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

Great. Maybe we can start. Thanks, everyone, for joining. I'm Ravi Shanker, Morgan Stanley's Freight Transportation, Airlines and autonomous trucking analyst. And we have literally saved the best for last. And we are very happy to have with us Aurora Innovation CEO, Chris Urmson. Chris, thank you for coming back to the TMT Conference.

Thanks for having me. Great to be here.

Before we start, as you pointed out, I have to tell you that for important disclosures, please see Morgan Stanley's research disclosure website at morganstanley.com/researchdisclosures. And please read these disclosures. So Chris, it's probably been the most important and successful 12 months in Aurora history. You guys have cleared a lot of milestones over the last 12 months. So for those less familiar with the story, can you just give us a really quick recap of what's happened over the last year and what catalysts we can look forward to in '26?

Yes. It has been a heck of a year, and it's -- '26 is shaping up to be even better. So we focus on delivering the benefits of self-driving technology safely, quickly broadly in trucking. In '25, we launched the first driverless trucks operating daily in the U.S. We're the only company operating them regularly every day. It's exciting to have them on the road. Over the course of the year, we went from operating in daylight conditions on a particular route and good weather to now operating day and night across a variety of routes in the rain and more wind and that kind of thing. '26 is going to be even more exciting for us. So at this point, we're running kind of a handful of trucks on a daily basis driverlessly. By the end of the year, we'll be running a couple of hundred trucks driverlessly. In Q2, we're going to launch our second-generation hardware on our newest truck platform from international and have that operating again driverlessly, nobody on board, serving customers. And we're going to grow across the Southern U.S. over the course of this year, demonstrating our ability to take this generalizable, verifiably AI technology that we have and apply it in ways that really meet customer need. So it's going to be very exciting on that front. Those couple of hundred trucks, we expect to end the year using those to generate $80 million revenue run rate, which is going to be a nice step forward for us on the revenue front. And then we continue to make great progress with our partners. With Volvo at the end of last year, we had the first trucks come off the assembly line where they had taken our hardware, integrated it in the factory and had it roll out. It is a big step because we're not just working on the technology and making the technology talk to each other, but actually working with their manufacturing teams so that they can actually produce these things at scale. We've put in place a capability to upfit trucks with by wire and the Aurora Driver that's going to allow us in the back half of this year to be producing 20 trucks a week, allowing us really to grow aggressively to meet that customer demand. And then Aumovio, our partner for our third-generation hardware continues to make strong progress so that we'll be in place with the capability to build tens of thousands of trucks coming in 2027. So really just going to be a whole lot of fun this year.

Got it. Maybe just to take a step back here, there's a lot of focus. When people talk about autonomous, they think of robotaxi, not as much on the trucking side. But it almost feels like trucking is a better use case, a better application for autonomous maybe to begin with, and that is why you guys chose trucking over robotaxi, which you will eventually plan to get into. So can you just walk us through your vision of how autonomous trucking rolls out? How does that impact the economics of trucking? And what share of the market do you think Aurora can ultimately capture?

Yes. So -- you're right. We think of trucking as the right first market. It's a very hard market. Actually operating these 70,000 pound things on the freeway at 70 miles an hour is extremely challenging and make sure you can do that with conviction and the safety and safe operation of the vehicle. The great news is we've cracked that. We expect over the course of the next several years, to basically get to the point for customers where they're not operating our system, they're really not going to be competitive. The economic opportunity for them is really profound. We can take the asset, the truck that's one of the biggest parts of their cost structure and double the utilization of that, moving from 11 hours a day, which is what you limited to with a person driving a vehicle to approaching 24 hours a day. Similarly, one of the big challenges that operators have is a combination of high turnover in the driver pool and the fact that over the next decade, we expect to need 1 million more drivers, and we just don't know where they're going to come from. Aurora Driver will provide a way to complement the people they have driving trucks today, that's basically a safe, scalable pool of drivers. On the economic side, on the bottom line, we expect to have some huge benefits to customers. Today, the cost of a person driving truck is about $0.995 per mile, plus another $0.15 to $0.20 of indirect costs. We expect to price at $0.85-plus per mile. So that in and of itself is a significant economic benefit. You add to that the fuel economy benefits. So we expect between 40% and 34% improvement in fuel economy, which is huge when you think of fuel as one of the big 3 costs for these drivers -- for these companies. And then, of course, the safety benefit of this is real. You won't talk to a trucking company who doesn't lead with safety is the most important element and the combination of the Aurora Driver's superhuman capability that is paying attention 360 degrees to be able to see further at night than a person can, be able to react without ever being distracted, it's just going to be profound.

