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

Okay. Great. Thanks, and welcome back. Another really exciting presentation along more of our technologically sort of focused coverage here and companies that we've been talking to. We are very pleased to be joined by Aurora Innovation. And joining us from Aurora is Chris Urmson, who is the CEO and Co-founder of the company. So Chris, I think I'm going to turn it over to you. Thanks, first of all, very much for coming and joining us here at the conference. We're really excited to have you and hear more about the story. And I think I'm going to hand it over to you to run through a few slides, then we'll leave plenty of time for Q&A. And just as a reminder, for folks on the webcast and those in the room, there is the opportunity to ask questions. The instructions are on the tables around the room and certainly you can submit them from the webcast. I can go ahead and get those questions up to Chris. But with that said, Chris, let me pass it over to you. Thanks so much.

Great. Well, thanks very much for having me here, and thanks for showing an interest with what we're doing in Aurora. So today, we're going to show a little bit of the company, do some Q&A. We have our kind of customary safe harbor statement upfront here, so please be aware. Now our lawyers are hopefully happy. So we're going to -- Aurora was founded about 5 years ago with the mission to deliver the benefit to self-driving technology safely, quickly and broadly. And I look forward to sharing with you a little bit about the benefits we see, a little bit about our technology and then obviously answer questions. When we think about the benefits of self-driving technology, they're pretty profound. There is the obvious benefits to safety. So in the U.S., we lose about 40,000 people on our roads every year. Over the last 2 years, that's ticked up substantially. Globally, 1.25 million people die every year. So this is a really meaningful problem. Benefits in expanding access to folks, improving our logistics, which is particularly pointing right now as we're dealing with the global supply chain crisis. I don't want to say that all of it could be solved by self-driving vehicles, but we can certainly make it more robust to the impacts we've certainly felt over the last 1.5 years. And then, of course, improving the quality of life for those of us who have the privilege of driving today. Even if you like driving, even if you enjoy your vehicle with your commuting, that's just not that much fun. And so maybe we give you back that time in your day to be more productive or more restful or get home not hating life, which seems like a net social good. The economic opportunity, as many of you understand, is just gigantic. Trucking alone in the U.S. is a $700 billion market. And we think there's a profound impact to make that more efficient, having huge impact there. And then globally, the scale of transportation is phenomenal, as you all know. What are we building at Aurora? Well, the Aurora Driver, and it's a common core of software, hardware and data services that enable vehicles to drive safely through the world. And we've been thoughtful about engineering that system to work across different vehicle platforms from big tractor trailers all the way through to a light passenger vehicles and sedans. And we use a combination of different sensor technologies. So LIDAR, radar, camera, and we're particularly proud of our FirstLight Lidar, which is the innovation we developed in-house that allows us to see further and instantaneously measure speed when compared to other sensors. And that's a critical enabler for you all to drive at high speed, particularly if you're pulling 70,000 pounds of goods in your trailer. Let's just give you an idea of what the data looks like from these sensors. So on the left, you can see our LIDAR. We take that data and we render it in 2 different views. One is the euclidean tensor view, which is kind of this top-down view of the world. And then the other is kind of a radial view of kind of what a camera might look like. And by taking both views of this data, we're able to get the maximum kind of information from it. We take the radar data and similarly cast it into the euclidean tensor view. Camera data, it looks a lot like the range view from the radar. And then, of course, we take our high-definition maps, and we really think of that as just another sensor input that can leverage -- it's what computers do well, right? It store information, allows us to access it. And then we take this and we combine it through our sensor and tensor algorithm, which extracts information from these different layers, the 20 different sensors we have, passes them through our convolution engine and ultimately come out with a model of the world around the vehicle that will allow us to drive safely. And then this allows us to deal with all kinds of different situations on the road, whether it's merging onto the freeway or dealing with 4-way stops or weather or what have you, this robust perception system enables that capability. Obviously, just having technology isn't enough. We actually have to bring this commercial product to market. And for us, it's really been about building world-class partnerships. So today, we're partnered with 2 of the top 3 trust manufacturers in North America: PACCAR and Volvo Trucks. They make up about 48% of the U.S. market. We're partnered with the #1 car manufacturer in the world, Toyota. We're partnered with the #1 ride hailing platform in the world with Uber. We're partnered with FedEx, the largest carrier in the U.S. by tractor trailer cab. Uber Freight and US Xpress, 2 forward leading technology companies in the space as well. And we see this by bringing the Aurora Driver and then partnering with these other amazing companies, we see the opportunity to have the biggest impact most efficiently in the marketplace. And so that's what we're about as a company and really excited to take questions and get into it. So thank you.

