Hello everyone, thank you for joining us for our next and last session of the Auto Tech and AI Conference. We really saved one of the best for last. With us today, Richard Tame, CFO of Aurora. Aurora presented several times at this forum in the past. You're gonna hear a lot more in his own words, but just a quick background from our side. Aurora was founded in 2017 by Chris Urmson, former CTO of Google's self-driving team, which is now Waymo, and is one of the pioneers, honestly, within the autonomous driving space. The Aurora driving system, which just recently introduced the final driving capabilities for its Level 4 self-driving trucking launch, is vehicle-agnostic, so it's also built to operate passenger robotaxis in the future.
Aurora has partnerships and has been working with several of the industry's most important players, such as Toyota, FedEx, Volvo, PACCAR, and more. Richard has been the CFO since 2021 and has been at Aurora since 2020. He previously served as Senior Director of Finance at Lyft, where his focus included its self-driving business. Richard also has a wealth of experience at several other Fortune 500 tech companies, including Microsoft, Amazon, and Facebook. Richard, thank you so much for joining us today. I know you have a couple of slides to give a little bit of overview of Aurora. Why don't we go through those, and then we'll come back for a little fireside chat after.
Perfect. Thank you very much indeed for having us, thank you to the audience for tuning in. Yeah, I think they probably saved the best session for last, we'll try to live up to that billing. Next slide, please. I'm gonna give you a brief overview of Aurora, then we'll jump into the Q&A with Chris. Looking forward to that. First, as is customary, also to keep the lawyers happy, here's a safe harbor statement. I'll give you a few minutes to read that. Not really. Stacy, next slide, please.
What I'd like to do is to talk briefly around the promise of self-driving technology, why there's never been a greater need for autonomous vehicles, and why we expect at Aurora that our technology and our path to commercialization will enable us to deliver this technology at scale. Together, we're building the next generation of self-driving technology, and we're focused on moving both goods and people autonomously. There's just really a vast opportunity, as I'm sure you can imagine, to transform transportation, and that's really what motivates our work in this space. Next slide, please. This is a quick look at our Level 4 self-driving product. We call it the Aurora Driver. If you're familiar with autonomous vehicle companies, many will have a similar diagram like this.
It shows the different components of the system. The Aurora Driver's hardware, software, infrastructure, and development tools are designed to work across vehicle types. The same hardware and the same software powers Class 8 trucks, all the way through passenger vehicles, all the way through to light commercial vehicles, ultimately. We think that, or know that this common core technology approach is unique in our space. The commonality is awesome because what that does is it ensures that the lessons learned, the development efforts that we make, the hardware improvements that we have over time, and importantly, cost reductions that are made to the Aurora Driver, you know, it benefits every vehicle that it powers. Something that the truck learns, the car picks up automatically. Something that a car learns, the truck picks up automatically.
What we do this for is we think it's super important because we believe that it's going to enable us to deliver on both our priority market, the first product we're going to launch is in the trucking space, and then down the line, the same hardware, the same software will allow us to go into the passenger mobility space. We're setting ourselves up to address not just one, but both of these significant markets in time. Next slide. You know, I don't need to tell everyone on the call, I'm sure, that these markets are massive. Aurora plans to first launch in trucking, which we see as a $700 billion market in the U.S., and the trucking industry has attractive unit economics for what we're trying to do.
We believe that this technology that we're building has the potential to meaningfully reduce the cost of ride-hailing. We believe that when you do that, you could thereby start unlocking the $1 trillion Personal Mobility market via shared mobility. Further down the line, we also expect that the Aurora Driver would work, and we'll be able to expand into Local Goods Delivery. Next slide, please. We're motivated by the immense impact that our technology can have, the most important thing being safety. It's the first part of our mission statement, you might remember. Safety is fundamental to everything that we do at Aurora. Just a kind of a stat that sort of is kind of mind-blowing when you think about it, is specific to trucking. There's approximately half a million truck crashes every year on U.S. roads.
