Aurora Innovation, Inc. (AUR) Earnings Call Transcript & Summary

Okay. We're going to go ahead and good start it. Good morning, everyone. Welcome, and thank you for joining us today. I just want to give a special thanks to all of you out here at our autonomous trucking terminal in Palmer, Texas and those via the webcast for joining us today. We really, really appreciate your time and interest. We have a very exciting program for you today. We're going to take a deep dive into our roadmap to launch Aurora Horizon, our autonomous trucking service. First, our CEO and Co-Founder, Chris Urmson will provide a brief overview of Aurora, our market opportunity and our business model. And next, he'll share some great technological progress we're making with the Aurora Driver, review our key milestones on our roadmap and outline the framework we committed to sharing, which will enable the investment community as well as all of our stakeholders to measure our progress toward launch of Aurora Horizon. We'll then have a safety discussion in the format of a fireside chat with our Vice President of Safety, Nat Beuse; and Dr. Jeff Runge, former NHTSA Administrator and Chair of our Safety Advisory Board. We think this will be a really valuable discussion to talk about a critical aspect of self-driving technology development and deployment. And next, Sterling Anderson, our Co-Founder and Chief Product Officer; who will review some progress we've made on hardware advancement as well as our vehicle programs. Kendra Phillips, Vice President of Service Delivery will give an update on our autonomous trucking pilots as well as a look into how we're building out our operations. And then lastly, Chris will close out with an overview of the regulatory landscape for autonomous trucking, and then we'll open it up for Q&A. Before I turn it over to Chris, 1 housekeeping item I need to note. You're all familiar with this, but I have to read it. We will be making forward-looking statements within the meaning of the U.S. Private Securities Litigation Reform Act of 1995 in today's presentation. These statements are based on current expectations of the management of Aurora and are based on assumptions that are subject to uncertainty and changes which may cause actual results to differ materially from those expressed or implied in such forward-looking statements. More detail can be found on this slide, which is available for download, along with the balance of our presentation on our Investor Relations website, which is ir.aurora.tech. A replay of the keynote presentation will also be available on our website for 30 days. So welcome all of you to watch that. And with that I would now like to introduce our CEO and Co-Founder, Chris Urmson.

Thanks, Stacy. Stacy always gets to do the fun part of the presentation. Well, welcome, and again, thank you all for joining us here at our South Dallas terminal. Really appreciate the opportunity to share with you what we're up to at Aurora and keep you abreast of our progress. I want to make a particular thank you to some of our partners here. So we have Eric Fuller here from U.S. Xpress. We have Matt Culham, who is the Chief Engineer on the vehicle program with PACCAR. [indiscernible] from Volvo Autonomous Solutions is here. Stephanie Cook from FedEx is here. And then a little later, we're going to be joined by Matt and Jason from Covenant. So please seek them out over here. I'm sure they'd be happy to share with you their experience working with us. We're going to have a little bit of a keynote session right now, and then we're going to get you in the back to meet some of our team, see some of the technology up close and personal. I think give folks an opportunity to experience the truck on the road around here in south -- just South of Dallas. We also offered folks to get a ride down to Houston. I know at least one person took us up on that. So have a fun day with that. But it should be a good experience and look forward to you learning from that. So Aurora was founded in 2017 with the mission to deliver the benefit of self-driving technology safely, quickly and broadly. When we set out to build the Aurora Driver, we really set out to build a driver that would work across different vehicle types. But of course, we're focusing now on delivering Aurora Horizon, our product in freight first. And we've built this driver in a way where we can transfer the capabilities of, we developed on light vehicles to trucks, to trucks to light vehicles, we showed the case that transferability in Q1 of this year. I think I skipped something there. The Aurora drivers, this combination of self-driving software, hardware and data services that enable vehicles to drive themselves. Again, it's designed to be operate across Class 8 trucks and light passenger vehicles. Trucking and Horizon will be the first product. [Presentation]

So we've brought you here today to share our vision for how we're going to deliver this and how we can transform this concept into reality. While many companies have been focused on ride-hailing first, we've been focused or we're focused on delivering what we think is the fastest way to build a business in this space, and that's through scale and through freight. As you know, the America's highway system is really very self similar. A mile of freeway in Texas looks an awful lot like a mile of freeway in California. It looks an awful lot like a mile of freeway in Minnesota. So as we deliver the technology that operates here, we expect to be able to scale it and operate it across the country. Longer term, we intend to leverage this technology into ride-hailing as we expand from driving on the freeway and our terminal to terminal type operations, our trucks will then span to go from warehouse to warehouse. Our passenger vehicles will expand to drive from airports to convention centers and hotels and ultimately into city centers and all the way between stores and distribution centers. The opportunity in freight is incredible. The U.S. market is about $700 billion. When you compare this to ride-hailing today, which is about a $35 billion market, it's a much bigger opportunity. Over time, we do want to lose track of that ability of personal mobility because we expect that personal mobility space be about $1 trillion opportunity eventually. We focused the last 6 years in actually building the core technology to capture this opportunity. We've also invested heavily in building partnerships because at Aurora, we want to focus on the thing we can do best and then find like-minded partners to work with us to scale and deploy the technology. We see immense opportunity for -- we've got these incredible partners in PACCAR and Volvo. Together, they are 2 of the top 3 truck manufacturers in the U.S., and they deliver almost half the vehicles in the U.S. We also work with incredible partners that take those trucks and vehicle -- there's trucks and drivers to do work in the world. FedEx is the largest less in truckload carrier in the U.S. Winner is a top 5 full truckload carrier. Schneider is one of the largest multimodal carriers in the U.S., and newer freight, of course, is enough incoming significant technology forward freight broker. We also work closely with US Xpress and Covenant and collaboration. So thank you again for those of you who've been able to join us here today. First and foremost, Aurora is a mission-driven company. We're motivated by the incredible opportunity to improve safety on the world's roads and on the U.S. roads. It's fundamental to everything we do here. In the U.S., roughly 500,000 truck accidents happened last year at about 5,000 or 5,500 people died in truck fatalities. The vast majority of these were due to the -- are addressable with the technology we're building at Aurora. We can make this much, much better. We see an incredible opportunity to improve the businesses. So the challenges that our customers face first is a shortage of drivers. We're delivering a scalable, stable driver supply. Hours of Service limit how much our customers can use their vehicles on any given day. This will allow them to get better revenue per vehicle and allow them to move freight more quickly. Fuel cost is one of the largest contributors to their cost structure. We expect these vehicles to be significantly more fuel efficient. And then, of course, higher insurance costs continue to increase. We expect both the combination of safer operation and better visibility into the events that happen on the road will allow us to manage and bring those costs down over time. Solving these pain points makes a whole lot of near-term value for our customers. It allows them to have better safety. It allows them to move freight more quickly, that leads to better revenue generation per truck and reduced operating costs. This will allow us to improve both their top line and their bottom line. To meet the needs of our customers and most effectively monetize our technology, we're using a business model of driver-as-a-service. This model is basically a subscription model where we expect really in the long term, these Software-as-a-Service type revenue structure or margin structure and we're going to be providing this technology to the fleet owners and basically pay as you use it or pay as you go away. This model aligns very well with what our customers need. It increases their revenue potential. It allows us to operate really as a valuable technology partner and supplier to them rather than a competitor in the space. They get more - they operate their trucks more that allows them to have bigger revenue opportunity that increases our revenue opportunity as well. It allows them to have really consistent pricing. The variable part of the cost structure becomes less variable with us. And the customer continues to operate their business as the way they know how to today. right? We're slotting into it in a way that feels natural. They have drivers on their fleet today, we provide a driver to them. We're going to talk a bit more about the 3 parts of the pillars of our technology of our product, the Aurora Driver, which is the part that's kind of constantly improving every day. The operations service and delivery this is the part of our system that enables our customers to actually integrate the Aurora Driver onto their businesses, and then, of course, the truck platform, the work we're doing with our amazing OEM partners. At the end of last quarter, we shared with you our roadmap between here and actually delivering the Aurora Driver commercially. And we're -- we've got basically 3 milestones between here and commercial launch, getting to feature complete, delivering the Aurora Driver at a point where it's ready to go and then, of course, launching it. Today, I'm going to focus first about the Aurora Driver. Now as you think about driving vehicles on the road and operating our service, imagine taking a trip with me. We're going to head out of the terminal. We're going to drive down surface roads. We're going to get on to the freeway, along the freeway we have to interact with traffic. We may encounter emergency vehicles. We may encounter debris in the road construction. We have to be able to navigate all of that. Ultimately, we'll exit the freeway, get back on the server streets and pull into terminals. That's the set of capabilities we've been developing at Aurora, and that's what you can see reflected in our roadmap. Over time, we release new capabilities, we are also constantly improving the features that we've already have on the vehicles. Many of you may have been here last year and got to ride. You will see it's another experience today with the vehicles. And you'll unfortunately probably not get to experience most of what we've been working on over the last year because these are the more rare events that are out there on the road. Here are some examples of some of the things we've been working through. So here's the Aurora Driver, navigating a lane closure. So here, it's dense traffic. I think this is down near Houston. The Aurora Driver is slowing, creating room, understanding this driver on the left is trapped, leaving space for them to cut in and operating in this kind of heavy traffic. Here's another example where we're having to deal with contemporary construction science. This is on the drive between Dallas and El Paso, driving into the setting sun. On the side of the road, we're seeing a sign that's reducing the speed limit. The Aurora Driver understands that and reflects up on the top left here, but the speed limit is reduced from 75 to 65 miles per hour. In Texas, they had this amazing road structure where we were able -- I'm sorry. [indiscernible] So that's one of the capabilities that is actually -- do we have [indiscernible]? Today, are we able to drive on the service lane? Yes. So I think that's next on the road map, right? But yes, we'll drive partially in it today. We won't move all the way into a fully over into the service lane.

[indiscernible] We will have Q&A opportunity at the end. So we'll just take that questions at the end.

