that they had right after the announcement, they've talked about the first product coming out in the second half of this decade, but that's the first of a string of products that will come.
We've spent a good deal of time with their teams. This isn't something that we just came to with the cursory level amount of diligence. There was a lot of work at a detailed technical level validating that, in fact, the sort of tapestry of tier one ECUs can be replaced with a much simpler, much more advanced zonal architecture that we've developed. So we've done a lot of work there. What has us excited as a technical team and as an organization is we've this chance to scrimmage, so to speak, with the technical teams within the brands at Volkswagen Group gives us a lot of confidence of just how this will play out, a tremendous amount of confidence.
The alignment from the top between myself and Volkswagen Group CEO Oliver around our objectives, our joint objectives, which translates then into the brands, which translates into the technical teams. The pull from the technical teams within each of the brands to get our technology into vehicles quickly because of all the positive aspects you heard from with Wassym and Vidya is a backdrop that we couldn't ask for a better backdrop than that. We're bullish on deploying this in as many vehicles as we can. And then, as you've heard a lot, leveraging the benefits that come from that in terms of economies of scale and sourcing and overall cost structure.
Revenue?
Do you want to talk about, Claire?
The way to think about it is I spent a little bit of time as we talked through some of the JV consideration and that $2 billion going towards both background IP as well as Volkswagen Group's ownership stake in the JV itself. We also spoke about some of the benefits being related to revenue in the future. But given the materiality of their organization and the collaboration that we'll have together with them, in the near term, as we focus on Rivian and Volkswagen Group, you can think of it more as a cost center for both parties. And over time, as we think about the roadmap for the future, this is certainly an extensible platform that could be shared or provided to other parties in the industry as well, creating high margin recurring revenues through potential licensing.
Hi, guys. James Picariello, BNP Paribas. My question is on the bridge to 2027 as it relates to the 5% increase in revenue per unit. Because in 2027, Rivian should be delivering over 100,000 R2s, right? Which will, of course, have a much higher price point than the R1. So yeah, just any clarity in the assumption or the implication there?
One point to clarify, that slide is related to just R1. So that was really looking at, as we think about Q4 2024 to 2027, what is the trajectory for R1's revenue, fixed cost, as well as variable cost?
Got it. If I could just sneak one more. The 2024 guidance had this deliveries or implied deliveries growth of low single digits. Just based on what you've seen through the first half year, it does seem as though the R1 is selling quite well relative to that low single digits. Just curious if you have a bias change of any kind to that rate for the year.
No, staying consistent with the guidance that we've provided historically and reinforced today as well. It's important to note, as I mentioned, in Q2, the vehicles that we're selling are largely all Gen 1 product. And so as you think about the production volumes we have right now for Q2, those will be out in the market being sold over the course of Q3. And so there is a trajectory from a ramp perspective and trailing effect as we think about the delivery cadence over the course of the year that won't necessarily immediately get up to scale delivery volumes as a whole. And so that's really reflected in; it's not a lack of demand confidence, it's really just the dynamics of a new product ramping up within our facility and the timeline associated to get those into the hands of our end customers.
Thank you very much. Colin Ducharme with Sterling Capital. I wanted to hone in on the unit pricing maybe from a different angle. That would be just relative pricing positioning versus vehicles in same category and maybe link that to financial assumptions as we roll forward in the next few years. Very fundamentally, you're building a brand, especially as you begin to penetrate with the R2, the mass side of the market. Where do you want Rivian units priced relative to other EVs out there in same category versus ICE vehicles in similar categories? It's important for us as investors when we see fairly fluid pricing environments as we've seen in the EV space, particularly in the last few months, to make our own assumptions on where Rivian needs to slot itself and its products as we see those competitive actions.
So do you view your relative price positioning as correlating to those market fluctuations or not? And then how will that change your financial assumptions in the next few years on the unit pricing? Thank you very much.
Yeah, I mean, in the context of R2, this is an important point. So the average transaction price of a new vehicle sold in the United States is just under $50,000. And the R2 pricing was designed to really sit around the average price of a new vehicle. And the most popular segment for new vehicles in the United States is a two-row, five-passenger crossover SUV. And so R2 is everything about it has been thought about as being designed to be cross-shopped with the 92% of things that are not EVs. From a size of the vehicle, pricing the vehicle, of course, it's impossible not to compare it to Model Y. But there will be cross-shop with Model Y, no doubt, but it's contemplated in every way to present a very different type of choice for consumers.