There's only 365 days in a year, kind of a phenomenal number of accidents that happen now. As well as the safety aspect of it, we believe that autonomy can also solve a number of real pain points that there are in the trucking industry today. We believe we can alleviate driver shortages. The industry is short about 80,000 drivers today, and that shortage is projected to get worse. It's projected to double by the end of the decade. What the Aurora Driver can provide is a scalable, stable driver supply, which is super important to the industry. Currently, there's hours of service limitations around truck drivers. They're limited to driving for 11 hours, and then they must take a break. The Aurora Driver, being a computer system, it doesn't have those hours of service limitations.
It can operate 24/7. That means that we can deliver freight faster. We can get things where they need to go faster because we're not limited by having to take a break after a certain number of hours. That can potentially cut transit times in half. The trucking industry suffers from high fuel costs. What autonomous technology does, what the Aurora Driver thing can do, is it can enable more efficient vehicle operation. Also, there's an ability to reduce fuel use and emissions by 10%, maybe even upwards of 20%-25%, depending on the speed at which our trucks will travel. If you don't have hours of service limitations, you don't have to take those breaks, there's the potential for trucking companies to decide to maybe drive a bit slower and take advantage of better fuel efficiency that comes from that.
Everyone's aware of very high insurance costs in the, in the trucking industry, and the Aurora Driver will be safer. And also, when there are accidents, there'll be much more data for fault attribution, and we believe over time, the insurance industry will recognize that, and the cost of insurance will come down significantly. In the longer term, there are all sort of, you know, direct benefits that would happen in the short term. In the longer term, we believe that the technology will really transform the way that freight moves, because it's gonna bring previously impossible 24/7 truck utilization and faster freight than ever before. If you take a trip from Houston to L.A., that's ordinarily gonna take two to three days today.
That could be completed in 24 hours in an autonomous truck, which means you could reach the entire U.S. within a day with only two or three distribution centers. A much different proposition to what you see today. Then, of course, very important to the trucking carriers, our potential customers, is because the trucks can drive 24/7, they have the potential to kind of double their revenue per truck, which we know is an important measure for them. It's a super exciting technology with a massive potential impact. Next slide, please. To meet the needs of our customers, and most effectively monetize our technology, the business model that we're going to introduce is called Driver as a Service, and that's structured to be highly capital efficient. We don't want to own trucks. We don't want to operate trucks in the long term.
We want to do what we can do best in the world, which is to provide the Aurora Driver, the software and the hardware. What this means from a business model perspective under Driver as a Service, is that we anticipate to get to very attractive SaaS-like margin structures. The way that we expect to kind of get our revenue is we expect to provide our technology to fleet owners and operators, and then, in return, we'll be paid a fee per mile for the driving task. This model, you know, is very attractive to us from an economic perspective, but it also aligns very well with our customers' needs. Next slide. We believe we're very well positioned for success in trucking. We're successfully executing on our roadmap today. You can see that in our investor presentations.
It's on our IR website, and we've outlined our path to commercialization. At the same time, what's happening in the industry is that the competitive landscape is clearing, and you might have seen recent news that a key competitor is de-emphasizing their autonomous trucking side of the business. Another player is likely winding down, and another player has quite a bit of uncertainty with respect to their path forward. We believe that we're the leader in the AV trucking space, and that we're best positioned to commercialize this technology to generate significant value in the freight ecosystem and for our shareholders. 2023 is turning out to be a pivotal year for Aurora as we continue to execute around our roadmap to launch Aurora Horizon, our autonomous trucking service, which we expect to do by the end of 2024. Next slide.
If we drill into our roadmap, you can see what we're working on, what we've done, and what we expect to do on the path to commercialization. You'll see it split up into kind of three components of the business that we think are needed to bring our autonomous trucking service, Aurora Horizon, to market. The top section is the Aurora Driver. That's our self-driving technology that I just walked you through. In the middle, we have the truck platform section, and for the truck platform, we're working with our OEM partners, PACCAR and Volvo Trucks, and that really helps us to operate at scale in future.
We think it's super important to have deep relationships with OEMs in order for them to provide a redundant truck platform, so it's got the redundant brakes and the redundant steering that you need to have a safe product on the road. The last section, we call it commercialization, and what that section does, that part of our roadmap, is showing what we're doing in respect of our customer relationships, our operations, and what infrastructure is necessary to effectively integrate Aurora Horizon into the existing freight ecosystem. Something that sort of Aurora's been keen on and has been sort of focused on through its life is we're not doing a science experiment here. We're really trying to build a business and build a massive business for the future.