Yes. Thank you, Stacy. And as I'm saying in Texas, we have this great road structure, they call it a Texas U-turn or Texas Boomerang. This is the ability of basically change direction around the freeway without having to come to a stop. This is a technology we developed first on the Toyota Sienna and have now transferred back to the truck. So here, the driver is coming to the left side lane, makes the little turn here and then has to merge into traffic as it comes around the corner to get back and head in the opposite direction on to the freeway. Of course, we have to deal with other drivers on the road. So in this case, the Aurora driver is attempting to make a lane change the vehicle in front of us or from behind us, swirls over and is making a lane change aggressively to the right. The driver is making a split-second decision that it's going to kind of pull back, create space, and then once it's safe to do so, move back into the left lane again here and then continue safely on its way. We shared with you at the end of Q2 that we've been -- delivered our fault management system. This again is the ability for the Aurora Driver to understand when something is broken, take the appropriate remedial action, in this case, pulling this side of the freeway and stopping, getting itself out of harm's way and reducing the risk to the drivers on the road. And again, this is not of those sexy things you think about in self-driving, but it's one of those critical elements of actually delivering a product. We talked in our road map about this quarter, delivering -- managing foreign debris in on the road and also driving on repainted or construction shifted lane markings. So we've delivered both of those. In this video, what you can see is, and we'll wait for it to loop back again here. But what we'll see is driving towards this chunk of road here. There's a bit of debris in the center of it, the vehicle detects it right about now and then able to make a little shift to the right around that bit of tire carcass. And then for added complexity, of course, there's a stopped vehicle here that had happened to encounter on the side of the road, and it makes a nice courtesy lane change to create space around that vehicle. So on our path to commercialization, we have these 2 major milestones for Aurora Horizon. The first is to get to feature complete, which we're working towards at the end of Q1 and then to be Aurora Driver ready, which we're working to at the end of the year. Feature Complete means that the system does everything we think it needs to do to launch the product on our initial launch lanes in Texas. We'll note that it's ready because at that point, we won't have any more policy interventions. Policy interventions are where we ask our operators preemptively to take over because we know that the driver can't handle certain situations on the roadway. We'll expect everything to be able to maybe not fully reliably, but at this point, operate in a way that we have confidence that we can unleash it on the road. Aurora Driver Ready means that our safety case is complete for everything except for the base vehicle platform that if we had that platform, the driver could be out on the road running safely. So how are we going to share with you the progress we're making between Feature Complete and Aurora Driver Ready? Well, we committed to providing a measure for this and what we're going to share with you is our Autonomy Readiness Measure. And this will allow the investment community to follow along and allow us to continue our tradition of being the most transparent company in the AV space. So we've been saying for a while that safety case is the critical thing to actually having a commercially relevant product and that we will be ready to launch when the safety case is complete. So as a reminder, the safety case is this structured argument as to why we have confidence in the safety of the vehicle on the road. And at the top level, it's made up of these 5 things. First proficient that when the vehicle is operating in nominal conditions, it's driving well. Staying in its lane, it's making space for other vehicles. The second is that it's fail-safe. It understands when something is failed or broken, is able to take the appropriate remedial action to keep itself and others on the road safe that is continuously improving that we have process in place to close the cycle so that the driver gets better, and we get better as a company and we learn from our mistakes, that is resilient. The things that we may not have expected to encounter in the road that we handle gracefully and keep the world safe and then that the company is trustworthy. We operate with integrity. Well, the combination of these things we'll document all of this, and we'll have a collection of evidence that gives us the confidence that we've done the engineering work. We've done the process work. We've done the business work to deliver a safe product. And what we'll be sharing with you is the percentage completion of the safety case framework. So this is -- will begin at the end of Q1 through -- we'll begin with when we're Feature Complete through commercial launch. We expect that by the time we get to Aurora Driver Ready, there will be about 95% is where we'll be with this Aurora readiness -- Autonomy Readiness Measure. 95% that remaining 5% is about the final bits of integration with the autonomy-enabled truck platform that we have to do once we have those in hand. And then by the time we get to commercial launch, that safety case will be 100% complete. And we're targeting Aurora Driver Ready at the end of next year. In addition to sharing what we think is the launch bar, right, that fundamental measure of are we done yet. We're also intending to share with you a measure that allows you to understand indicatively how the Aurora Driver is performing on the road. And the measure we're going to provide for this is percentage of miles in autonomy. This is kind of how good or how much the time is the vehicle operating. Now this measure is going to be made up of 3 factors. So for each mile that we operate, we're going to ask -- basically did everything just work the way it should do the easiest possible kind of most obvious measure of this. So it's everything operating normally. And if yes, then that's operating in autonomy and we'll count that in the metric. The second is, did we -- were we able to address whatever may have been off nominal using Aurora Beacon by providing input from offboard? If that's the case, then that also will count towards this percentage of miles in autonomy because that is commercially relevant how we're operating the vehicle, how we intend to operate Horizon and operate the service. And then finally, what will assess is anytime that our drivers have an intervention on the road or disengage the system, we'll ask the question, was that required and we'll do off-line analysis using a combination of simulation and other analysis tools to assess, did they actually need to take over. Now this turns out to be a subtle but very important point. So we train our operators to keep themselves, keep the vehicle and keep others on the road safe. And we want to make sure that we don't create this weird perverse incentive where they think at any moment, like, should I take over right now? Is this going to mess up our metric externally? We don't want them to ever have that question. We want them to do the right thing on the road, do the safe thing on the road and take over. And then after the fact, we can ask the question, did they need to or not. And so we'll be tracking that. And we think this is really an important part of having safe culture, right, is making sure folks don't have weird incentives to behave in ways that don't result in safe outcomes. Now it's important to understand that this percentage of miles in autonomy won't get to 100%. And it's not a launch bar, right? No matter how good we are, things are going to happen on the road, a tire will blow out. A piece of hardware might fail, something where we can't solve it remotely or through just pure software on the vehicle. And so while we expect this to trend towards 100%, it won't actually get there. And it's not really our bar for launching. It's not a bar at all for launching. Our bar for launching is that autonomy readiness metric is our safety case 100% complete. And so that's kind of the measure we're going to be sharing with you. We think it will allow you to have really good clarity into how close we are to launching and the progress we're making. And so with that, given how important safety is to us at the company, and how that's one of our #1 priority. It's right there at the beginning of our mission. I'd like to invite Nat Beuse, our VP of Safety, up to talk -- have a little bit of a fireside chat with former administrator, Jeff Runge, Dr. Runge. Nat, I've known since his time in government. I think he was probably one of the first people in the federal government that was actually thinking about automated vehicles. He's been an incredible asset for us at the company. He's allowed us to really understand how regulators think, allow us to engage with them, allow us to have built a very strong report. So thank you, Nat. We'd love to have you up here and Dr. Runge.

Good. I think better on my b*** than doing my feet.

Good morning, everybody. So I'm Nat Beuse, spent many years at the agency in a bunch of different programs, everything from consumer information to regulation to basically leading the agency's efforts on AV, cybersecurity and basically everything else in between all of that. And actually, Dr. Runge here is a former NHTSA administrator, who was one of the first administrators that actually worked for at the agency and because we didn't currently work him hard enough, the time he came to the agency was right around the time SUVs were coming into their own. If folks remember, that was also around Ford Firestone, and there's a lot of interesting things happening in the vehicle safety space then. But he was board with that and decided he was going to go over to Homeland Security. Shortly after that got stood up after 9/11. And so he's had a really long career in a bunch of different aspects of safety, and I've been honored to kind of have known him I guess, almost 20 years now. It's been a while. And we're going to have a conversation just about some of the things that Chris talked about, but some of the things that I think we both believe are kind of fundamental to AV safety. So Jeff Runge, thanks for joining us on here.

Absolutely.

I think one of the things that we've talked about before is this idea of our Safety Advisory Board. At least for me, personally, I find it always helpful. to assume you don't know everything. And it's always helpful to have outside perspective sort of help shape what are your gaps and what are you missing. And so this idea of the Safety & Advisory Board really kind of serves that purpose. We're in the day-to-day working on the vehicle, working on the tech. And it's always helpful to have a group of experts, guides you, and I'm going to turn it over to Dr. Runge to kind of talk about his perspective of the Safety Advisory Board and why did you join it? And what value do you think it provides to Aurora?

Thanks, Nat. We only have 20 minutes, so I can't expound to in much detail on this. But I will tell you that I've been -- I was an emergency physician working in one of the country's busiest trauma centers for 2 decades. And my research was in traffic, injury control and trauma care. And frankly, I got burned out because of the revolving door of trauma that kept coming into the emergency department and road crash injury was the leading cause of those injuries. It's still the leading cause of death in Americans between the ages of 2 and 34. And the fact that there was not a national crisis declared around that motivated me to leave my practice and come to NHTSA. There were 43,801 deaths on the road when I got there. When we left, we introduced some regulation that required technology, namely electronic stability control and side curtain airbags, that technology reduced -- is largely responsible for reducing the number of crashes down to about 32,000 once those -- once the vehicles got into the fleet and those things started to change. So I was convinced that technology could be the answer. Human behavior is very difficult to change. I'm the Click It or Ticket guy, okay? I get the fact that we need to do human behavior change. But technology has to step in when human behavior is not capable of doing this. So I've been very bullish on autonomous driving now for a long time with basically hopefulness. Because I don't care if you're disabled, if you can't see or you can't hear or you're too old or you're too reckless, Autonomy has a solution for this. So when you ask me to come and join back in the Uber days, I was skeptical, but hopeful. And what I'm seeing now, and I think everybody on our Safety Advisory Board would agree, we've got people from an audit background, aviation, product development and aviation, trucking, road safety. And everybody is convinced that, and Chris, I think you sort of alluded to this, that the safety case is the business case for this, okay? We -- and you have to be safe. And I joined the Board for that reason, and it's why I'm here today.