Great. Yes. No, absolutely. Well, thank you for that overview. That was very helpful. And so let's dig in a little bit. I guess one of the things that I wanted to maybe start with is your technological approach to sensing. So there's multiple companies that are sort of pursuing similar path to the market, but with different strategies around sensing. So obviously, FirstLight Lidar that is something that you've brought in-house. Obviously, that seems to be important to you. So can you talk a little bit about how you differentiate your approach to sensing technology?

Yes, absolutely. So at a high level, there's kind of 2 -- kind of 1.5 schools of thought, one which is use a variety of different sensors so that you have complementary failure modes and you can see the world as robustly as possible. And the other is just use cameras. And maybe by the way, I said that, it's pretty clear which side we come down on. We think if you're building a technology that's going to be out there on the road. You want it to be safe, you want it to be robust, you want to get the benefits from radar, camera, LIDAR, right? The countervailing argument for camera only is effectively people drive with 2 eyes. Therefore, that should be good enough. I think like when we build cars, we didn't build them with 4 legs. We built them with wheels because that's just dramatically a better way to engineer it when you do it from the start. Getting to our specific approach to that multimodal sensor model, we use the FirstLight Lidar, which is our in-house technology. And what's really powerful about it is it's got the special way to measure distance. It's called frequency-modulated continuous wave. Most of the LIDARs that are out in the market today, if you follow the space, the other companies out there, they're primarily pulsed LIDAR. And what that means is they send a super bright burst of light out in the world, and they wait for a super bright thing to come back and they say, "okay, I knew how long that took, light travels at fixed speed, and so I've got a distance." The challenge with that is if there's a really bright thing in the world like these lights or other LIDAR sensors, then it's really hard to tell whether that was your light or something else. And so you get noise and inherit from that. And also, you can only spend -- we only send so much bright light out in the world because otherwise, you'll start to burn people's redness, and that's just not desirable. Whereas with frequency-modulated continuous wave LIDAR, we're actually sending a continuous pulse out in the world, and we take the outbound energy and the inbound wave form and interfere them and look for the shift to them. And it turns out that through physics, this is 10 to 20x more efficient than pulse LIDAR is. And so that means for the same amount of energy, you can see further substantially further. It also has this cool property because you're looking for exactly the right pattern, the right frequency, you get to ignore all the other things. So instead of being -- and for those of you who have a physics or electrical engineering background, it's the difference between being AC-coupled and DC-coupled, right? You get to ignore all the stuff that's just kind of right there and look for the stuff that's exactly what you're looking for. And then finally, because we're looking at the wave form, we can actually look at the change in frequency and shift of that, as the light moves through the world. In the same way that when you hear a siren from a police vehicle or an ambulance go by changes in pitch, that's called the Doppler effect. Well, we can measure that with light. And so for each point we get back, we don't just get where it is, but we get how fast it's moving. And so that means we can react more quickly to things in the world, and we can differentiate a pole in the road from a person jogging down the side of the road from a car coming out at us at 60 miles an hour. And you can imagine how valuable that is when you get that half a second quicker ability to react and respond to things in the dynamic world.

Got it. That makes -- certainly makes sense. And then how you -- can you talk a little bit about how you marry all of the sensing technologies in this? So obviously, you're using multiple and I think layering them on top of each other. At your Investor Day, you spent some time going through that and then ultimately, the object recognition process and that. Can you talk a little bit more about how you've developed that process?

Yes, absolutely. So maybe -- I don't know if this will work to go back a couple, but this is kind of the magic of some of the deep learning work that we do, the machine learning work that we do. So we had this view that we call the euclidean tensor view, which is taking the data that comes in and projecting it effectively into something that looks like a map, right, a top-down view of the world. We take that from -- and then in some cases, like the camera data and the radar data, we also generate kind of a radial view, which are these horizontal stripes you see on the left side of it and the images there. And that allows us to get the benefit, both from kind of the true 3D geometry of the world, but also these kind of -- the fact that we have this reasonable angular density from both the camera and the LIDAR view. We can evolve that data together. We run our machine-learned algorithms on that. We can then extract feature maps from that. We then run our neural convolution engine on that, and that generates the output of the system that allows us to know where vehicles are, where pedestrians are, where cyclists are and whatnot in the world around the vehicle. We do this many times per second. We have this novel approach of what we call sensor to tensor, which is generating the objects. And then we have another algorithm that updates those based on individual measurements. And this kind of combination of bottoms-up and top-down measurement allows us to be more reactive and more robust in a variety of scenes, including in situations where there's heavy weather. So I think we posted recently some videos to our website showing the truck driving in a monsoon rain storm in Texas.