Got it. Understood. Just looking at the autonomous trucking space, like you said you guys are leaders, kind of you're the only ones doing what we're doing right now. There are a few -- a handful of other companies who are trying to attack similar or adjacent parts of the business. Can you just talk about what makes Aurora unique? Is it a tech stack? Is it OEM relationships? Is it you and your genius, kind of what really sets the company...

Let's not be the last one.

Highest multiple.

That will be a short shortfall. So I think these are like all of what you mentioned, except for the -- but the fact that we have verifiable AI, the fact we're the only company running this stuff for real on the road today that our verifiable approach -- verifiable AI approach is just critical when you think about we're the tip of the spear for this physical AI revolution that's happening, making sure that we don't just have a thing that we think works, but that we know works and have conviction in that. When we talk about the OEM partnerships, the fact that we work with 3 of the 4 OEMs in the U.S. today, no one else can say that, right? And obviously, you can't have a driverless truck without there being a truck. When we look at our hardware supply chain, our second-generation hardware will be produced by Fabrinet and now it takes us up to about 1,500 units. But then the partnership we have with Aumovio is a huge competitive advantage. As we look from the hardware we have in vehicles today to our second-generation hardware, we expect the cost of those components to drop in half and the durability of those components to triple. As we move to the Aumovio hardware, we expect that cost to come down by another -- or the cost to come down by another factor of 2. At that point, we really can not just scale because we have the quantity, but scale because we actually can have the price performance that we want the cost of goods sold to build a hell of a business. And then the relationships we've been building with customers are deep. These folks have been with us for several years. I'm a big believer that they care about safety. They care about that this will help them build their business. Once we start working with these customers, once we are able to have a strong partnership and build value with them, it's going to be very hard for someone else to come in and play, right? Do you really want to put the safety and risk your business when you have something that's working well with somebody who might come in late with some other offering.

Got it. Understood. Just one follow-up there. Kind of you mentioned AI, obviously, a huge theme of this conference for the last 3 days. It's come up a couple of times. But can you just talk about your view of how AI plays in the development of this technology? Because you guys have been around since 2017. You've been one of the first -- the first time I heard of simulation and saw what you can do and kind of blew my mind kind of several years ago. At the same time, you also have had actual rubber on road and are testing in the real world as well. What is the right balance between simulation and real-world testing in your view?

You need to use the right tool for the right job, right? So we -- first and foremost, simulation is great, but actually a simulated test is actually what you want. You need to not just be able to kind of make the world move forward in some simulated way. You need to know whether it was good or not. And that actually turns out to be quite hard and another step beyond just simulation. For us, it's also important that, that simulation is not just pretty pictures, but it's actually grounded in reality. So when you get a return back from an object in the world from your LiDAR simulator, is it actually coming back at the right probability? Does it have the right intensity, right? These are things that if you don't put attention to, you get something that looks pretty, maybe looks to the human eye as representative, but has a meaningful difference in the statistics of it that are going to cause problems when you actually see -- compare the performance relative to what you see online. So for us, we've been a believer from day 1 that it is less about the quantity of data, it's about the quality of data that you have. And so we target collections where we need to. We have a variety of test tracks we'll go out to gather data, but then we can amplify that data, we can create scenarios that we wouldn't otherwise be able to safely generate in the real world through simulation.