Yes, I think that's right. And I think when I think of the product expertise that we have on the board, One of the points that you've made to me a couple of times about the safety case and why it's relevant is the fact that it's a claims-based argument, right? So it's not this idea that we don't understand what we need to solve. You actually break all of this down to these different claims and then you do the work to answer the claims. It's a very powerful tool from that perspective. And I think, to me, when I figure out and think through the value of such a thing, it was very new in the AV space. Most of the folks in AV space for the last decade or so I've been talking about it, it's some other metric about how safe and human and all of that. And I think we fundamentally that's actually not the answer. The answer is you actually have to do the engineering. This is an engineering problem, go solve it. And I think other industries have used this technique and what we've done is kind of modified it. And so I would ask you, it's not that simple, though, right, it took you, the rest of the Board and myself many conversations that actually really unpack the safety case and why it's so valuable.

Right. And the safety case is kind of jargon and maybe I should explain a little bit. But you think about this, you are investment analyst, you get it, it's a spreadsheet. And Column A is a series of claims of what the Aurora Driver can do. It's a long list. And each of the other columns beyond Column A is the evidence that those safety claims have been met. So it's a matrix of claims and evidence. And the safety case is built upon capabilities that they want to attain -- and every -- there are milestones in the safety case where they cannot proceed beyond the next stage of development until the safety case is made, which makes Nat's job very difficult, but it also instills confidence. And when the Safety Advisory Board probes convince us that these things have actually been met. And that reminds me that, look, we have to commence ourselves first. And so they're not -- it starts with safety. Safety is not only a reason to have this, but it's also -- it has to be safe if the benefits are going to be attained.

Yes. One of the things we've done in our safety case is incorporate both product aspects, organizational aspects and operations aspects. I want to talk about the organization piece for a minute. One of the things that's been very powerful in other industries is the idea of a strong safety culture and kind of how you manage risk within a company. You can have a bunch of processes, but how do you know you're actually using those processes? Are they actually effective? Are they working? Are they doing the intended thing that you thought they were supposed to do? But another thing you have to measure is actually the employees themselves. The employees we actually believe in the things that you're saying and how do you actually know that -- and so one of the things we've done within our safety case is incorporate this idea of a safety management system. This is an idea that's really -- I would say a lot of people point to aviation as being like the poster child for safety management system, meaning you had very safe aircraft, but we're missing a lot of different pieces in order to really make that system what it is today that we all kind of take basically for granted. And so we're installing that now today and Aurora have actually shared that with others in the industry. And I wanted to ask you, Jeff, because you've been in a lot of other different areas as well. How do you see the power of the safety management system really playing out?

Yes. So once again, I want to make sure that the jargon is understood. Big S, big M, big S, safety management system is an internationally accepted standard for how to manage safety across an organization. And Aurora is the first company in the automotive space that I have encountered that actually has fully adopted safety management system, or is working on it. You're very, very close. We have aviation experts who go around auditing airlines around the world to see if they are compliant with safety management. This has basically 4 pillars. It's safety policy. Your policies in place where safety has its preeminent role in your operations. Second is the ability to manage risk, understand risk, manage risk put in countermeasures for risk and hazards. And then the third one is safety assurance where you actually measure if what you're doing is actually meeting the mission of the organization. And if your risk management procedures are actually working to manage risk. And then the fourth one is what you just mentioned, which is safety promotion, externally, internally. How do you beat safety? Safety management culture, if you go into a company that has a safety culture and you ask where is the safety office. Everybody is self-identified, it's right here. It's not down the hall and to the right. So it's complicated. It's transparent. It's open. You can look at safety management system. It's -- it was promulgated by the international aviation -- Civil Aviation Authority, ICAO. The U.S. was slow to adopt, but the FAA requires it now. It's also spread into maritime and rail, Coast Guard uses it for their operations. So it is -- it's a great target and a methodology to manage safety across your company.

Yes. And I wonder if you could talk a little bit too about how you've seen this play out, particularly in the auto sector, where this isn't paid attention to and kind of the ramifications of that?

I mean, listen, I'll start with my pedigree. I mean there's no safety management system in medicine. And yet the Institute of Medicine came out and said, well, it's a crisis of medical safety. Well, first of all, there was not a safety culture in medicine when I was practicing, it was blame. It was malpractice claims. It was all that -- safety management is a -- it's a permissive environment where everybody's job is to root out problems and solve them.

Yes. I think that's really, really powerful. I think one thing we also talk about and how we use the safety case is this idea of transparency. So one of the ways we're leveraging that safety case, as Dr. Runge mentioned is even with the Safety Advisory Board, we can have very detailed conversations about very specific claims. The other way we're leveraging that is actually also with regulators and with the public. So not everybody has the same questions. And so what this allows us to do is actually share that framework, and we have shared that framework for them to zero in on whatever particular thing they want to talk about. So for some stakeholders, it's like tell me how you understand what operational design domain you're in and that you actually have evidence to prove that you understand that. For other stakeholders, it's more around tell me about your safety culture and how you know your employees are reporting safety concerns. And for yet others, it's like, "Well, just tell me that you meet the rules and I'm good to go. And so we're able to kind of service all of those stakeholders with this kind of universal framework. And I want to talk a little bit about public trust. So Jeff and I in our career when we work together and even now more directly, there's always this aspect of public trust that's really, really important and motor vehicle safety, partly because the public has a really that perception of risk, particularly when it self-imposed risk, but they kind of get very agitated when it's the risk that someone else is imposing on them. So what do I mean by that? Driving a car is likely probably the one of the riskiest things that humans do on a day-to-day basis. And yet, as Jeff pointed out, we're up to 40,000 deaths a year, and it's kind of like not the necessary action, at least in our humble opinion, that there should be. And yet, if there was a bus crash today, and no one got injured or one person maybe got unfortunately, fatally killed, there'd be front page news across the country about that one particular crash because it was a bus -- it was a driver and it's kind of like, well, there was them, not me kind of attitude from the public. And so we're in this space with self-driving, where we have to work really hard day in and day out to explain what we're doing on safety in a way to convince the public and build that kind of public trust. And Jeff, I wanted to know if you could talk about a little bit from what you've seen. What are the things that we're doing right in terms of bringing that public trust and building that awareness.

Well, first, let me talk about the ideal, okay? So you build public trust with consistency, predictability and transparency. When -- in driving, there is a -- you depend upon the predictability of what other drivers are doing. And this is going to be a learning curve with public interacting with autonomous vehicles. 95% of motor vehicle crashes involve human error. So if you can take the Aurora drivers potential capability out you will create a very predictable, very consistent vehicle that people will learn how to interact with on the roads, much more predictable than the must thing that cut you off this morning on I-45, okay? Now you saw it, you were ready for it. You saw them coming, and you backed off. I saw Chris' demo where he saw somebody -- you are driver -- sorry, he/she saw somebody cutting in front of them and they stop back to a low space. I don't always do that. I kind of want to punish them a little bit for like being a jerk, okay? So the Aurora Driver is intended to be predictable and consistent. The transparency piece is a piece that we talk about all the time. Right now, nobody has a clue what you all are doing to safeguard safety to make the -- to drive the capability out of the Aurora Driver, right? So at some point in time, you're going to have to share not just the story, but the spreadsheet with -- not really the spreadsheet, but the idea -- what's in the claims that you make. I've made the case all the time that -- and I love NHTSA. I talked to them every week. It was the best job I ever had. But the regulatory scheme that NHTSA uses for automobile regulation certification will not cover this. its self-certification. If they can pass a test, they can sell a car. This is more like how the FDA approved drugs where a drug company comes in and says, "My drug will do this. These are the side effects, these are the benefits. We know from testing that we've done ample testing Phase I, Phase II, Phase III studies, here are the data on our claims". And the automobile regulatory scheme that we have does not anticipate this. I don't know what's going to happen, but I do think that through that sort of leadership role now that Aurora has developed that you've developed through your whole career in autonomy, that you're in a position to sort of lead that discussion among government regulators.

Yes. I think it's an important point to end on is -- and I think Chris is going to talk about this later, but it takes a long time to do regulation. Even if you started today, Jeff started one on electronic stability control...

It took us what 5 years, right?

Yes, that I had to finish and then I've worked on other ones, backup camera is one of them that took nearly 10 years, 10 years. And so this is a journey, I think, is what Dr. Runge is trying to say, and we have to start that journey at some point. And I think we believe the way that we are operating now in the current structure is the way to do it and to really deliver this technology in a way that first and foremost, safe, but in a way that's very transparent on what we're doing. We're not hiding behind the magic stat. We're actually full showing all of the things that we're thinking about and inviting folks to challenge on whether we've actually thought about the right things. So thanks, everyone. I think now I'm going to invite Sterling Anderson to come up and talk about some of the great progress we've been making on hardware.