Got it. Yes. So that can break through the barrier that we've initially seen meaningful weather issues being a challenge for sensing technology. Do you think where does sort of the more intense weather dynamic kind of fit into that? Obviously, most operations naturally are starting kind of in the Sun Belt. Is there the opportunity to build this out from a sensing perspective, so you'd feel comfortable with snow and ice and those types of operation situations?

No, absolutely. And that's part of the insight we brought to developing the Aurora Driver. It was that understanding of what are the challenges we're going to face, not just in the initial deployments, but as we scale this to have a truly national and global impact. And so the sensor technologies we developed, the algorithmic approaches we're taking, we're engineered to operate across different environmental conditions. We'll validate them first in more benign conditions and then move to harder stuff over time. But one of the benefits of not being just a California company, right, being a company that -- and Pittsburgh and has offices in Bozeman, Montana. Up in Seattle, up in -- and then San Francisco and Mountain View, is that we get weather. And so from basically day 1, we've been operating in the snow and we've been gathering that and using that to inform our approach to the problem.

Got it. That's exciting. We don't hear that much about that. One thing -- so there's multiple companies in this space. One thing that certainly is differentiating yourself from others is your approach, both on the truck side as well as the automobile side. So can you talk about how or why that makes sense to pursue both simultaneously?

Yes. We -- so to be clear, our first product will be in trucking and freight, and we're very excited about that. But when you understand the problem the way we do, you see that it's really -- there's a tremendous amount of commonality between the applications. You think about what's actually hard about building a self-driving vehicle, and it's understanding the world around the vehicle, then figuring out how you should move through it. It doesn't really matter whether you're in a Toyota Sienna or in a Peterbilt 579, right, understanding what those other vehicles are going to do, predicting how they're going to evolve over the next few seconds is at the core of the driving problem. There's stuff around the edges that are much more like device driver type things, if you're familiar with operating system. So being able to talk to the truck, being able to talk to the Sienna. There's certainly differences between trucks and Siennas, right? This truck has -- well, it's much heavier. The gear shifting is much more discrete. It ends in the middle. It's more susceptible to wind loading. But those things are all kind of around the periphery of the problem and not at the heart of it. Now you can certainly simplify elements of the problem by not considering the broader application, but the overhead and actually kind of designing the system so that it can work across the 2 is relatively minimal, if you've architected the system correctly. And so we think that there's a tremendous opportunity, therefore, the synergy that we get by operating both trucks and light vehicles is really valuable. And we're going to be showcasing what we're intending to showcase some of that at the end of the quarter.

Got it. And so let's talk about the process to development on the truck side. So can you give us a sense of where you are in the testing process? Sort of what is your path? I know simulation is something that you spend maybe more time on than your peers do. Can you just talk about sort of where you are generally in that process?

Yes, absolutely. So maybe talk about simulation first.

Sure.

So I had the privilege of helping found what's now Waymo and kind of building that team for many years and that technology with that team. And we pioneered the notion of let's get a bunch of vehicles, drive them out on the road and gather all the data we can because we'll learn super rapidly. And it's a really good to kind of get started, but it turns out it doesn't get you to the finish line. But it's important to realize that fatal accidents in the U.S. roads happen about 1.15 per 100 million miles. So trying to just kind of brute force gather data and hope you get the right coverage, just won't work. And so our approach has been, okay, let's be smarter than that. What happened in every other engineering discipline? Well, it moved from let's kind of draw it up and make it in a shop and then beat on it and see how it breaks to, let's design it in CAD, let's use computer engineering -- or computer-aided design techniques to test it, look at what happens when you heat it, when you shake it, refine it in the computer, then bring it out to the world and validate it in the physical world. We're taking that same kind of approach here. It means we can be dramatically more efficient. We can drive orders and orders of more equivalent miles and simulation than we can drive on the physical world than anyone can drive in the physical world. And so we think that's immensely valuable and allows us to move more quickly. And it was one of those insights that came from being able to kind of restart the program, restart the development from clean sheet when we founded Aurora.