We think it's very important, as well as making progress on the technology side, as well as having a solid truck platform, that we also understand that this has to be a commercial product. It has to be useful and usable with our potential customers. That's why there's that focus on the commercial operations section, so that we don't get to the point where the truck can drive itself and then try to figure out what to do with it. That's all sort of embedded into the development work. I'd like to call out that in Q1, at the end of Q1, in the earnings call, we announced that we'd achieved a critical Feature Complete milestone.
What this means is that we've implemented all the capabilities that we believe are necessary to operate on our Dallas to Houston launch lane, and we've removed all of the policy interventions for that lane. We think this milestone is an inflection point on our path to commercial launch, as it unlocks the final phase of refinement and validation to close the Safety Case for our launch. We plan to spend the rest of the year working on this to achieve our next critical milestone, which is slated for the end of the year, which is Aurora Driver Ready. That's the next major milestone on the path to commercial launch, which we expect to be by the end of 2024. We've been a leader in transparency in the AV industry through our life.
We were the first company to create and publish a Safety Case framework, and we've taken that a step further, given feedback from investors, about how to kind of track our progress. What we did is we created, an Autonomy Readiness Measure, the ARM, and that's enabling the investment community and all of our stakeholders to track our progress we're making against closing our Safety Case. Often people ask us: "Well, what's the bar to launch? How do you know when this is gonna be launched? How do you know when it's safe enough to launch?" To us, the closing of our Safety Case, that's when we know that we're ready to launch. The ARM, the Autonomy Readiness Measure that we're publishing, tracks that.
With our first quarter results, we shared that the Autonomy Readiness Measure, the ARM, was already at 44% as of March 31st, 2023. We've done nearly half the work in closing the Safety Case. That metrics demonstrates to the strong progress that we're making towards closing our Safety Case, and this shows that we're still on track to achieve Aurora Driver Ready later this year. Next slide. Just before we shift to Q&A, I want to spend some time highlighting our philosophy and strategy around bringing our self-driving technology to market, because, like I said, this isn't a science experiment. This is a real business that we're trying to build. We have spent the last six-plus years building the technology to capture this tremendous opportunity.
Like I said, we really just want to focus on what we do best, which is the self-driving technology itself. We want to go around, and we want to partner with industry leaders to support commercialization of the Aurora Driver. This slide sort of outlines some of the partnerships we have, and you can see that it's an awesome slate of people who are really experts in their field. If you kind of started with a blank sheet of paper and had to fill in these boxes, you would be pretty happy if you got to this position. On the OEM side, we're partnered with two of the top three North American trucking OEMs in PACCAR and Volvo Trucks. They together make up nearly 50% of the market.
We're currently out on the roads in Texas, pulling loads for customers in autonomy every day under the supervision of vehicle operators for now, for people like FedEx, Werner, Schneider, and Uber Freight. You might have seen that we recently announced a partnership, a pilot partnership with Hirschbach as well. That's another example of how we're trying to build a business. We're getting experience with different trailer types. In this case, Hirschbach's an expert in refrigerated goods, and that's a kind of use case that could very much benefit from autonomous technology because of the hours of service limitations don't exist. We also have a couple of collaboration partnerships with U.S. Xpress and Covenant.
On the on-site fleet maintenance side, we have a pilot program with Ryder, that supports our pilot operations, we also expect that ultimately that will support our commercial operations at scale. Super exciting from my perspective, I, you know, I was able to work on this deal. We announced an exciting first-of-its-kind partnership with Continental, who, as you know, are one of the world's leading technology manufacturers and Tier 1 automotive suppliers. Under this partnership, Continental will co-develop and manufacture a future generation of the Aurora Driver hardware kit, they'll manage the complete life cycle for that. From design through manufacturing, service, and warranty. Continental has already begun investing what may ultimately be $300 million plus in NRE to scale the Aurora Driver hardware.