Thanks, Nat. All right. One of the hallmarks of our developmental approach over the last several years has been our willingness to invest strategically in not just the software but also the hardware. When done right, this co-evolution, this joint development of hardware and software can be really powerful. Can you all hear me okay? Yes. Okay. It can be a really powerful tool. It not only accelerates our path to market, but it also allows us to do it with a far more performance integrated system than we could otherwise do or we had called together set of off-the-shelf solutions, many of which don't even exist at quite the specifications required for a problem like this. We've included in our hardware roadmap, a high-level overview of the key phases left between here and commercial launch. In the next few minutes, I hope to unpack with you a couple of illustration of both the progress that we've made, but also define what each of these phases means. The first of those phases that you see here, at least in the time frame that we've shown is where we are or were last quarter. This is with an autonomy development platform, a kit of self-driving hardware designed to be representative in all the ways that are meaningful to the self-driving system. For things like sensors, this is the range, this is the resolution for things like compute, it's the bandwidth with which to power the software that we're developing. But importantly and notably, this development kit was not designed from the outset for reliability in the kind of harsh environments of the road for trucks or for the kind of redundancy failover mechanisms, self-health monitoring that's required to operate without a human fallback. That's this Phase II. And I'm excited to note that over the last several months, the team has made substantial developments in the development -- substantial progress in the development of this driverless ready hardware kit. What I mean by this is by incorporating both the functional safety systems, the health checks, the error checking, the failover mechanisms and even the redundancy throughout the system, whether it's in the computer, the sensors, the electrical systems to allow it to operate without that human fallback. And this is actually rolled out in almost 2 dozen vehicles that we've designed, developed, produced and introduced to our fleet over the last several months. This is actually in operation today in those fleets. The final step, and I recognize now it's a little hard to see at this resolution is what we're calling the hardened driverless hardware. What I mean by this, it is taking that driverless ready hardware that we've just integrated into our fleet and now hardening it. What I mean by that is preparing it for both manufacturability but also high levels of reliability that it requires to operate in these networks at high uptime, so these trucks are performing for our vehicle -- for our logistics customers. Along the way, the team has not only developed driverless capability. They've not only developed and made significant progress on the hardening of that capability, but they've also realized several -- I'll call them surprise improvements to some of the performance of these systems that have been really meaningful. And I'll show you a few examples of where that's happened. To start, though, before I show you those examples, I just want to orient you physically as to what we're talking about when we say the Aurora Driver hardware. What you see here is a truck rendering, effectively integrated with the hardware called out in blue from the sensor modules. This is everything from radars to LiDAR, to camera, to IMU, to GPS and a variety of other systems required to perceive both the world around it, but also keep track of the health and safety of the system itself, but also the computer, the networking systems, the electrical systems required to tie all of this together. The computer that we've rolled out, you'll see here on my right, your left, has been designed to -- in these trucks has been designed to allow a safe operation with a human fallback and also to withstand the harsh conditions, the trucks it operates will experience. To achieve this driverless capability, this redesigning to introduce error checking, self health monitoring required for the computer to know when a failure has occurred as well as key redundancies and failover mechanisms to fall back to in the event that one does happen. In addition to this, the team has actually -- we joke about this computer looking relatively bomb-proof. But the team has invested heavily in the robustification or the hardening of this computer so that it can operate per the demanding exacting automotive standards stipulated in ISO 16750 set of guidelines. And we'll talk a little bit about some of the test and development that went into doing that. but it's really, really important for the kind of high uptime environments for a highly performant computer like this to operate that it is designed for that ruggedness for that reliability. And that's what we mean when we talk about that final stage of hardening in the computer already operating in our trucks, it's gone through much of that. And so the team has made tremendous progress there. My apologies. I'm going to show you on the left here, another example of what's happened here in preparation of this driverless-ready hardware. This is specifically the camera system. So the team has also made substantial progress on the development of our cameras to both quadruple their resolution, but also to introduce our proprietary image processing pipeline that's allowed us to realize substantial better dynamic range and performance out of these systems. What I mean by the dynamic range is effectively the ability to resolve high differences and brightness between different elements of seen. What you see here is a couple of examples. The image on the left is a very common one in the automotive space. As 2 megapixel imager run through an image processing pipeline provided by some suppliers. It is -- this is an image that the Aurora Driver would have seen through that imager. On the right, you see the images that are running on the Aurora powered trucks. This is incorporated into the hardware that's in our vehicles operating today. In fact, the vehicles that you ride in today have had these -- what you'll notice is a couple of things. First, kind of obviously, tone mapping and things get a little better. Dynamic range gets substantially better. Your ability to resolve both kind of the darkness of the tunnel and also the brightness of what's outside of it is a really important element of that progress. And then finally, you'll see that kind of that much better resolution obviously translates to better performance of a perception system that's based on it. Our proprietary FirstLight LiDAR, I'm going to point you to the rightmost screen now, and we'll kind of do this stance back and forth. You can watch the left one if you want. But it's designed also for this ruggedness. Also to ISO 16750 standards. It's introduced a number of other things, both the kind of fault handling and error checking required for that driverless-ready capability, but also getting ahead of this robustness in extreme operating environments. So up to and including its integrated liquid cooling, self-health monitoring, but also integrated cleaning systems. So the sensor is able to keep itself clean in the event that it's operating in weather conditions, crickets are a thing here in Texas right now, as you may have observed walking in. Crickets bugs, other issues that would otherwise obscure a sensor. This integrated cooling system has been designed to help us deal with. But a final note here, and this is one of those kind of surprising moments that we realize as the team developed this is the priority here was make this system redundant, make it capable of driverless operation and make it rugged. It was not Hey, increase our lead, increase the lead and performance that this sensor gives us. If you recall, this sensor, Aurora's FirstLight is based on a novel approach to LiDAR -- to the use of light in detection and ranging, which is called frequency-modulated continuous wave. You may hear it referred to as FMCW. The benefits of FMCW, by virtue of the way that it modulates light and the way that receives and measures it is -- are manyfold. Among them, though, is this ability to perceive at much longer ranges than traditional pulse modulator units can do. In fact, a year ago, we showed an example of juxtaposing what the FirstLight LiDAR sees relative to what a traditional pulse modulator LiDAR unit see. And if you recall, and you're welcome to look at old materials, but that juxtaposition showed a full 7 seconds of additional time beyond what a traditional unit would have seen that the FirstLight LiDAR can actually see, resolve, detect and prepare us to respond to enactor a full -- at highway speeds is the full 2 football fields further along the road. As the team developed this driverless capability and kind of this robustification of the sensor, one of the things that came back with was, "Oh, hey, by the way, we've increased our lead". We're no longer looking at a 7-second incremental lead time to a traditional sensor. It's now about 8.5. At highway speeds, we're now pushing 280-ish meters, so nearly 3 football fields. That's not 3 football fields total, that's 3 football fields in addition to what a traditional system based on pulse modulated LiDAR would have seen. What you'll see in the trucks today is a set of effectively range markers. And I want you to watch those. And what you'll see is even before this sensor, you'll probably see a lot of tracks of vehicles that we're detecting out to about 500 meters. If you look at those as kind of those blue boxes that will give you a sense for the kind of perception ranges that are so significant not only to the safe operation of these trucks, but also the comfort of that operation as it relates to how similar they are to a human in terms of the way they respond to things that are significantly down the road. So I mentioned a few examples in the last few pages of -- in the computer, in the cameras and the LiDAR where the team has made substantial progress in terms of kind of the ruggedization and hardening of our system, but it goes beyond that. In fact, across the Aurora Driver system, our hardware team has been developing for several months now in an environment that replicates the harsh conditions of the road. This is everything from vibration testing, the thermal testing to water ingress testing to combination tests, shake and bake, we call them or other things where effectively it is exposing and giving our engineers the opportunity to make small changes in refinements to the hardware that make it operate highly reliably on these roads. And this -- the result of this, I know I'm saying these words like reliability and ruggedness a lot, it matters. The value to our customers and our partners is the Aurora Driver is able to operate at high uptime. It could operate without these hours of service limitations, but it can only do these, and our customers are really going to care about this, if it can do so reliably if it can operate for years on end in a highly reliable way. That's the significance of some of the development that the team has made over the last couple of quarters. So that's on the vehicle hardware side. Now as it comes to kind of the relationship between the hardware and the platform, what we've also been doing is begun the work with our OEM partners in preparing this hardware and this integration for launch. And I'll talk in a minute about the platform development that we're doing. But along the way, what we've also done is built up a fairly sizable operation in Pittsburgh, I mentioned just over the last couple of months. It's nearly 2 dozen trucks that we built, brought up and integrated with the fleet, and we've done it through this operation that's allowed us to prototype pilot and test everything from the creation of these harnesses at much more rapid speeds than the industry would otherwise be able to do to the development and integration and build of these trucks equipped with this hardware. So this video shows, I think we may have looked once so all advanced, but it shows some examples of where that's happening. I should note that, that work is solely in preparation for that launch. And as Kendra will mention here in a minute, as it relates to our operations, as it relates to our hardware development, there are many of these things that we do internally now to move faster, to develop a more performance system. We don't intend to do them long term, but what it allows us to do is bring in our OEM partners and say, this is how we've approached this. We've worked through these kings, bring in their manufacturing experts and it makes for a far more seamless introduction into their production lines and processes than we could otherwise do where we did just kind of cut straight over without that ramp-up. Final element of the roadmap that I'm going to highlight before I'll pass it off and we can talk about kind of the service delivery, it's the truck platform. And I'll share kind of a little bit about the evolution of these platforms. Again, the roadmap calls out a few different phases. First is kind of the autonomy development truck platform. That's what you've seen today. That's what you're ride-in today. These are effectively stock trucks that have been upfitted with a set of changes required to allow them to operate with a self-driving system but not necessarily operate without a human fallback, a prelaunch truck platform that we're phasing into, we'll start to introduce some of the platform changes required to operate that safely without a driver. And then finally, the scalable autonomy-enabled truck will form the basis for the launch platform, and I'll talk about what each of these means. Before I do, I do need to call out the couple of OEM partners that we're working with on this. in a tightly integrated co-development, not only of the technical side and the engineering of these drugs, but also on the commercial side. [indiscernible] is here from Volvo. He runs some of that for them, the business side, the deployment, the servicing and even the manufacturing. So this is a wide-ranging development that we're doing together, specifically designed to meet the needs of the Aurora system. As Chris mentioned, Volvo and PACCAR are the 2 of the top 3 truck manufacturers in North America. Both partners have deeply experienced development teams, sizable support operations. They've got extensive history and relationships with customers that matter. Feel free to ask Eric later if he cares. But I think if you're a carrier, it matters that the OEM is willing to stand behind the product that they've provided to you. And this is something that this tight integration is allowing us to support. Now to be capable of driverless operation, it's not sufficient that we simply take these effectively today stock trucks off the line, I'll fit them and deploy them. for a couple of reasons. Primarily, they need a set of changes to control interfaces, thermal systems, electrical architecture, cybersecurity architecture, steering and braking systems to introduce the kind of redundancies, failover mechanisms and other systems that are required to operate these safely at large scale in a driverless capable way. So what that looks like then as I mentioned, you'll see out here in the trucks you're right in these autonomy development platforms. built basically on stock trucks. What we've developed over the last several months with several months, I say several, several could mean a lot of things. We've developed since beginning and even before these partnerships were officially announced early last year, we've been developing with PACCAR and Volvo the set of changes required for this prelaunch platform. Now one of the most significant areas of risk for these platforms when we commence them. And one of the conversations that I had with leaders of these companies was -- our -- we have to secure award commit to a timeline key suppliers for key systems. Some of these key systems include the steering system. They include the braking system, and to -- that was widely recognized and acknowledged as one of the key risks for the program when we started. We were pleased to announce in August that we've retired that risk through the awarding of the suppliers for these major components that we were initially concerned about. Not only that, I'm pleased to report today that over the last month, we've actually produced a set of vehicles with one of these systems integrated by the OEM from the supplier. And it's begun its testing in the preparation of these trucks for that driverless capability that's required ultimately for commercial launch. Now when we talk about the prelaunch platform, that's effectively what this is, just think of it as a transition platform that allows us to integrate many of these systems, whether it's redundant braking, whether it's a variety of other systems required for that launch platform in a phased way. And it's being done in conjunction with our partners at PACCAR and Volvo as well as with the suppliers that provide those systems. Ultimately, that will result in -- what I've shown here is a scalable autonomy-enabled truck. Now you'll notice that it's got a couple of holes in it. This truck initially on production will have been built with all of these systems that we've developed with these OEMs and all of the interfaces as well for the Aurora Driver hardware. This is a thermal hookups. It's electrical systems and interfaces, it's logical systems such that effectively, the Aurora Driver sensor modules will be integratable directly with that truck, the computer will drop right in and the truck will be ready for business. In the long term, post commercial launch, we will be working with our OEM partners, not only on this kind of scale production of these trucks that are prepared to receive the Aurora Driver, but actually will start to line side of the computer, the sensors and the rest of the system such that it rolls directly off the line and into the hands of customers who have purchased those trucks that's not actually even shown on that roadmap, but that's where that goes post that initial commercial launch. And with that, I'm going to introduce Kendra Phillips. So Kendra has joined us, if you don't mind, brag about you for a minute as you come up. But Kendra has joined us from Ryder about 7 months ago. where she served as the General Manager of a $200 million book of transportation business. She subsequently served as the Chief Technology Officer for Ryder's Supply Chain and Logistics business. She also led their venture capital work. We're super excited to have Kendra. She is deeply experienced in what it takes to deliver a valuable service to our customers based on these platforms, and she'll talk to you a little bit more about what that entails.