Got it. And just to close the loop on the tech side, not everyone is using the same tech stack in the industry as you would expect.

I hope not [indiscernible] stuff.

Fair enough. Do you think it's kind of there's going to be one singular approach to this? Or do you think there's going to be 4 or 5 different players with 4 or 5 different technology solutions kind of -- but all of them really work. I think the biggest -- I won't say controversy in the industry, but kind of the question mark is around the need for etchy maps or not. So what's your answer to the mapping question? What's your answer to LiDAR versus vision only? Or do you think all of them work?

I guess my answer is that we need to deliver a safe system and we deliver it quickly, right? And that's kind of the start of our mission statement. And so when I think about maps, what are computers really good at? They're really good at storing and recalling information. We should lean into that. It turns out it's a hard engineering problem, but done well, you can actually leverage it. I also think that when I drive near my home, I'm a much better driver than when I drive in some other part of the country, not because I lose my skills in another place, but because I have less understanding about the world, and I'm reacting in real time rather than be able to kind of have a cash model of what's there, I'll be able to use that and exploit that and make my decisions. So it seems fairly obvious that be able to know what's coming in the future rather than having to figure it out instantaneously is the right answer, assuming you can do the engineering to make that happen. It turns out we've been able to do that. And our ability to build maps is now really an operational problem. It's not a technology problem. And when I say problem, it's just -- it's a thing we have to do. We know how to do it. We're not at all concerned about the ability to scale those maps. When it comes to the sensor suite, we are strong believers in multimodal sensing. We believe that because they have different failure modes and benefits, cameras have higher resolution, but don't see depth directly, can get obscured by things. Radar will punch through things, but it's lower resolution and doesn't see color. Our FirstLight LiDAR can see further than anyone else can and allows us to get geometry at a range that you wouldn't otherwise be able to. So we see that complementary set of capabilities is really important. We see that play out in practice when we're driving through weather, when we're driving through dust doors, we're driving at night. And so the question really is not -- is why wouldn't you use these things, industrialize them, bring them to a point where they don't impact the cost negatively and make it work. If you ask anyone working in AI, would they like more data or less data? They're going to save more. And this gives us this complementary set of data. So yes, we're big believers in multimodal sensing.

Got it. So maybe a follow-on question is we were discussing this before the start of this session that when I speak with investors on your story the goalposts keep moving, which is a very good thing because it shows that you guys are moving to the goalpost. And I think the most recent one or where we are right now is it's gone from does this work? People are now convinced about to can this scale, right? And one of the questions we're getting is, does the need for mapping constrain your ability to scale, right? So can you address your confidence in scaling, both in '26 and as well as when you start commercial production next year from the perspective of routes and operations and kind of where the truck can work and then also scaling from a volume perspective and kind of getting up to 200 trucks and then 1,000...

Yes. so on the map scaling side, literally no concerns, right? That this is a thing where we have built the technology. We know how to roll it out. We were able to open up 10 lanes, go from 1 lane to 4 lanes to now 10 lanes very quickly. It's really going to be driven by customer demand and scale of vehicles. It turns out if you have a handful of vehicles that you can operate driverlessly, opening up 20 lanes is not useful because you don't have any trucks to put on them. So for us, there -- we'll continue to advance and improve the efficiency of that process, but this will not be a limiter for us as we scale and build the business this year. We've also unlocked the ability to get a couple of hundred trucks on the road. So we now have a partnership with Roche, where we'll be taking these international trucks. They'll taking our by-wire overlay. We'll be taking our Aurora Driver second-generation hardware kit and installing them. It turns out upfitting is a very common practice in trucking. So we'll be using this as an upfit facility. And then these trucks will be deployed into market, like I said, in the back half of this year at 20 trucks a week. So far, we're getting very positive signal from customers. They're excited for this as we start to unlock these long lanes like Fort Worth to Phoenix, which is 1,000 miles, no human can drive that in a day, at least not legally, right? And so the fact that we can start to unlock real value for customers over the course of this year is going to be a big way that we see this kind of supply that we will now have get absorbed.