And so when you think about numbers of trucks in terms of some of the metrics we can think about, how many trucks do you have operational today? What's sort of the pace of activity in terms of collecting data in the real world side relative -- and then maybe proportionately relative to the amount of miles you're capturing on the simulated side?

Yes. So we generate -- well, we've gathered billions of miles in simulation. We have, I don't know, 4.5 million miles or something like that, of physical world data. And today, we operate a fleet of something like 18 trucks with really the focus of learning with our partners and kind of validating and testing the work that we do in the development of our system.

Got it. And so how do you think about the path to commercialization on the truck side? So getting to the point where you're moving consistently revenue loads for your customers and then ultimately, moving towards that driver out. That is at times a somewhat controversial discussion and maybe it's probably a stand-alone topic to talk about. But sort of how do you think about that process? And what is your specific path to market?

Yes. So we're working with partners today. We're pulling loads on a daily basis for customers. We're really excited about that. We're doing it with our truck actually driving in autonomy. We have operators on board and are pulling those loads. So we think that's -- the goal for us is not to be a carrier, right? We work with amazing carriers. We want to provide them a driver to complement the human drivers that they have a shortage of today and do that efficiently and well and in partnership with them, not in competition. We look at that question around driver how and what does that mean? Again, for us, it's really about, are we delivering commercial value? Are we on the right path to that? And are we doing that in a way that's safe? And so as we think about getting to a commercial operation where there aren't people on board the trucks, we see this much more as a continuum. That today, any one of you could get in your car, you can get on a freeway. You can wait until there's a pocket with not a whole lot of traffic around you. You take your hands off the steering wheel and close your eyes. And for some amount of time, your driver out. Now I would not recommend you do this, and you wouldn't want to do it for a prolonged period of time, but you have a self-driving vehicle for that short period of time. And so we think about it as, okay, how long could we reasonably expect the vehicle to operate without a driver on board, without them having to touch an interface? And are we doing that in a commercially relevant way? Are we doing that in a way that as we scale this up, that it's a real business? And so for us, we'll continue to push in that direction. We'll focus on closing our safety case, the set of arguments that explain why the vehicle is safe to be on the road doing that in a way where the mitigations are actually commercially viable, where we're not taking 1 driver and turning them into 4 and operating around -- operating with them on the road, but are actually really truly delivering the value of what we're building.

Got it. That makes sense. A couple of questions that have come in on the webcast. I want to make sure we get those included here. So one of them is, what is your value proposition or the moat around your business? And how is that different relative to maybe some of your peers in the space?

Yes. So I think there's a number of ways we think about the value of the company. So one is -- yes, it starts with just the quality of people we have. If you kind of poke around the industry and you look at the density of high caliber folks, I think we're in an incredibly strong position there. They're building amazing technology. We have a substantial IP portfolio. At this point, I think we've got 1,000 and something patents pending and awarded. So that's all well and good. Then there's the partnerships that we've put in place. So again, PACCAR, Volvo, Toyota, these are world-leading brands. One of the things we're really proud of with our partners is that they're choosing to work with us, and they're -- they've got skin in the game, right? There -- each of these partners is running a tens of millions of dollar program to develop a vehicle that's compatible with your Aurora Driver and in partnership with us. I think that's a really meaningful investment. We think that, that kind of partnership is one that stands the test of time and will actually deliver value for our investors and for their customers. So we think that's really important. I think on the technology side, some of the investments we've made are meaningful, whether it's what we're doing with our proprietary FirstLight Lidar, we don't see others be able to do that, right? The range that we can see, the quality of data we get back, the ability to measure speed instantaneously, right? It's really important to actually deliver a safe product because you can drive on the freeway with much less sensing or eyes because most of the time you're driving on the freeway, nothing happens. Most of the time, the vehicle that's in front of you kind of stays about the same distance in front of you, kind of cruising along, maybe somebody goes past you briefly. But most of the time, it's kind of quasi-static. When things get exciting is when it's not. When you're moving at 65 miles an hour and there's a thing out there that stops suddenly, and you now have to manage that load and react to it quickly. You have to react to it with time to stop. And so that's where the difference between kind of getting by and actually having a viable product comes, and I think that's one of the cool technologies we have. And that's why we chose to bring that inside as well.

Yes. Makes sense. What are some of the milestones that we should be thinking about over the course of 2022, 2023? I think the industry, in general, starts to think about 2024 is a period of time where autonomous trucking might become a little bit more scaled into the market. So how do we think about your specific milestones along the way?