Great. Thank you, Sterling. Good morning, everyone. As Sterling said, I've had an extensive background in logistics and transportation. I also unfortunate enough to have an extensive background and bring emerging technology into logistics and transportation. In my time doing so, I've learned there's 2 key elements to success. First is a compelling value prop for your customer, which as you heard earlier from Chris, Aurora absolutely has. Second, the thing that often gets overlooked is your product has to fit into your customers' networks in the way they operate today. And that's what I get to help Aurora build. With a service delivery team, we're designing, building and implementing the technology and processes that will need to bring the Aurora Driver to our customers to enable them to operate at a commercial launch. So let's talk about what we're doing. First, today, we have pilots with some very impressive names in the industry. We're running autonomously from Dallas to Houston and Fort Worth El Paso every day with these customers. You can see it's FedEx, Werner, Schneider and Uber Freight. The benefit of these pilots is it allows Aurora and it allows our customers to learn how we can operate successfully together in their networks that they have today. We've had a lot of great lessons learned out of these pilots, and I'll share a few. First, our customers' trailers don't always come to our terminals in perfect working condition. This happens in transportation. One of the things we're going to be building is on-site maintenance facilities at our terminals. The value of this is will allow us to provide maintenance and repair our customers' trailers so that we can get them out on the road, complete their load and ideally do so on time. The additional benefit of having maintenance at our facilities is around the Aurora Driver tractor. If we can do preventive maintenance or repairs to the Aurora Driver tractors here, it will allow us to drive a higher uptime, which really enables that utilization target we want to bring to our customers. Second, you can see that Aurora has been very thoughtful in the types of companies that they've partnered with. We have companies of all different modes of transportation, less than truckload, parcel and brokerage. By doing so, we've learned not only how these customers operate but how these modes operate. And that's been great for us. What we've seen is that some of these modes are very heavy in the middle of the night and early morning while others are running more traditional transportation times. What that means for us is it's important that we have a diversity of customers across different modes so that we can work to level load our days and level load our weeks to really push that throughput through our networks. As you can see, we're also really proud of what we've accomplished thus far with our partners. 550 loads delivered, over 150,000 miles driven, and we've done so 100% on time and with 0 cancellations. What you're going to hear from us is we're looking to grow these pilot partnerships in the future. We're looking to really ensure that we get our lessons learned, and we're building a product that works for our customers at launch. So what we're operating today is our terminal to terminal network. And this is a high-level view of how that will work. By the time we're operating commercially, both Aurora and our customers, we'll have complete understanding and confidence in our ability to deliver and to deliver so in a way that works for our partners. During the pilot and early commercialization phases, we're operating this model so we get it solidified. And then we're able to handle it to our customers to operate in their networks while Aurora transitions to Driver as a service. Today, the way it works, our partners bring their trailers from their facility to our terminal. They dropped their trailers at our terminal and Aurora Driver takes over. The Aurora Driver moves autonomously from the origin terminal to the destination terminal. It then drops the customer's trailer at the destination terminal and the customer comes and takes it and brings it to the next location. So let's talk about the services we're building to enable us to get to launch with this product. So we're going to zoom in on different components of this. The first thing that happens or will be happening is that customers will send us a request electronically. We're building out the Aurora Beacon technology. That technology will be integrated with our customer systems to enable all this functionality. We can integrate with our customers' ERPs, their transportation management systems or their order management systems, whatever they may prefer. So the customer will send this letter request electronically. Aurora Beacon will receive that and we'll manage that through Aurora Beacon scheduling application. We'll look to make sure we have available capacity on the lane at that day and time and ideally, we'll accept that load. Then the customer will bring their trailer to our terminal as they do today. They'll drop that trailer. And ideally, they'll pick up another trailer that has completed its autonomous mission and drive that back to their facility. Now that their trailers at our terminal is time for us to take over and get it on the road. We're going to be doing that in a few different ways. First, we'll do traditional things you see in transportation, such as a pre-trip inspection weighing and fueling, all to be done here at our terminals. We'll also do things for autonomy, such as calibrating our sensors and ensuring our operating domain is safe. Once that truck is ready, we will have our terminal operator drive it over to our launching zone, excuse me. They'll enable autonomy and they'll send it on the road. So here's where the fun stuff happens. The Aurora Driver takes over, and it brings that truck from the origin terminal to the destination terminal. The goal of this model is to enable our customer to focus on the local halls while the Aurora Driver focused on the longer-haul routes, driving efficiency in both networks. So now while those trucks are on the road, I mentioned earlier we have a Beacon suite of technology. We will have a command center staffed with personnel who are using this technology to monitor the trucks every step of the way. I mentioned our Aurora Beacon scheduling application that's used initially. But once it's on the road, we'll be monitoring via Aurora Beacon fleet monitoring and also Aurora Assist. Let's talk about what both of these do. Aurora Beacon fleet monitoring allows us to see all of our trucks that are on the road and the status of all those trucks. We'll know if they're on time or running late, and we'll have 4 warning systems to see if they run into -- could potentially run into issues of weather or traffic conditions. This will also be integrated to our customer systems. We'll be sending them lat longs and ETAs, so they know where the truck is and when it will arrive at the terminal. Aurora Assist. The best way to think of this is like an extra set of eyes and ears for the Aurora Driver. Let me give you an example. The Aurora Driver is going down the highway and knows the speed limit 65 miles per hour, but it encounters a new construction site, one we haven't seen before. And there's a temporary speed limit time that says 50 miles per hour. At this point, the Aurora Driver will trigger a session to our operator back in the command center and ask for validation. The operator will look at the scenario and give the appropriate input. The great news is we actually have all of this technology here for you guys to see later. So please stop by and see it's live, you'll see it in action. All right. Our Aurora Driver has done its job, and we're bringing that load into our destination terminal. The Aurora Driver will pull in autonomously and park in the landing zone. At this point, we'll disengage autonomy and the terminal operator will perform the post-trip inspection. Go and drop the trailer in the yard, and then our goal will be to have our standard operating procedures to enable that tractor to be turned as quickly as possible and really ensure we're getting the utilization we want out of our equipment. And at same time as that's happening, Aurora Beacon will be sending information to our customers, letting them know that the truck has arrived and it's ready for them to come pick up. Once the customer picks up the trailer at the terminal and exits the terminal, the load is complete. So as you can see from what I've shared, we're thinking about this business, not just from the view of autonomy, but how we seamlessly integrate into our customers' operations as they exist today. We have a focused team developing this work, and we'll be really excited to share it with you as we continue to mature our product. When we transition to our driver as a service model, we'll have confidence and our customers will have confidence that our products will be able to effectively integrate into their existing networks and drive the value we all want from autonomy. And with that, I'll hand it back to Chris.