Got it. And just to clarify to the audience kind of when you say scaling up to get to commercial production, like eventually in '27 when you have the full launch, you're not actually going to make anything. So there's no CapEx cliff kind of there's no...

No, we don't see a CapEx cliff, right? So we work in partnership with others. So the hardware that we're putting in the vehicles this year is our Gen 2 hardware. Next year, as we -- as the Aumovio hardware is produced, we have this phenomenal partnership with Aumovio. It's an incredible team, incredible company. We're working together to deliver that third-generation hardware. The Aumovio folks have invested and they're investing roughly $350 million to allow us to do the engineering work together to put in place the manufacturing, and ultimately, they finance the hardware, and we pay them based on the utilization of that hardware. And this is actually one of the most lovely bits of partner alignment that you can imagine. Our customers get paid more, the more their truck drives. We get paid more, the more that customer drives the truck and Aumovio gets paid more, the more that we drive that truck for a customer. And so through the stack, we have this complete incentive alignment of we want that truck out there operating effectively and serving ultimately our customers, the shippers.

Got it. Just to close the loop on the topic of scaling. You guys did your first Driver Out run on a public road end of April. You just crossed 250,000 miles with just a few trucks in your fleet. So that was a pretty good scaling right there. Obviously, no reported incidents or any issues there. So what have your biggest learnings been? Like has this been like exactly as you expected? Like what have been the biggest surprises, positive or negative?

Yes. It rains a lot in Texas. You can look at the weather and you can see like here's the rain. It turns out somewhere around 40% of the days this last year, we weren't able to operate because of various weather conditions. And so that was -- as we -- as kind of entertaining, but it is what it is. We now have the ability to operate in rain and we'll continue to enhance the kind of the level of rain that we can operate and the level of environmental conditions we can operate over the course of this year. So taking that off of -- out of play is a big deal for serving customers. You don't want to be Hirschbach and say, sorry, the berries can't get to the store today because it's raining out. So that's a big step for us. I think the other -- like there's this thing that you have to have belief when you're building a company, building technology that you're going to get there and you're like, yes, we're doing the things. I believe we're doing the things. This is going to work. And then there's a moment where you actually did it and you see how it played out. And I think in terms of the process we use for releasing and how we've made that repeatable, we can push new releases on a regular basis, the fact that the performance fee on the road for the most part, aligns up with what we expect, right? Like that's actually been a big deal because it gives us continued confidence that we're kind of on the right path. And this will scale, right? If we kind of test one hypothesis and that succeeds, then kind of the extension of that hypothesis becomes more likely to succeed. And that's what kind of this place we're in right now.

Got it. Maybe switching gears a bit and talk about a few numbers. You referenced the $80 million revenue run rate for 2026. Can you just talk about the big moving parts there and kind of what will get you up to that level, obviously, scaling up the fleet, but also give us a few more details on the Detmar partnership that you guys announced.

Wonderful. Yes. So we get paid on a utilization basis. So we drive for customers. Today, we own and operate the trucks. We effectively operate like a trucking company, but we do it under the flag of our partners. And so that's reflective of miles we expect to drive heavily weighted in the back half of the year. The Detmar partnership is one of these really wonderful ones where we were not thinking about this as an application. This customer saw the progress we're making. They've seen some of the press about what we were doing. They reached out and said, "Hey, we're moving sand between this mine and this distribution center. We need to drive on freeways. Nobody else seems to be able to do it and you. And so of course, we looked into it. They basically want to create a virtual -- not treadmill, basically virtual conveyor belt with trucks. So the more the trucks drive, the more sand they can move, the happier they are because their customer has basically an insatiable demand for this. So this has been one of those places where just by -- somewhat by happenstance, customer reached out to us, and this will become an incredibly interesting bit of business, and we expect to be able to grow that dramatically.