So we have a couple that we've shared recently. So at the end of Q1, we're working to demonstrate this concept that we've been talking about of transferability, the fact that the Aurora Driver is engineered for tractor trailers and engineered equally well for light vehicles. And so we're working towards demonstrating that transferability to the Toyota Sienna at the end of this quarter. In Q3, we want to demonstrate the fail safe behavior of our vehicles. So this is the ability for it to detect failures on board, react to those and bring the vehicle to a safe stop and handle those because, again, having a vehicle operate on the road in kind of nominal conditions is not that hard. But be able to deal with and making it robust and productizable so that it's safe, is actually where the hard work lies. And so we look forward to showcasing that, what we think is a major milestone. And then again, for end of '23, is when we're looking to launch the product and actually have vehicles operating in a commercially viable way without operators on board, serving customers.

What has been the customer response? So you have some significant partnerships, as you highlighted on the carrier side. What has been, as you have come public, have had more sort of public information in the market. Does that attract customers to the name?

Yes. I'm not sure if it's being public or just that we do a good job. And the way we approach partners, I think, is one of true partnership and try to learn from one another. So for us, the process we're going through with partners is one where we first kind of approach them. We learned from one another. We have a conversation around getting their kind of nonbinding level of interest and kind of volume of interest. We'll then move to a second phase where we collaboratively look at the routes that we have available, the lanes that we have available and allocate the supply that we have. And then ultimately, we'll get to binding orders. Today, that initial statement of intent or statement of interest that we have substantially exceeds our ability to deliver the driver -- the number of drivers through 25. So we're feeling a really great place about the level of interest from customers. Now of course, that's all nonbinding. We'll see what happens as it all shakes out, but it's a really exciting level of interest we're seeing.

And the receptivity of the carrier fleets in general to the idea of the technology, weaving it into their existing fleets is quite high at this point?

That's what we're receiving, what we understand, right, is that these are companies that have great businesses that are challenged by a shortage of drivers. And so for them, that opportunity to kind of build their business, to increase their revenue per truck, to tackle some of the kind of painful -- pain points in their business is real. And we see the Aurora Driver as a pathway and solution for them in the that space.

Got it. I want to get into a couple of these questions here before I do want to move on to the partnership with Uber and some of the stuff on the passenger vehicle side. But a question coming in about OE partners or your peers all have partnerships with OEMs, for the most part as well. What do you think is sort of the addressable market, given the fact that we've seen some pairing up amongst the names and the producers themselves. Can this become more universal outside of these set relationships as they stand today? Or will they stay relatively fixed going forward?

Yes. I think that most of the companies are not going to make it in the space. I think it's a really hard problem. I think you need to have the right partnerships. You need to have the right team. You need to have the right experience. And so yes, I don't think the dance cards are going to -- are full yet. We're pretty excited about having roughly 50% of the U.S. market aligned with us.

A big number.

It's not bad. And so we'll continue to see how we can build that.

Okay. Makes sense. A couple of more questions, but maybe some that naturally transition over to the passenger vehicle side. So maybe the first, just before we leave truck, and I think this is more of a passenger vehicle dynamic, electric vehicles. I think as it stands right now, the technology around the Class 8 heavy-duty trucks is a little bit challenged from a battery capacity perspective. But any thoughts around EVs, as it pertains to the Class 8 trucks.

So we'd certainly welcome it, right? That, obviously, we're all with on the planet. We would like to continue to do that. And so reducing greenhouse gas is important as our partners bring EV vehicles online. We look at that. Powertrain is a relatively minor difference between -- from our perspective. It's obviously a lot of work from theirs. But from the Aurora Driver's perspective, it's a relatively minor difference. And so we're excited for it.

Okay. Makes sense. So let's talk about the relationship that you have with Uber and what that might ultimately open up for you from a market perspective on the passenger side? So can you talk a little bit about what you sort of see as a path of commercial operations on the passenger side? So what are the first steps to getting there? What are the sort of end use cases that you can develop?