Thanks, Kendra. So I'm just going to take just a couple of minutes to revisit what the regulatory landscape looks like. So in the U.S., regulation is divided between federal, state and local governments. The federal government regulates in general, the safety of the vehicle itself. And through FMCSA basically puts in place minimum safety requirements for motor carriers. The state governments regulate the safe operational vehicles on their roads. So driver licensing, for example. And then, of course, local government regulates access to street and permission to operate within those regions. At Aurora, we've known from day 1 that this kind of ecosystem of people are, well, they're part of the ecosystem, right, that we have to engage with them. We have to build trust with when we have to help them understand the promise and opportunities and also the challenges with the technology, so they can be informed in generating policy. The good news is that today, if we had a truck that we were confident in the safety of, we could operate in the vast majority of the U.S. So the blue -- the dark blue states here are states where we could go and take a vehicle operate either because there's explicitly a law that allows it or there's no law preventing the driverless operation of trucks on the road. The light blue states allow us to do testing. And today, we're working, for example, with Pennsylvania, working with the state DOT, working with other partners in the state to try and move that from just a testing regime to a regime that will allow a safe operation of driverless vehicles as well. California, is the outlier here. So California is a place where you're allowed to operate light vehicles without drivers but not yet heavy vehicles. That's a state where we're again, we have obviously one of our Headquarters in California. We have a large amount of our team there. The technology has been developing there. We've been building relationships with California DMV and DOT, and we're starting to see them move to implement the second part of regulations. The history here is that while the law was passed, I think, 2015 to enabled driverless vehicles, the DMV bifurcated between heavy vehicles and light vehicles. They passed all the regulations they need for light vehicles but not heavy. And so we see that progressing again now, which is great news because California is obviously an important state for freight in the U.S. We'll continue to engage with these folks. We don't need regulations to change. We do anticipate that there will be regulatory action over time, but by being actively engaged with them by being trusted voice by being transparent in the way we have we think we can help ensure that any regulation that does happen over kind of the next decade timeframe, we'll end up in a place that is supportive of safety on the roads and enabling of this product in the market. And so with that, just to sum up, we're in a really exciting moment for Aurora. We have an incredible group of leaders that you get a chance to spend time with today. We have this technology that between the way we're using machine learning on the vehicle, the way that we're using amazing sensor technology like FirstLight that allows us to see the road the way we're thinking about how we harden this so that we can actually be a viable product. We think that's an incredibly powerful technology footprint that we have. We have these incredible partners, right? We are spring-loaded, that as we deliver on our part of the technology, they will be able to come to the market with 2 of the top 3 truck manufacturers in the U.S., an incredible set of customers who are ready to embrace this and see the benefits for their business rollout. We have a launch plan that makes sense. And of course, we are leading with trucking, but then we'll come back and pick up the light vehicle work over time. And so I think with that, I'll invite Sterling and Richard to come join me up here, and we'll turn to Q&A.

[indiscernible] development if you talk a bit more about availability rate and then how long it is going to take today [indiscernible]. How long does that take?

Yes. So maybe Sandor, I ask you to come up and answer. Sandor leads our vehicle integration hardware programs. And do we have a microphone?

Thanks. That was a multipart question. I'm not sure I can answer the part on the chips first. So we're actively sourcing chips both fortunately -- unfortunately, our scale is quite small, so we are able to secure the supply we need. That being said, it's a constant thing that our supply chain team is working very closely with key providers to make sure that we have what we need. And we haven't been impacted to date.

Okay. And then there's a question around upfit. I think was the second part of that.

I'm sorry, could you repeat the question?

How long that it takes upfit truck [indiscernible]

So the prebuild pieces are coming later. So right now, we're using standard vehicles, and we're using a standard Peterbilt truck today. The upfit takes about 3 weeks per vehicle. But we are able to do, as you saw in the video, multiple vehicles at the same time. So we banked out the almost 2 dozen in about 2 months once we've worked through the design.

Notably, that is for like current purposes of development. Our expectation and what we're working towards with these OEM partners is to pop them off the line. That's the whole idea of this series production, scalable truck is to be able to produce them at much higher rates.

I think we'll start over here.

Chris, 2 questions for you. When you talk to your partner customers and other prospective customers and you show them the -- your vision of how the network works, i.e., the long haul is done autonomously and then they do the last effectively intermodal for the road. What is their response to that? Do they accept that like, hey, I need to completely approve my supply chain kind of, how are they reacting to that? And the second question, completely unrelated is, on the hardware side, how close are you to commercial-grade hardware production. Obviously, you're doing FirstLight yourself, but for other hardware products, kind of is that a bottleneck that kind of could prevent your timeline to commercial launch?

Maybe Sterling, do you want to take the second question, and we'll get Dave to come up and talk about the partner engagement and the intermodal.

All right. So let me handle the first one. I think it's a great question. And it's important to recognize this is why we have pilot partners and launch partners. We've selected partners who have a like vision with us, right? So if you think about delivering the benefits of self-driving technology, it's a long runway of how we're going to roll this out. And so we've selected companies that understand that starting off in a terminal to terminal operations may not be the ideal solution for every use case. It still creates enormous value overall, and our partners are willing to work with us to actually build out that deployment model until the technology matures to a point where we can go, as Chris showed you before, where we eventually get from warehouse to warehouse and depot to stores. So that's part of the reason why you select like-minded partners who have a vision of rolling out the autonomy in the future. So every one of them is a little bit different in terms of the deployment model and how to benefit it together. And so we work with each of our partners to develop that strategy. So that's how they think about it. Ultimately, end-to-end will be the ideal at the end of the day, but there's a great value for terminals for a lot of reasons, flow-through volumes and things like that. So there's enormous opportunity to redo the network to a more efficient utilization overall. So I hope that answers that.

Yes. If I can add just a couple of things to what Dave said. One of those is the terminal-terminal model is not only an opportune introduction phase for a self-driving system. It's also for our partners. So a couple of things. First, many of our partners are wrestling with this idea of, okay, how do I intermix autonomous systems with our drivers, right? We were just talking with Eric about this last night. Effectively, what this allows them to do is start to transition their human drivers to more regional halls that they prefer anyway. And the other thing that allows us to do is convert these highly valuable assets that can run nearly 24 hours a day into very high utilization drop and hook operations, that you simply can't do if you're going to endpoints and doing live loading and waiting around at warehouses, you just can't do that. So it allows much higher utilization of the assets. So it is a little different, but there are also benefits to that phase in for each of these customers. And what also note is one of our pilot customers right now. We were talking about this last night, one of our pilot customers is actually like this isn't actually unique for them on one end of the load, right? So we do one of these routes. And on one end, they were already using local dray. And so it is unique on the other end, but they're fairly accustomed to this kind of a model in some case.

[indiscernible] question about scalability of...

[indiscernible] The thoughts on maybe the hardware [indiscernible] kind of color is that right now inside of [indiscernible]

Not one. So I think what should have been apparent that I guess wasn't is, as we've gone through the preparation of driverless capability, we've simultaneously been doing a lot of the ruggedization and hardening already. And so while we called that out as a separate phase of the roadmap, the computer that's in our trucks today has already been developed relative to the ISO standard 16750 as well as -- and as we roll in kind of the FirstLight LiDAR and some other things those have also been developed with respect to that. So I expect that is not going to be a bottleneck for us.

And then over time, we will continue to iterate these generations of hardware, reducing the cost, improving manufacturability and doing that kind of appropriate with the scale of the fleet that we're operating.

All right. So [indiscernible] possibly work out to in the year is about 30% of all the [indiscernible] every day working really hard. So my question is that following up on this with 2 aspects. So first is in 2024, 2025, 2026 as a percentage of hours in the year, what do you think your utilization rate will be per truck? And the second is with respect to how much you increased utilization rate, how do you think about risk there and the more you realize these trucks more potential there will be wear and tear for driving this [indiscernible]?

Do you want to speak.

I can speak at a high level as it relates to kind of what's been modeled or whatever in the forecast. That's a Richard question. At a high level, our expectation for the kind of utilization of time on these trucks is that it improves over time. We are holding a bar internally for their launch as it relates to the reliability. I don't think we've published publicly. But our expectation is there is this upper bound -- most -- much of the 70% downtime that you just referenced is specific to human drivers and their limitations. They're the unique limitations of a self-driving system are far fewer in nature. Kendra mentioned a few of the steps that are taken at the launch and landing of each of our trucks to include these pre-trip inspections. We expect and we've already seen in some of our -- with some of our pilot partners, that those are catching issues that contribute to that downtime in current operations that we wouldn't see, right? So it turns -- when you actually check a trailer as per FMCSA direction, you're actually catching these things that would otherwise leave you stranded on the side of the road. So we expect incrementally fewer of those on account of that more rigorous inspection process that we're going through with these trucks and trailers. And so the downtime attributed not to the human causes, but to the kind of base platform and trailer causes, we expect to decline. Now as it relates to the final piece, which is, okay, well, certainly, how much incremental downtime do you expect to see on account of the Aurora Driver? Again, not a number that we have kind of highly refined right now or certainly not published, but it is something that we expect that many of these improvements that we've made to the hardening of that hardware will lead us to not particularly dissimilar downtime to what any other mechanical system would encounter on that truck in the long term.

And then I think a lot of time when you think about risk, you normalize it based on operating hours or on a per hour basis. And so we -- our expectation is that we're not creating unreasonable risk on the road. So front row?

Thank you. Chris, I thought it was a great point you made, which is we're going to start terminal, to terminal and move beyond. I think there's a perception that when you launch commercially, that's it, right? That's the capability of the system. Can you talk a little bit about what Gen 1 of launch technology won't be able to do? And maybe as you build out your strategic roadmap over the years that follow, what Gen 2 will be able to do, what Gen 3 will be able to do just help us understand that?

Yes. So maybe it's worth bifurcating between hardware and software. So on the hardware side, there we've engineered the hardware stack to be able to operate across the range of capabilities. So all the way from the freeway driving we're doing to urban driving, and we've kind of built that experience early on, and we've helped set the requirements there. And so the hardware will really be moving down a cost reduction curve where there will be discrete steps as we kind of change through design or componentry in that system. And of course, we'll pick up volume benefits along the way. On the software side, we expect this to be more continuous that we'll be adding incremental features. So at launch, we'll be driving from our terminals, we'll be driving on the roads on the surface road merging on to the freeway, driving down the freeway, getting off the freeway and then coming to a terminal. We expect that we'll have to deal with construction. We'll have to deal with emergency vehicles on the road. We'll have to deal with other drivers behaving in ways that are semi-erratic. We'll do that through a combination of the onboard Aurora Driver and then import from Aurora Beacon. Now it's important to know Aurora Beacon is not somebody behind a steering wheel driving the vehicle down the road. They're providing input, think of it like a dispatcher, right, where, as Kendra mentioned, we might have the thing on the road notice the reduced speed limit sign and then say, "Hey, did -- are you sure that was a reduced speed limit". Let me just make sure so that I'm not creating something a moving chicane on the freeway. So that will be kind of the launch package. And from there, we would expect it to expand to more surface type engagements. So as we would think of perhaps in the initial places, we don't have to -- I don't know what to think we wouldn't have to do. We don't have to do a U-turn on the road back into a dock. Maybe initially, we're turning across 2 lanes of traffic but not 4. And so you could imagine us expanding those. And those will be driven -- those features will be driven by the economic expansion model that we have. So as we move on to new lanes, we'll look at the delta between capabilities, and add just those to unlock the next lane. But again, we expect one of the benefits of operating in the freight domain is we expect much more similarity and much more scalability from the core technology. Maybe back of the aisle.