Got it. So just a follow-on from there. What have the customer conversations been like? Because this is historically viewed as a fairly conservative industry where people are somewhat skeptical about technology. But at the same time, if you can give them technology that will actually save them money, improve their productivity, people are all in on it because math is the only thing that matters in this industry, right? So have the customer conversations ramped up after you've past these gaps and these goalposts? Or are people waiting for a final launch to kind of fully roll out?

So I'd say, first, there's a distribution of customers, right? There's no one kind of persona. There's different flavors of risk and appetite. What we have seen is that ability to move from the hypothetical of what this could mean to your business to the practical of, no, this is real. It's operating driverlessly. It's about to start scaling out to places you care about, really has changed the tenor of conversation. At this point, we effectively have more demand than we can possibly supply in the next little while here. And so that's just a very exciting place to be as a business. And what's been even more exciting is that the -- the conversations have changed, particularly with folks who've been with us for a while from I'd like to get access to, to I'd like to get access to and ensure that I have priority on this. I don't want someone else to come in, right? And they ask, we get exclusivity? The answer, of course, is no. But for folks who've been with us for years, we want to reward them. We want to say, like, thank you for making the bet with us. We'll help you grow and build your business. And then we'll expand from there as well.

Got it. So maybe moving through some of the capabilities of the truck itself. Obviously, you referenced that you can drive through rain and fog as well, kind of you noted how often it rains in Texas. Obviously snow has been always kind of big question mark that people have had. Just talk about the kind of capability of the truck and when do you get to a point where you can cover the entire country. And to what extent does that unlock more of the market for you?

Yes. So obviously, being able to drive more places just intrinsically unlocks more market, right? if you can't go the place the goods need to go, it's not helpful. What we expect is over the course of '26 is we'll unlock the Sunbelt and then we'll start to push north from there. I expect this will just -- this will actually move very quickly. Today, like as you said, we operate in a certain set of weather and environmental conditions. But we're already doing the work to start pushing towards snow, that doesn't sow in the summer. So we've got trucks out gathering data in the Rockies right now and chasing snow to make sure we've got data sets that we need. We look at this, and we don't really see this as like there's a thing made of it, but it's not really a big thing, right? Like people talk about it because early on, we want to get something on the road. It's useful to be places and then we'll kind of add capabilities in. At some point, we'll get snow in there. I think we're planning to do light snow certainly by the end of this year.

Got it. Sounds good. I'll open it up to see if any questions in the audience. If not, we can keep going. So kind of...

I know it's a smaller private. Can you speak to Gatik as a competitor and how they are going to market versus you?

Yes. I don't know a whole lot about Gatik. They have been working on box trucks. It's hard to tell exactly what's happening. First and foremost, though, I look at this space, and it is gigantic. We believe we have a multiyear lead on folks. We welcome competition. We'll continue to execute, but I don't have any particular insight into that company.

Come back in a second. You referenced the way you're building up revenues over the course of the year. The margin trajectory, I think, is mostly tied to the second generation, eventually a third generation of the kit. How much visibility do you guys have into the math there and that cost reduction?

Yes. On the cost of the hardware, we have very good visibility into that. The hardware that we're building for Gen 2, we're manufacturing that today. We understand what the cost is. We have the supply agreements in place. So we know what that cost is. We feel confident in that cost structure. When we talk about the Gen 3, if anything, I'm more confident in that. One of the reasons you work with a company like Aumovio is their superpower is the process of manufacturing and the supply chain management. And so on a weekly basis, we're looking at the bill of materials costs. We understand the assembly cost, but we understand what the full kit cost is. So we feel very confident about where we expect to end with those components.

Got it. And any particular concerns around supply chain or other issues, like -- because LiDAR was a huge thing for a while and can you source them as such.