Yes. So we're obviously excited about that. It's -- we think about this opportunity to serve our freight partners and customers there and then to help serve the ride hailing space as well. The relationship with Uber is deep and special. So they're our largest shareholder. They're still a minority shareholder. We have a decade-long agreement with them to have access to their data, that is really a super power. And let me kind of put in context why we see this as such an exciting and important partnership. So it goes back to our core philosophy of let's go work with great companies that know what they're doing and then -- that are excited and need the technology. And so we think about trying to replicate Uber. It feels like a bit of a fool's errand, right? Billions of dollars have gone there. They have a really great product. Let's go support that. And beyond that, if we try to replicate Uber, well, it turns out you have to serve all of the customers in an area because otherwise, they'll go use Uber or they go use Lyft because they just want the convenience to be able to click a button and get a ride. And if they have to think, does this my whatever go there, then they're not going to use it. And so for us, the approach was let's go work with this great company, let's feather our capability into their network. So early on, the Aurora Driver will operate on routes that look from very much like trucking routes. They'll start somewhere, like an airport perhaps. They'll drive somewhere a longer freeway and then pull off at something like a commercial district or a hotel district and back and forth. It turns out that's a small fraction of the trips that people take Uber for, but it's an interesting meaningful fraction. And it turns out that high-speed capability is essential to unlocking most of the economics in ride hailing. And maybe 40% of trips or something like that require high-speed driving, but 60% of economics do. And so as we go into that market and partner with them, we expect to be attacking that space from that end. And we think that's really the right way to enter it. And then as we get our vehicles deployed in the market, we're serving some of the trips, humans are serving the remainder of the trips. We can then use the data we have from Uber to understand, should we add this feature or that feature, right? It creates this crystal ball that allows us to understand the ROI of each engineering investment when we move -- going forward, and it will obviously guide our deployment so we give you -- most efficient as a company in that space. And I think that's incredibly exciting, and that's a very powerful way to be able to build a business.

How does that sort of work into the timeline? So you mentioned before trucks going to be the first product to market. So how do we think about the passenger side?

Yes. So we're working towards the end of '23 for our truck product, and we're working towards the end of '24 for the ride-hailing product. And again, very heavily leveraging the same capability. So our launch will not be in downtown San Francisco, right? It will be on a -- like I said, something that looks like driving from an airport to driving to hotels or maybe it's hotels to a convention center, but something of that vein.

And would be -- the idea would be island at -- and a random airport, I walk out, I hit my Uber app to go to my hotel for that evening. And it's almost -- it's agnostic to me as I walk up, the Aurora Driver pulls up.

I imagine there's a choice of it that you say, no, I don't want one of those or I do.

Not that quite comfortable with that yet.

But yes, that's really in -- that it's really about providing you the end customer a seamless, safe, easy access to transportation. And so we'll kind of fit into that network that they have, right, and provide a service through that.

Okay. So you'd be a portion of that fleet ultimately. Okay. Makes sense. Can we talk a little bit about the regulatory environment. So the safety is paramount in this because we need to kind of get to this point and prove that out overtime. And I know that's something that you guys are working and take extremely seriously, and then there's the sort of regulatory piece. So I guess, first, maybe let's talk about the safety case and sort of how comfortable you are with where you are and what maybe the next steps need to happen to kind of prove that out? And then I'd like to talk about regulatory and the process there.

Great. So we do take safety very seriously. When I said the -- we deliver the benefit of self-driving technology safely, quickly and broadly, it was not an accident, but safely was at the beginning of that. And so we've been trying to adopt best practices from other places in our approach to safety at Aurora. So we're one of the first companies, if not the first company, to have a safety management system, which is a concept from aviation, where we bring kind of practices about culture and how we operate and how we track failures internally. So that we can make them better and avoid them in the future. And then we've built our safety case framework, which is really -- think of it as like a structured explanation, structured argument as to why the vehicles will be safe on the road. And that consists of -- the first part of it is proficiency. So basically does it drive well? Does it drive safely when everything is working? The second is around fail safety. So this is when something breaks, do you behave in the right way, you detect it and then mitigate the risk from that. The third to continuous improvement or something to that effect, which is, are we learning from the failures we've had on the road? And then there's 2 more columns, which are really about the processes around the company and the operations that we have. And so we've put a lot of effort in that. We're the first company to share a safety case framework broadly and how we're executing against that to launch the product.

Got it. Makes sense. And so let's talk about the regulatory dynamic here. So clearly, there's the ability to operate and operate commercially in a number of states around the country. California is sort of the exception as it stands right now, something we're waiting to get kind of full approval there. Any thoughts around that process? How long will that take to play out. When you're in the market in 2024 with a commercial product, do you feel confident that California will be part of that process?