That allows us as Aurora to avoid near-term capital commitments to develop and manufacture our hardware. We really believe that this is a testament to the Continental's confidence, both in the autonomous future, but also Aurora's leadership position in that future because they're putting their own money to work now for something that will pay off into the future. What's super unique, very unique about this partnership, is that it's a Hardware as a Service structure. This hasn't been done before, but what it means is that Aurora will pay for the hardware on a per-mile basis as we get in revenue in the future, as the hardware is used on the road in our product. This aligns with and it supports our highly capital-efficient SaaS business model that I just touched on before.
Financially, we believe industrializing our hardware kit through this partnership is gonna help us achieve the commercial scale and the cost structure necessary to support our long-term profitability objectives. that's all in the freight world, and, you know, super awesome partnerships in the freight world. When we're ready to target the ride-hailing market, we also have very strong partnerships on that side with both Toyota and Uber. With that, I think we can kick the slides out, and, Chris, I look forward to taking your questions and any that the audience might have.
Richard, thanks so much. You know, while you were speaking, we have an internal chat, and all we kept on saying is, "I wish every company would do something that looks like your slide nine" I appreciate the visibility and the clarity into the process for, you know, particularly for autonomy, which, you know, sometimes can be opaque to the outside investment community. If it's okay, I would love to jump into those, you know, at least think about the pieces as we move into commercialization in 2024. If we focus on that slide nine, the next six months before we get the Driver Ready, it looks like it's increased validation.
Yep.
Then I would love to. Maybe you could talk about what is being done to validate the hardware. Also just a little bit more on the metric that you referenced for the Safety Case, being, you know, obviously the last piece to get to Driver Ready. If we could just dive into that a little bit more, because sometimes, again, we don't realize how long it takes once we even feel Feature Complete...
Yeah
to make sure that this, you know, it's ready to be on the road for X number of years.
Yeah, for sure. No, thanks for the compliment. Yeah, we, it's kind of interesting, right? When you're in a world in the sort of development phase with super exciting technology, you don't have revenue yet, and it's
Yeah
you know, that's what people are used to seeing. We worked, really worked hard internally in terms of trying to create something that people could follow along with, because we wanna build that credibility, right? We wanna say, "This is what we're gonna do. This is how we're gonna get there," and then, you know, every quarter be able to tick the boxes that we've done it. Appreciate you calling that out. Yeah, I think if I touch on the ARM, the metric first.
Yeah
Then we can talk about sort of Feature Complete through to the Aurora Driver Ready milestone. The ARM was, again, people were asking us since we went public, like, "Oh, how can we tell what your progress is? When are you gonna launch? How do we know?" It's like hard for them to see on the outside. We live it every day, so we can see the progress. I get to, you know, drive in a truck, ride in a truck down to Houston, and I can see the progress time over time. What we felt is, you know, there's various metrics that are put out into the world, something like RITM is one that.
Mm-hmm
often people try to kinda push out. We think that, it isn't great as a metric, and it isn't great as a measure of progress towards commercialization for a number of reasons. One is, you know, what we think is an intervention might not be the same.
Mm-hmm
Company X or Company Y, so it's not really comparable. Also, most importantly, you know, safety is the first part of our mission. This has to be a safe product, and if you introduce a metric that is which essentially, you know, could cause your vehicle operators, who are there for the safe operation of the vehicle when they're supervising it in autonomy, to not pull it out of autonomy when they're concerned because they're trying to hit a metric, we just think that's a terrible idea. We, we looked at it and we said, "Look, like, what's the launch bar to launching a commercial product in 2024?" What that launch bar really is the Safety Case is complete. The U.S. is a very permissive from a regulation perspective.
Mm-hmm
... very different to Europe, as I'm sure you know. Europe's, like, more of a type certification thing, where, yes, you do have to go and get permission for what you wanna do from the various bodies. In the U.S., I think, 46 of the 50 states today, if we had a self-driving truck, that we as a company were comfortable with, we could put it on the road, and it could operate without a vehicle operator. Without a driver.
We wouldn't have to get permission from anybody, even, you know, sometimes you go to Texas and they're like, "Why isn't this a product already?" You're like, "Oh, okay, we've got a Safety Case to complete." That's the kind of, the sort of regulatory environment that we live in in the U.S., which is kind of awesome, but at the same time, it requires the company to make that call as to what the bar for launch is.
Mm-hmm.