Yes. We continue to pay attention given that we're building something as advanced as we are, there are certain kind of single source risks that we have in the supply chain. I think most of the companies at this conference probably have some kind of supply chain risk with NVIDIA as a single source. And of course, we use their Thor SoCs. We're one of the lead customers for that. But no, we look at this. And again, part of the advantage you have with working with an Aumovio is that's what they do, right? They make sure they have a core team that we would never be able to hire that handles supply chain management and make sure the suppliers are credible and have alternatives lined up where it can be.

Got it. So 2026 will be when you launch the Volvo program, you'll take the observer out of the PACCAR truck and you have the upfit on the international program as well. So...

We haven't said any of that.

Maybe that was a pretty clear overview there. I tried -- sorry. But if all of these happens, right? So when do we get to a point where a driverless truck is the only Aurora truck that's running on road kind of at what point does that happen?

So we will be running trucks with operators on board indefinitely because we will always be developing and improving the system, but that will become a very small minority of trucks. By the end of '26, my expectation is we are already at a small minority of the trucks that are on the road, right? As we think about just scaling to hundreds of trucks out there, we can't have 500, 700 drivers sat by hoping -- or hoping they don't need to go drive that day. So our expectation is later this year, the vast majority of trucks that are operating [indiscernible].

Got it. Sounds good. And just on that point, right? So again, you guys have been hitting all of your catalysts and your key milestones, some of them ahead of schedule. But it still seems like there is some level of conservatism on the OEM side, kind of going back to the need for an observer kind of after you pull the driver out. So what is the risk that you guys do everything you have to, but maybe some of the OEM side or maybe the carrier side move slower than you would like in terms of adopting this technology?

Yes. There's risk in everything. If any company here tells you there's no risk in their business, I wouldn't buy that stock. But what we can do to mitigate this, one is we continue to have good strong working relationships with them. With the partnership we have or the relationship we have with international, where we're buying stock trucks, upfitting them, we've basically taken that kind of schedule risk from the OEM and taking that as one that we can control on our side of the ledger. And so we think that's an important way to kind of help mitigate that risk. But I think the biggest thing, though, is getting these trucks in the hands of customers at scale. There's nothing that motivates a company more than their customers saying, if I can't get the thing I need from you, I'm going to go buy it from your competitor. And so while we have enthusiastic partners, they're working well. I think as the reality of this is in the market, people are using it, they're demanding it becomes a thing. I think that will just help provide a virtual feedback -- a virtuous feedback loop through our partners.

Got it. That's good. Last call, if any questions from the audience? Maybe there's a couple -- a few minutes left here. So kind of just to hit on numbers again. So you're targeting breakeven gross margin on a run rate basis exiting 2026 and positive free cash flow in 2028. Can you just talk about the balance sheet and kind of the cash used and liquidity until you kind of ramp up?

Yes. We feel like we're in a very good place. At the end of the year, we had $1.5 billion in the bank. So that's a really enviable position to be in. As we look at how we expect the business to roll out over the next couple of years, we think that will carry us through to that run rate free cash flow or free cash flow on a run rate basis. The provisos on that are that we will use our ATM to fund the tax obligation and our RSUs, and we'll likely use it to cover the cost of our [indiscernible] but modular that, we feel good about the capital we Have. At some point, we're going to want to make sure that we have a responsible amount of capital on the balance sheet. We'll find the right time when -- basically when the catalysts have been reflected in the stock and find the right time to put a little more capital on the balance sheet. But we feel very good about that. It's a great position to be in where it's like, yes, we just execute. This is all going to work.

Got it. Speaking of medium to long term, kind of you guys obviously have the partnership. Uber kind of has a historical stake in the company. And you initially started at us doing both commercial trucking and robotaxi ride-hailing. So when do you think -- I know you guys are going full speed ahead on launching commercial truck, but what's the time horizon for ride-hailing and maybe kind of local delivery as well?