Three questions on the commercialization of everything. First of all, has anything changed in terms of the drivers of service unit economics?

So we're still talking to customers. Obviously, pricing isn't finalized yet, but I think our expectations are that what we said in the model are still supportive. So if you think about something like the pricing of the product, for example, we still think that this technology is going to come into the market slowly. And it's going to be a small share. So our rail competition is going to be human drivers. And we don't expect the human driver rates are going to go down significantly over time. In fact, they might increase. So we feel like our initial unit economic thoughts are still supported. You'll also see some of the competitors who started off through their kind of go in public process, suggesting that they would be doing discounts against human drivers. And you start to see over time then bring in their expectations back closely to where we always were. So yes, so we're pretty confident that the unit economics are around the same that we thought of. Again, hardware is a cost factor. We still expect that to kind of come down over time as we scale. There's no unobtainium, as Chris likes to say, in this. So as we increase the scale and we'll just go down to normal hardware cost life cycle. So we expect the unit economics to be about the same as what we've said previously.

Okay. And then the second and third one is given the advantages you have around LiDAR and some of the other AI and ML, why not commercialize that either through licensing, looking at other tangible markets outside of trucking now as opposed to waiting? And just while I have the mic, the third question would be, why not just focus on one route and fill it up as much as you can and maybe hit the S curve a little quicker than you currently think?

Yes. So on the commercialization in the long term, we would love to drive everything in anything and look at adjacent fields, but we think there's a real value in getting a product on the road getting it out the door and building a business. And we just -- we are convinced that freight is that right incredible opportunity. Do you want to take the...

Yes, as it relates to kind of growth of volume, our model, Chris alluded earlier to the model that we've used effectively for that kind of lane expansion. We're working on this closely with each of our customer -- pilot customers that we're working with. But the key consideration here is if you look at effectively the return on any kind of given investment tends to scale with the number and complexity of endpoints. If you're having to deal with more novel things, your investment in terms of kind of incremental development rises. And you need to believe that you can pick up the kind of volume to warrant that. So as we look at major freight lines, we look at the kind of volume that we could realize with these partners at each of these locations. And we say, okay, what is that kind of like ROI sequence curve look like in terms of kind of the return, in terms of number of loads, relative to that investment. And you're right, there are areas, for instance, like a Dallas Houston Lane, where what is it, 8,500 trucks a day. I don't know, I don't even remember if that's round trip for one way, but it's significant. It's a heavy lane, right? There are other lanes like Fort Worth to El Paso if you get on to Phoenix and get on to L.A., it's one of the biggest rate quarters in the country. So there's no -- you saw the kind of the notional lane expansion rollout, it's not an accident that these heavy lanes are prioritized. And once we turn them on, the opportunity to scale up volume does come at a significant discount relative to incremental development of new endpoints.

And I think, might have been going to say the same thing, but I think it's important to remember like that we want to do both, right? Like we want to have a really big business. There's a massive opportunity. So we want to scale breadth wise, but also depth wise on the lanes that we can do. And that the mix of that, I think will vary dependent on what land it is and what the partners want and what the customers want, but you want to take full advantage of the opportunity that's there. So go breadth wise and not just kind of get stuck going depth-wise straightaway.

Yes. And I guess just maybe to say that another way, like I very much look forward to the day when every truck here in I-45 is driven by Aurora. But as we build the business, we're going to look for what's kind of the easiest way to expand and grow volume. And is that kind of exploding the lane we have. Or at that point, is customer acquisition challenging enough that it's time to push to another lane and go out. And so that's part of what the business team is going to be constantly evaluating its like, is it realistic for us to go from whatever percent to whatever percent on this lane? Or do we now need to be expanding somewhere else. Maybe in the back.

Ken Hoexter from BofA. Just the last time we were here, there was a lot of discussion about mapping and digital maps and Aurora's vision and the like. And it was kind of different than some of the other autonomous companies. And it was -- if I remember, Sterling in your presentation, you was taking about constant learning. Maybe you can just refresh us. How is the structure of the mapping and the ability to get new lanes from Aurora? Because I think it was different than most of the other companies, which I think was...

Okay. Do you want to take it?

Sure, I'll start and then you can jump in to or Karl, if you want. So on the mapping front, you can develop a more robust system more rapidly. If you bake fixed features of the environment into the map. This is everything from what are the traversible pathways through a particular intersection allowed by this light versus that one? To do otherwise is to require that the system is constantly attempting to infer at every point in time, not just kind of what should I do, but what's around me. What can I do? What is the law allow, and so it allows for a much more robust rapid introduction of the technology. Now that map bakes in notably fixed features in the environment, not the dynamic ones. And what it allows you to do is focus when operating on stuff that's moving on stuff that you care about and you have to kind of actively negotiate with. And what you guys will see on your [indiscernible] today, as you get on the highway and it is an active negotiation when you're merging, particularly hopefully, some of you see heavy traffic and you see how kind of the Aurora Driver navigates that. But it allows you to focus on those things. So it allows for a much more performance system. Now the world changes. Construction changes the world. Some of these other things change the world. And so one of the key elements of our architecture that we probably talked about a year ago, I don't remember, we did at this point, but it is the fragmentation of our mass into a set of, we call them [ shards ] or effectively locally consistent tiles that we can rapidly update and turn around in hours to ultimately near real time. As opposed to a globally consistent one that's historically been more of the purview of I think some of these -- many of these companies that use maps that requires them a large-scale optimization that takes on the order of weeks to update. In addition to that, one of the things you probably saw on our roadmap, I don't remember if we talked about it, we didn't talk about kind of detect lane lands, but there's also this kind of real-time element of, hey, even before we send back the data and update the map and every vehicle in the fleet learns kind of what the new structure of the world looks like. If anything changes real time, you need to be able to detect and respond to that. And so the team -- what the team has developed over the last quarter and is continuing to mature is the ability to online real-time detect where the lane lines are even if they've shifted and allow the truck to actually deal with things like construction barriers and other things that would cause us to diverge from what the map might otherwise have suggested. But at the end of the day, this ability to both have an understanding of what the world should look like, the ability to deviate from that as necessary, but the ability, in particular, for the fleet to learn from the collective experience that every vehicle has really comes down to maps, right? That's the secret of -- people talk about fleet learning, the primary kind of core central element of fleet learning is maps. It's every vehicle knowing what happens to the other vehicles at a particular location in space.

Yes. And it's really -- like I maybe take it up a level, like think about what computers are really good at, they're good at storing and accessing information and so let's use that. And there's kind of -- you can think of 3 schools of thought. One is no map or absolutely minimal map. And if you're just driving on a freeway, maybe you can get away with that because there's an awful lot of structure there. Then there's -- but we think there's advantages to doing more than that. When -- then there's kind of this monolithic map right? So one big map, think of it like a giant world atlas. The challenges with that is it has to be globally consistent for it to make sense. And it turns out that whatever you're using to measure the world is going to have error in it, and so that error gets baked in somewhere. And so when you go to update the map, it becomes very complicated to reoptimize to make that error fit. And then finally, there's the approach that we're doing, and we think -- we don't know that anyone else is doing this, which is much more like a road atlas, a set of pages or [ shards ] as Sterling said, where each local area is consistent, and that area is about a couple of hundred meters across and then it's linked to the next one. And so when something changes in the world, we just take out one of those pages, put back in the next one and again, allows us to maintain it more quickly. And it gives us the kind of benefit that you have when you drive near your house versus when you drive kind of somewhere else randomly in the country.

And can I just -- I think maybe another part of the question you were potentially asking is we don't actually believe that it's particularly resource intensive in terms of either time or like vehicles in order to kind of expand that map over into another domain.

So it's effectively self-maintaining as well. So it's built by the very vehicles and sensors that are used just for autonomous operation. So when any vehicle detects a change to the world, that change will trigger a map update if required. So it's not a -- it sounds like there's a giant dedicated mapping fleet. It's the very vehicles that are doing commercial operation.

Brian from JPMorgan. So I just wanted to ask you, you're talking about percent of miles in autonomy. Why did you go that route? What did you consider? And we all talk about engagements, but you have a different policy than others. So why did you go that route? What did you not consider and why? And then also, when do you think you're going to be able to talk more about cost per mile because you mentioned the competition is human. Obviously, it's quite heavy right now. You've got a couple of milestones laid out. But when do you feel like you'd be able to talk about that and then the steps to get to something that's competitive and reliable?

Yes. Yes. So when we thought about that measure, we were trying to find something that was objectively measuring what does the product have -- how do you know whether we're performing in kind of the direction of the product, right? The Aurora readiness measure is we're good to go. That is the thing that is the bar. When that gets to 100%, we're confident in the safety of the vehicle, we can be out on the road. But we understand that, that is a little more abstract, right? And we want something to be able to share that's indicative of performance on the road. We looked at things like [indiscernible] disengagement or intervention. And when we think about that, a, it's really subjective and it can create this perverse incentive that I'm concerned about for our operators, right, that when they are driving, if they have to think, is this going to change our performance this quarter versus keep some unsafe we just don't want that in the measure. And it's -- we expect that there will always be some form of remote support that's enabled and obviously, some port of on-site support that's enabled. Over the long term, when we think about cost, right, we expect that there will be cost of the vehicle driving without any support then there's going to be some amount of cost that's associated with remote support. And then there's a more meaningful cost associated with on-site support. And so in the long term, we expect thinking that, that will be something that will be important drivers of our business that we would care about. And so this feels like it bleeds into that over time. For us today, as we think about the phases of our business, the -- as we talk internally about this, the first risk is getting to the point where the Aurora Driver operates without a driver, right? That is fundamental to the business. And so that is what the whole team is focused on. Over time, you're right, the economics are going to matter. But I think that most people in this room would believe that if we can get to the point where the thing is actually operating autonomously in a commercially relevant way that we're going to be able to bring the cost down on that. And so we'll share that as it makes sense. And if that becomes the prior for the company. But we're already engineering the future generations of hardware that see those discrete costs. And part of the benefit of operating in freight is that the unit economics are so much stronger than the ride-hailing, and so that opportunity to start to build volume and have positive unit economics is much closer. At the back. At the very back there, sorry.