The bar for launch is the closed Safety Case. We put out this structured safety argument that kind of nests down, and you can see it on our website, and it kinda nests down to all these contentions that we have to have evidence for before we can say that we've completed the Safety Case, and thus, we're happy to have the product on the road. What the ARM does is it tries to track that progress-
Mm-hmm
... along the way. Currently, we're at 44%. By the end of the year, at Aurora Driver Ready, we'd expect to be at 95%. We'd say the Safety Case is 95% complete. Before we launch, it has to get to 100%. The reason that at the end of the year, at the Aurora Driver Ready milestone, we'd expect to be 95%, is that at that time, we won't have the redundant truck platform that we think we need in order to launch the product on the roads. We expect that to come in 2024. Once that comes, there'll be another period of time whereby you're validating that platform and the Aurora Driver working on that platform. You'll close off the last 5%-ish of the Safety Case. You'll be at 100%. You can launch.
The work that's going on behind the scenes, that's measured by the ARM, that's all of the work that individuals are doing to prove, gather the evidence for the system working as intended from a safety perspective.
Maybe that leads into the second question, which, you know, hardware is obviously important, not only for 2023.
Mm-hmm
... before Driver Ready, also getting to full launch, obviously the cost down after. Could you talk a little bit about your sensor set? Who, you know, you're partnering with for external hardware. You also have, you know, internal capabilities through Lidar. Finally, the deal you have with Continental is something I, you know, I really don't think exists anywhere else. The fact that Continental would be doing multiple components is pretty unique, meaning some of those components, and that's why I want to clarify or just get to make sure I understand, won't even be theirs, but they'll be able to bring it all-...
Yeah
... as one one suite. if you could just, those three parts, I know it's a long question...
Sure
Who are you using, internal, external, and then Continental?
Yeah, I'll try. If I get off track, feel free to kinda bring me back to the focus. No, it's an awesome question, and, you know, it highlights as well, you know, that we really wanna focus on what we do best, right? That's the self-driving technology itself, and to the extent that we're able to partner with world-class companies throughout the ecosystem, that's core to our strategy and our philosophy, and that extends to the hardware side as well. We absolutely buy what we can, and we're certainly happy to use off-the-shelf components for a system, where they're available and they're performant. For example, today, we'll buy kind of off-the-shelf, so cameras, radar, you know, short- and mid-range Lidar.
The thing that is unique, the thing that we have, which is proprietary in our hardware kit, is the FMCW Lidar.
Mm-hmm.
That's Frequency Modulated Continuous Wave Lidar. We call it FirstLight. This is something that came out of an acquisition that the company made in the past of a company who'd been working on this technology. FMCW Lidar is, we think, much more performant and much better suited than traditional pulsed Lidar is for being able to drive at highway speeds. That's because it can see further, it can also tell you the velocity of the object that it sees. For example, we can see, I think, four football fields down the road, and we can tell if an object on the side of the road is moving towards us, accelerating away from us.
Yeah
or stopped. This just gives a fantastic advantage. It gives extra time for the system to figure out what to do, and it can be the difference between having to slam on your brakes.
Yep
... or making a nice, graceful turn to the left. You know, we also build that in-house right now. That's our technology. It's a competitive advantage. We don't think anyone else has it. We also currently build our computer in-house to match the requirements that we need for the system.
Mm-hmm.
We have a mix, right? Again, like, if somebody had a Lidar that was better than our Lidar and it was cheaper, we would absolutely figure out a way to use that. Right now, we don't think there's anything that is comparable to FirstLight, so that's why we have it in-house. The Continental partnership is exactly like you said, it's the first of its kind, it's very innovative, and it's awesome. It just underscores, again, the approach, the partnership, partnering with industry leaders so we can take advantage of their skills and their 100-plus years of expertise. Under the partnership, what's gonna happen is Continental's agreed to co-develop and manufacture the, a future generation of the hardware. They're gonna integrate the Aurora Driver hardware components.
They're gonna put them together into serviceable modules.
Mm-hmm
... that will be then supplied directly to the OEM partners. The cool thing for us is that Continental, under this Hardware as a Service model, they're gonna manage the entire life cycle.
Yeah.
The design, then the manufacturing, then the service and warranty once it goes on the trucks.
Does that include the manufacturing of the FirstLight, in addition to some of the compute? I mean, obviously they have their own radar, their own cameras.