Frankly, no. Not yet in California. We hope they would be, but I think that's really for the government of California to -- it's such an important state for shipping. This is going to be super important for the economy, and we hope that California will act to address this as the legislator sharing California indicated they should. So we're -- we'll continue to work with them and partner to kind of move that forward. But the good news is that at a federal level, there's the freedom to operate and deliver the technology. And there's many states, the vast majority of states are pushing that forward and supportive of it. And so we're proud to be operating in Texas today. It's obviously one of the largest markets for freight in the U.S. It's going to be an important initial market for us. And we'll kind of grow out from there, as we learn more with our customers and explore and deliver on the routes they need.

Is there anything that you need in the near term from a regulatory perspective to continue to sort of build out and move along the plan that you have internally towards commercialization?

No. I think in the near term, we feel in good shape, right? There -- we don't see a regulatory barrier at this time. We -- one of the things that we believe, again, pretty deeply in is, this is partnership sense, right? And that's not just with our commercial partners, but with our partners and government as well. And so we've been engaged at the federal level, at the state level, at the local level and the communities where we operate to help those stakeholders understand the technology, allow them to have an informed perspective, allow them to make good policy because of that. We'll continue to do that. We think it's just kind of good business to behave that way.

Another question on the webcast came in around the OEM partnerships. And given what we've seen with supply chain challenges, we're just curious of their ability to continue to move the ball forward on their side towards the potential for scalability in 2024 and beyond. Have there been delays, hiccups or any other issues that have impacted that?

Yes. I think we're all being impacted by the supply chain challenges to more or lesser degrees. All indications are that we should be over this hump before the point where that's a real -- any kind of significant risk to scale in the business.

Got it. Makes sense. Let's talk in, in sort of the last couple of minutes we have, about the sort of competitive landscape as we look out several years from now. So there's a number of people in the market. I think you had mentioned that maybe some people will be successful, maybe some people won't be successful. I guess, what do you think the market can support? It's obviously a very large -- I'm afraid that market is very large. Very fragmented as it stands from a carrier perspective today. So how do you think about the ability for the market to absorb multiple players?

Yes. I think the -- from a market absorb or support, I think I'm not concerned, right?

No problems.

I think that for the foreseeable future, the competition for the Aurora Driver is really going to be human labor, right? And that creates a really great pricing umbrella to support, right? We think we're going to offer a product that will be able to drive without the boundaries or the limitations of hours of service limitation. That is just as always attentive, right, doesn't have kind of the natural human limitations on attention and doesn't need biological breaks, right? It doesn't need to stop for food, that it will just be able to operate. And this will lead to better revenue per truck for our customers, and it will leave to safer roads. And so we think that's all in that win.

Okay. And from a technological perspective, maybe that's where we begin to see some bifurcation of winners and losers.

Yes, I think that's it, right? It's a combination of business and technology and funding, right? We think that this is a really hard problem that it's a deceptively hard problem because you can make early progress that feels good and yet still not be able to kind of climb the S curve and actually get to something that's scalable. And this is one of the things where I think we're in a unique position at Aurora is that we have a founding team, and we've grown a team of folks who have kind of been there, whether it was my experience helping found and then leading what's now Waymo for 7 years or Sterling launching Model X, an autopilot for Tesla, and Drew and his experience in machine learning and what -- it's just -- I think we have a great foundation that's allowed us to make the right long-term strategic bets that will get us to a product and scaling.

Last question for you. Any lessons learned or anything interesting from going through the public process coming into the market and now being a public company?

I'm sure there are. No. So I think the process that we took was a little atypical for -- I think for anyone who went through this SPAC process where we really just looked at the SPAC vehicle as an opportunity to become public, and we wanted the experience for our investors to look like that of a high-quality IPO. And so we began that process by talking to the folks that we thought would be kind of the core tenets of what would have been an IPO or a SPAC, and we ended up with the SPAC. That allowed us to go through and kind of make sure we had the demand that was needed, the right point -- price point for that demand. And so that led to what was, I think, a very successful SPAC process, where we ended up with tremendous blue-chip investors. T.Rowe, Baillie, MSIM, Fidelity, right, anchoring it -- it's what you'd want on the covering IPO.

Absolutely. Absolutely. Well, listen, I think we are out of time. There's lots more to cover, but we don't have enough time to do it. But thanks so much for joining us today. I really appreciate it.

Appreciate it. Thanks so much.

All right. Thanks, everybody.

Thank you.