Yes. I think this is a really important idea. So the capability we're building is not just an ability to drive trucks, but it's a generalized driving capability. And even more broadly than that, we're developing the process and tools that allow you to launch a safety critical physical AI system, right? Something where if you do it wrong, people get hurt, and we want to make sure that we do it right. That travels. And so today, if you ask me if I could spend another dollar anywhere, I would spend it on our trucking business because the lead we have is profound, the opportunity is great. Just -- the more we can get that to happen, the better. But within the next couple of years, that business is going to be going, right? And we're going to be -- there's going to be execution to perform, but it will be off and running and folks will start to recognize that that's kind of on escape velocity, that's the point where I want to start looking, okay, where is the next best opportunity, the greatest ROI for us to take the competence we have and point at it. And today, it's not clear to me whether that is other elements of the logistics space and stay in that vertical or whether it's to start to look at ride-hailing and start to enter that competition in personal mobility, do that fighting downhill though, because we will already have scale in the system, which will help us drive the cost out of the Aurora Driver, allow us to go enter that market in a more competitive way, which is great. But there's also agriculture and mining, drones. There's lots of places where we can take that competence that we have, that physical intelligence, that verifiable AI approach and deploy it. And so it's going to get really fun. It's going to be a hell of a lot of fun building this truck business. But then as we start to send kind of green shoots out to go after some of these other adjacent areas, it's going to be very, very exciting.

Can you just expand on that a little bit because if you go back 10 years, the early days of autonomous driving, there was some skepticism that you could take a system off of a car and burn in a truck or vice versa. Does AI now let you do that kind of much easier than it could have been before?

I wouldn't say it's AI, but what I would say is it's the architecture and design of the system. So up until a couple of years ago -- maybe even 2 years ago, we were running our Aurora Driver on both Toyota mini vans and big trucks. So we have thought hard about how do we architect the system so that it can be applied. Frankly, when we go from moving a big box truck down the road to driving a little passenger car, it's easier. And so then when we think about, again, moving from Class 8 tractor trailers to Class 6 box trucks, like -- that's just going to go, right? And so it will -- we're very excited about going to take on those markets to serve customers in those markets to deliver the value there. But again, every dollar I can spend now on trucking is going to have a higher ROI than any of that. So let's go do that first.

Got it. I think one of the most common questions we still get in the space is on the regulatory side, and it feels like there has been a lot of positive traction in that area. Can you just go over how you see that as you kind of on the cusp of launching commercial operations?

Yes. We do not see regulations as a barrier for us to build a business today, period. If you look across the United States, the vast majority of the states today, you can operate driverless vehicles in them, driverless trucks, certainly across the whole Southern corridor. We expect what we've unlocked in '27 to be roughly a $50 billion SAM or 50 billion mile SAM. So lots of opportunity there. We continue to have very positive support at state levels. We're starting to have conversations in more states just to kind of broaden that in preparation for us to start expanding out of the southern corridor. At a federal level, we continue to see support, the regulations that we have, both from FMCSA and NHTSA support the technology we're deploying, the product we're deploying, and we continue to have strong relations there. And we are seeing this administration, at least so far, communicate very positively about the importance to the American economy and American competitiveness of automated vehicles. And that's from the Vice President through to the Secretary of Transportation and various others in the administration.

Got it. So Chris, take us home by telling us what you're most excited about in '26 and what we can look forward to in terms of catalysts.

Yes. I am just jazzed, right? This is going to be a year where we get a new truck on the road, second-generation hardware. We're going to be able to do that at scale. And it's going to be -- like I've been working in this field for 20-something years at this point. And we made a big step forward last year where we could put trucks on the road and nobody in them. But by the end of this year, if you were taking a road trip across the Southern U.S., you can't help see our truck, right? They're going to be out there serving customers, and it's just the first step to what will be an even more exciting '27. So getting to that point where we get to see this really take off and really go, it's going to be a lot of fun.

Great. We're very excited to see how that develops as well. So Chris, thanks so much for joining us.

Thank you.