Yeah.
Right, I guess the manufacturing of the Lidar seems to be, you know, the biggest task as a new business for them.
Yeah. I think they have their own.
Mm-hmm.
Like, they have Lidar manufacturing capabilities themselves.
Yep.
We have the, obviously the specifications for the FirstLight Lidar. Right now they're at small scale. We're putting them.
Okay
together ourselves. We expect over time, these become much more manufacturable.
Yeah.
Obviously, Continental can help that happen. We acquired a silicon photonics company, which will, you know, it can shrink some of the components that are in there now onto a chip. This all supports the, like, the future state of more scalable, cheaper hardware. You know, we've got the hardware engineers, we've got talented hardware engineers.
Sure
We're never gonna be experts in hardware manufacturing.
Yep
... or the cost optimization. We don't want to spend our money doing that, right?
Absolutely
... we'd better do it somewhere else, so.
it could be something like, it could be radar from Continental.
Yeah.
Okay, perfect.
Yeah. I mean, we would work closely with Continental, and we would and they know the specifications of the system. You know, as Hardware as a Service, they have the ability to decide what to do. The cool thing for us is if they can take a short-range Lidar or a camera or a radar off the shelf, and they can put it in the system, well, that's awesome for us because it's gonna be a cheap component because they've already, you know, they were already using it on something else.
Absolutely.
It's fantastic. Because it's like this Hardware as a Service model and we're gonna pay them per mile, the incentives in terms of kind of cost down, cost optimization, producibility and things are, everybody's aligned on it. It's really phenomenal.
If I put my CCU hat on, central compute, you know, what type of compute is sort of in that, you know, that standard hardware?
Yeah. you know, like I said, we build the computers ourselves now to get it to the right components. We use GPUs at the moment, kind of off-the-shelf things. Over time, you would expect there's, you know, if we scale like we would like to scale, you'd, there might be a world where we have our own silicon, but right now we're just using GPUs and putting those in the computer. Again, under the Continental deal, for the future generation, like, Continental will put that computer together for us.
Richard, this has been an incredibly interesting presentation. I want to talk a little bit about the early business plan and, you know, that relatively limited initial highway ODD.
Yep
... that makes this feel particularly attainable in the way that, you know, some other AV companies may not feel right now. With the launch plan, the initial target is to focus on these designated highway routes. Maybe for our generals in the audience, you could talk about why this is an easier rollout to begin with, why this is your initial focus, you know, what you're learning in this rollout that can be transferred toward a more true Level 5 unlimited ODD set of autonomous driving parameters.
Yeah, sure. Yeah, great question. I think the first thing to say is, you know, we're focused on Level 4 autonomy, and obviously L4 autonomy is focused on specific domains, whereas, like, Level 5 is, that's truly unlimited autonomy, right? That's like, take a truck, put it in Yellowstone, and have it drive to the Empire State Building. From Aurora's perspective, we just don't see that Level 5 becomes a reality in the near future, and we're not even, you know, convinced that it's actually necessary over time. You know, so we're an L4 company. That's what we're working on, and that means that we have kind of specific ODDs. You know, I think we would say as a company that, yes, specific highway-focused self-driving isn't necessarily easier than anything else....
you know, the frequency of accidents is probably lower than urban driving, but, like, the severity of those accidents, because of the speed that you're traveling at and the weight that you're carrying, that could be higher. We certainly believe focusing on highway-centric routes will definitely enable more rapid scaling versus what you see with the urban-focused ride-hailing applications, that's primarily because the highway system in the U.S. is pretty self-similar. A stretch of highway in Texas is gonna look very similar to a stretch of highway in Arizona, or California, or even up in Pennsylvania. you know, we're focused on that highway domain because we really think there's a significant need there, and there's really attractive unit economics in the freight market.
Those kind of long-haul routes, that's the sort of the sweet spot to introduce a product into the market from a self-driving perspective. Start to earn revenue, you know.
Mm-hmm
... start to generate profits, and have that expansion and that scale, fund, you know, ride-hailing into the future.