I would say this, I've been surprised at how little product differentiation there's been at that price point, meaning in part because of Tesla's success with Model Y, a lot of the products that compete in that segment have very, very similar form factors. Jeff, our head of design, has this really interesting picture he put on the wall in the design studio, and it shows the centerline profile of all the vehicles in this class. If you just take a step back, so you're maybe five, six feet off the wall, and you squint, they all look the same. The centerline's very similar, the proportions are very similar, the seating package is within millimeters of one another.
You can understand why the success of Model Y has led to others saying, "Let's create something in that ilk." We've been very intentional to make sure we're building a brand that gives customers a different choice other than the emblem on the front and the back and the lighting design on the front and the back. I think that's important. I think more brands need to make sure they're creating uniqueness in terms of package, design, aesthetics, product attribute positioning. So what are the trade-offs you're making between performance, on-road, off-road, storage, capability, function? We've seen that play out in the premium segment as well, where there's certainly less buyers just given the price of where R1 is. The addressable market is naturally an order of magnitude smaller than what we'll have with R2. It's the leading market share vehicle in that segment today.
So the best-selling R1S is the best-selling large electric vehicle today in the United States. We'd love to see that market share penetration in the $45,000 segment. But that was because of the same approach we took to decisions on that vehicle in a very analogous way. The Model S, or I'm sorry, the Model X and the R1S are both seven-passenger SUVs, both have four wheels, both have four doors, very similar pricing. But they couldn't be more different in terms of how they present a product option to customers. And that's not to say one's right or wrong. They're just occupying very different demand spaces. And we've attempted very much to do the same with R2, R2 relative to Model Y. And Kjell gave a statistic before that I'll repeat here because it's so important on this point.
Our mission is to drive the world towards a fossil-fuel-free future in transportation, but really broadly as a global economy. To do that, we need to have customers of our cars that are not buying EVs. We don't want to just be playing a game of sort of moving the existing EV customers around to different EV brands. So more than three quarters of our customers today, their purchase at Rivian is their first-time EV experience. Our hope is to continue that exact same trend with R2. In fact, higher, I'd love R2 to be 90% of our customers have never owned an EV before. That means we're pulling people out of a wide array of SUVs and crossovers and passenger cars in that space.
That was a long way of me saying we think that the pricing that we've laid out with R1 is appropriate. We've been very disciplined around maintaining a cost structure that can help in a healthy way support those price points and not in any way confusing the excitement around the product with the allowance of more BOM cost or more conversion cost. We've been very disciplined to say we must hit this COGS, which then supports a healthy margin on a price point that we've talked about.
Hi, guys. Thanks. Alex Potter with Piper. So if you can help me walk through, I guess, the mechanics of modeling this joint venture. Right now, if I'm looking at my model, it's pretty straightforward, right? There's a number of vehicles that you sell times the price, then you subtract out all the costs, and then there's EBITDA, right? That's straightforward. Now there's this joint venture coming in, which I guess in itself, it's maybe its own little entity that has its own revenue, its own P&L, but it's unclear how I'm supposed to take my Rivian model and change it going forward. Does the output of that joint venture become your COGS? And then there's an equity line that I'm supposed to account for the joint venture later on. And I've gotten a number of questions also from clients about the future revenue coming out of that joint venture.
Are they selling a system to you, which then you count as COGS, but the joint venture counts as revenue? And is the same going on with Volkswagen? So just any help you can provide on kind of the mechanics for modeling both the joint venture and Rivian's own P&L going forward would be super helpful. Thanks.
So Alex, one point that is complex is we are not yet at definitive agreements. So whether or not we will need to consolidate the JV within our financials is still yet to be determined as we work with our accounting partners and internal teams on the assessment there. It is a jointly controlled entity, as we've talked about. But because of that, it will have different essential outcomes in terms of the modeling and the consolidated versus unconsolidated outcome. So we are going to have to defer some of those specifics till when we can circle back with everyone when we're at definitive agreements and can provide more clarity on the roadmap forward.
We've got time for one more.
Thanks. Hey, guys. Mike Shlisky of DA Davidson here. Claire, your comments on becoming EBITDA positive by 2027, I wanted to clarify, is that just in vehicles or when we think about certified pre-owned service charging or any of those a big piece of that? And if we could just drill in a little further, what size or number of chargers do you think you need to have in your network to become EBITDA positive for that piece at least? Thank you.
Sure. The commentary around EBITDA positive in 2027 encapsulates Rivian's full business. So that would be everything from vehicle sales to charging to all of the core services that we have within the organization as a whole. Kjell spoke a bit about the service opportunity that we have. We'll have the charging network in there. We have financing, insurance. We've launched our connectivity or Connect+ offering, which is connectivity and infotainment features for consumers as well. So that encapsulates the full opportunity as we think about our path to positive EBITDA in 2027 itself. Maybe I'll pass it over to RJ if you want to comment on more specifics around the charging business.