Yes. I was going to add something to a question about unit economics that was asked a minute ago. But just as a matter of context. I'm sorry, as a matter of context, Matt McLelland, I'm the VP of Sustainability and Innovation at Covenant Logistics. And so we're an operator. We're a customer, one of the partners that are working with in the program. And in my role, I spend all day looking at this in electrification and hydrogen. But right now, Aurora is an area where we're putting a lot of time. With the question about unit economics, one thing I wanted to add, in fact, almost jumped in and was that we keep -- with 2 things. One is -- one of the biggest problems we have with our drivers is they don't tell us when something's wrong. Their goal is to get as many miles as they can. They want to get paid. They want to make sure that their lives aren't disrupted. So if they don't tell us that an error came up, if they don't tell us there's a clunk in the left wheel. We don't know and end up the repair for that truck and so being much more expensive. And so with the Aurora Driver, it will tell us small codes, big codes, things that we need to start doing to route the truck back to where it can be repaired. So I just would add that because we feel like the unit economics is going to be improved greatly because the driver is taken out of it. Again, not to marginalize our drivers, but it's just Eric here is an operator, he can tell you, drivers don't always tell us what we need to know. The second thing I wanted to add is because drivers -- we keep our trucks about 400,000 miles to about 2.6 years is when we flip it. And we do that for 2 reasons. The biggest one is because we need to keep drivers in new trucks, clean trucks. And I know that including our friends up here in the front, you haven't seen some of the trucks that come in our terminals after 2 years, these things are disgusting. And you just can't clean them good enough. And so part of the way to keep drivers interested, part of the way is to minimize driver turnover is to give them a new truck. So at 400,000 miles, we tend to turn it. But we don't have a driver, it does not get dirty, and we can keep it longer. So we can keep it on the road longer. So now maybe something we keep for 2.4 years or 400,000 miles, maybe we take that to 5 or 6 or 7. I mean these things were built to run. It's not uncommon to see trucks with millions of miles on the road. So the unit economics for these 2 reasons is one of the things that is part of the value proposition that we're trying to bake into our model.

Thank you. Maybe at the back over here. And then...

David Vernon from Bernstein. Thanks guys for having us down here and thanks to the broader team for coming out. I had a question for you around the 2023, I guess, definition of commercial launch. Could you help put some meat on the bones around the geographies that are involved in that, the number of trucks that will be running maybe a broad range of revenue that might be coming on an annualized basis as you get there? And then as you think about taking it from the narrower lanes that you seem to be focusing on now and then branching that out and reaching some sort of S-curve point, what are the incremental investments after you get to that point of commercialization that you're going to need to do to really kind of accelerate the reach of the technology? Is it investments in new terminals facilities, is it dealing with the regulators and trying to get through the interstate issues? Can you just kind of talk broadly around kind of what you're seeing as the key issues that will affect your ability to scale beyond the initial sort of lanes of launch focus?

There is a lot baked into that. So we'll see what we can do and keep us honest if we don't get all of it. So I think when we intend to launch in '23, what we've shared previously is we're thinking roughly over [indiscernible] vehicles and it will be in Texas right to launch, sorry, commercial launch.

So '24. Aurora Driver Ready in '23.

Sorry, I thought -- I think you were -- I misunderstood the question, sorry.

Are you talking about Aurora Driver Ready? Because you said 2023, that's the Aurora Driver Ready milestone.

End of '24.

Yes. End of '24.

Yes. Yes. Yes. Okay. So I'm answering the right question now?

Yes, '24.

So when we think to the commercial driver ready -- sorry, the commercial launch there we are thinking the scale of 20 vehicles, it will be here in Texas. We're focusing right now on Dallas, Houston and Dallas, El paso. We'll see exactly which of those make sense as we get to launch time, but that's where we're thinking. And I've now lost the rest of the question.

So I think we're not giving revenue guidance. So we're not going to give any suggestions about what revenue might look like on a kind of debt basis or an annualized basis at this point? And then I think the next question is around the incremental investments across the spectrum in terms of being able to scale after that.

Yes. And so we'll continue to -- one of the things that will enable scale is the engagement we have with our OEM partners, they're today building vehicle platforms that we expect to be scalable. The next will be our hardware. And so our initial hardware platform is from the Aurora Driver side is something that we expect to have a relatively small scale but then we have generations that we're already architecting and developing that will bring that scale further up. And we look forward to sharing more with that -- about that at some future time. And then on the software side, again, it will be very much -- okay, what is the small delta between this lane and that lane, let's validate that case. So if we would expect to be able to turn -- if we're turning across 2 lanes of traffic, we'd expect turning across 3 lanes of traffic to work let's put the validation work in place and then have confidence, okay, yes, that's ready to go. There will be some investments in terminals early on is our expectation, and that's baked into our financial models. But over time, we will move to depot to depot or partner terminal to partner terminal as the economics make sense for that.

Yes. And there's a period of time as well, whereby we might -- like this terminal, this kind of ours, right? We are going to operate it even as we expand out to terminals, we would fully expect to find somebody who's good at operating terminals and thus be able to kind of pay them similar to kind of like how airlines pay airports. So it's not kind of coming on to our balance sheet.

Okay. Looks like your...

Tom White, Davidson. Just a follow-up on the trucking platform. I think you expressed that you -- we're encouraged that some of the component kind of supplier contracts have been kind of locked in. And maybe this is a question for your partners here in the audience. But how are you feeling generally confidence-wise, about your ability to or their ability to kind of get that done? How could you guys kind of keep that on track? Would you ever consider, I don't know, helping them kind of co-investing to kind of get things across the line sooner?

Yes. So I feel incredibly excited about where we are with our partners. These are amazing companies. They understand how to deliver product. They understand how to deliver product on time as Sterling mentioned, kind of -- as we looked at the program, we saw -- and I think they would share this, if you talk with them, right? Some of the biggest risk in those programs is supplier selection, requirement development, all of the business and technical wrestling that happens prior to unlocking that. Once those commitments are made, suppliers, again, are very good at doing the thing they said they would on time, on schedule. So we feel privileged to work with these amazing companies and really excited by the prospects of that.

I'll add to your last -- the last part of that question, do you side jumping in just working together on some of it? Yes, -- in fact, with both progress and Matt Collins here, Chief Engineer for the PACCAR programs. You can ask him some of the questions. But in specific areas where we have expertise or preexisting hardware that can be brought to bear to satisfy a need. Our engineers are right there to help provide it and work through it. And so our engineering teams are very tightly coordinated in the development of these systems.

Yes, this is very much not a kind of fire and forget. We're working together on a daily basis. So one more. Okay. We answer that more quickly. In case Stacy will keep us on schedule.

Great. Maybe this is a good place to wrap up. Chris, you said earlier in the Q&A that your vision is at some point, every truck on the highway is an Aurora truck. I think you're being fictitious.

Not totally. I hope for that. That would be great.

There you go. So maybe if I can just push you on that, right? So what is your vision of what the trucking industry looks like 10, 15, 20 years from now? Are you the dominant player? Or do you think there are 2 or 3 other credible guys in the markets divided to 3, 4 ways? Is this a private truck fleet industry? Is it for higher industry? Does trucking take share from rail kind of what does the world look like 15 years from now?

Yes. So we are obviously working to be the dominant provider of automated vehicle technology in the marketplace, and we're working hard for that. I do not expect we will be the sole provider, right, that this is a $700 billion space. I think our customers would want options. I think the government would push for options, right? So I think that there will be multiple players in the space. But it's not going to be dozens. It's going to be a small number. And as we look at the landscape, we think we're incredibly well positioned to be one of those and if not be the dominant of them. Yes, I think there's an incredible opportunity for change in this space. that I think about that the transformation of the logistics networks where today, the number of warehouses that are needed so that you can service the U.S. within 24 hours, it's large. As you're able to push the utilization of trucks to closer to 24 hours, you don't have that hours of service limitation, that drops down to 2 or 3 warehouses. So that fundamentally changes things. we will be able to, over time, displace air, for example, again, because we use air for rapid shipments of high value things. We'll be able to compete with that, with trucking. So I think that will be interesting. I think again rail, I think also, ultimately, we start to steal share from as well. So I do think that trucking will grow. And that's just in the U.S., right? We think that as I think about Korea, I think about Japan, these are economies that need this technology. They don't have the workforce to support it. ultimately, Europe is going to be a very interesting market as well for us. So yes, there's a lot to come in. And it's one of these things where you think about over the arc of history when there's been fundamental changes, whether it's in communication or transportation, right, go from walking to the horse to the wagon to the internal combustion engine. This is kind of the next big step in ground transportation. Thank you. And so that will be the end for our Q&A, but certainly not at the end of the morning. Thank you, everyone, for being here, and thank you for everyone who joined on the webcast. Just before we break, I do want to introduce a few of the people we have around the room here and maybe they can identify themselves. These are some of the amazing folks at Aurora. So you met Sandor already, leads our hardware team. Alan over here leads our corporate strategy work. Yanbing at the back, leads our software team. Carl further at the back, leads the autonomy team here. I saw Veer somewhere. Veer is our product leader for the Driver. Gerardo leads our regulatory engagement government affairs. You've met Kendra, you've met Nat, you've met Dr. Runge. Dave Maday leads business development. I think I got -- is there anyone that I missed? And [ Lia ] leads Operations. I'm sorry, too clear in the front here. So please, they're around here. I think they're great people. We'd love for you to meet them, ask some questions, so they can give you much more depth on things where appropriate than I can. So thank you all, and please enjoy the day. Thank you.