Yeah. I referenced this earlier and I spoke about uptime, which is a customer-facing benefit, but one of the other elements about the charging business. In fact, I should say this: if you were to ask me 10 years ago how I would have predicted charging to evolve, I would have predicted that more investment would have flowed into it and you would have had a handful of highly effective networks that would have mirrored in some way the ways that the gas station infrastructure was built out in the United States. So pre-1900, there were essentially no gas stations in the United States. By 1910, there was like 100,000. Today, there's 160,000. And car companies didn't have to really invest in them. They were built on the demand side of vehicles becoming more popular.
What uniquely hasn't happened here is there's been very little investment in this space and there's really a surprising opportunity for this to be a space in which there's less than five, maybe two, maybe more great networks. We believe, of course, we believe this wouldn't be doing it if we didn't, but we deeply believe that we will be one of the great networks in the system. We're in the early days of building it, but the decision to vertically integrate those chargers was a decision based around cost and based around uptime.
By building them ourselves as opposed to what you see a lot of these other networks that exist outsourcing the hardware, our structural cost of installing DC fast chargers on locations and then the ability to iterate the design of those DC fast chargers over time is something that really is, I would say, the Tesla Supercharger network, which is an outstanding network and our own, have that opportunity. Then in terms of path to profitability, it doesn't require a large number of stations. It requires high utilization of each station in and of itself. The full objective is driving very high levels of utilization to the network. It's not as if these things need to be used all day long.
We haven't put numbers out there on this, but for a relatively small number of charging events per day, each charger itself becomes profitable. So key for our roadmap and very different than how the reason I gave the history of charging infrastructure, very different than how this looked 10 years ago, is we now have as a society, we have an installed car park of EVs that allow us to access that or we can access that car park to give us sort of instant scale. So opening up our network this summer to be an open we call open RAN, open Rivian Adventure Network, means that not just Rivian's, but Tesla's, Ford's, a plethora of brands can access our network and we get to benefit from the size of the installed electric vehicle car park that exists today.
We can be highly intentional as to where do we put our charging stations. So we see this as something that will absolutely be profitable for us, but importantly, because of the strength of the hardware, because of the way we design the equipment, we also believe it's a real opportunity for us to carve out a significant market share within this space.
Great. That concludes today's Q&A. I'm going to pass it back to RJ for some brief closing remarks.
Cool. Thank you, Claire. All right. Well, thank you, everyone, for joining us for this morning session. I know this was long. We sat in this room. Everybody's still paying attention, taking notes. Lots of hands raised still at the end for more questions. So for those that are dialing in remotely and aren't here, I'm sorry, this is the end of our time together. But for those that are here in person, we have a lot of fun planned for the afternoon. So we'll have tours of the plant. And we debated how to do this with this size of a group. We'd originally said, well, let's do a walk on the mezzanines. We're actually going to allow people to get on the floor. We'll have carts that allow you to go throughout the plant.
I'm really excited for folks to be able to see what we built and what I'd call attention to, and I referenced this before, is we've now gone through several launches. We launched our first products, the R1T and the R1S, a little more than two years ago. Since then, we did a major update on the EDV last year. We just completed a major update on R1. We've launched multiple drive unit lines. And what we've built in terms of our ability to plan and predict those ramps is remarkable. And as you go through the plant, you're going to see sort of vintages of how Rivian approached manufacturing and design for manufacturing. And the drive unit lines, when you get over there, are perhaps the best illustration of this. It's the newest part of our plant, and you can see the Enduro line.
You'll see right next to it the Ascent line. Take a look at just the way it's organized. You're going to be flying through there on a cart, but try to do your best stopwatch time studies. Look at the utilization of the team members. We continue to make progress in how we think about utilization of our lines, OEE of the lines, the material flow within the lines. I'm excited for you to see that. Then following that, we'll have some time in our delivery center. You'll get a chance to test our products and drive them. Of course, I highly recommend driving a quad. But for those that didn't get their questions answered, I saw a lot of hands up.
We'll have, in addition to the folks who are here speaking to the business today with me, we'll have a wider selection of the team that's available across these different activities. So engage, ask the questions. You'll find the team's not only excited to talk about it, but has deep knowledge of these topics. And hopefully, we can cover all the questions you have about the business. With that, thank you everyone for joining the morning session and looking forward to the rest of the day.