And-
The reason that we can do this, and the reason that we're so confident about this being the right path for us is also premised on the fact that we have the FirstLight Lidar, the FMCW technology. That foundational investment in that technology, that's the thing that really unlocked being able to drive at highway speeds. Other companies, if their Lidar can't see far enough, one of the reasons they're probably doing urban driving predominantly is because they just don't get confident that they can see far enough down the road to have vehicles travelling at highway speeds. That's, that's it. You know, the next.
If you dig another layer down, what we expect to do when we launch is what we call a T-to-T model or terminal-to-terminal model. The customer's trailer comes, is dropped off by a human driver at an Aurora terminal. We hook our tractor up, powered by the Aurora Driver, which then autonomously hauls the load down the freeway. Goes down the surface streets, goes down the freeway, gets off at the other end onto the surface streets, and then stops at one of our terminals at the other end of the lane. At which point it gets unhooked, and the human driver takes it to its final destination. That's kind of the, we think, the most easiest way to get this product launched and the cleanest way to introduce it into the freight ecosystem.
You know, we then have to go back to the fact that we have the Aurora Driver, which is agnostic between vehicle types. When the, as cars get better at doing surface streets, the trucks get better at doing surface streets, we can see it moving into denser and denser urban environments over time. When we're ready to launch a ride-hailing product, we can come from a different part of the market than other people. We can do those high-speed airport runs, right? Airport to convention center, airport to hotel districts that other people might want to avoid. That's because our technology is gonna draft off each other.
Richard, because this always comes up in the practicality of the launch. If, again, if I had my Jonathan Chappell, this, you know, it's fantastic because it goes across so many sectors. The terminal-
Yeah
... will that be situated within a customer's depot? I mean, you know, even in just the initial Dallas and Houston, where are those actual terminals? Are those with your launch customers? Then you would sort of, you know, it'd be customer by customer, you'd have to create more Aurora terminals.
Yeah. At first, we have four of them today, we have one in Dallas, one in Houston, one in Fort Worth, and one in El Paso. These early terminals, we believe that they'll be Aurora sort of owned and operated to start off with. you know, that terminal-to-terminal model in the early days will be Aurora, like, Aurora-owned, and that allows us to control it. What we really believe is, you're adding this brand-new technology to the world, you know, the customer wants to make sure that it works in the wild. You know, we wanna make sure it works in the wild. We don't wanna kind of interrupt what's happening in their depots, for example.
That's why we have this terminal-to-terminal model, and that's also why we expect to own and operate the first, you know, 20 trucks in the launch fleet that we expect, and then maybe, you know, 100+, the next 100+, before we transition fully into the Driver as a Service business model.
Mm-hmm.
We really believe in, you know, introducing this in a sensible way that doesn't kind of cause problems, and then that will really allow us to scale faster because everybody will be comfortable with the technology as it transitions to their ownership and their operation. We absolutely want to develop the capabilities to go what we would call end-to-end, which is from customer depot at one end to customer depot at the other end. There's no, you know, there's no technological reason why that can't happen. There may be some operational reasons why a customer wouldn't want an autonomous vehicle to interact with its human fleet in the early stages. Yeah, I think both the customers and ourselves are incentivized to get to that end-to-end model as quickly as we can.
Which means that Aurora doesn't have to own and operate that set of terminals.
Perfect. Just one more from me, and then, I wanna get to the questions online for the last couple of minutes. On this path to commercialization, everything goes great in 2023, we're now into 2024.
Yep.
Can you just remind everyone how that business model of the subscription service works? Just some, some numbers. You know, is that model going into effect with the launch customers, or do they need to, in 2024, get a little bit of a commercialization test, and then you get paid at a different scale as the business ramps over time?
Yeah, for sure. Right now, we have pilots, and you can see that on our roadmap, that we've ramped up the number of pilots that we're operating with some of the, some of the potential customers, with the FedExes, with the Werners, with the Schneiders. We're driving those in autonomy with a customer, with a vehicle operator supervising the vehicle. We get paid today for those legs, like any other shipper would. We don't recognize that as revenue, 'cause we don't believe that's our commercial revenue, so we just offset it in the P&L as contra R&D expense. It's, you know, it's not trivial, but it's not kind of like, it's not enough to make a business out of. When we, when we launch commercially, we expect to have a launch fleet of around 20 trucks.
Those trucks will be owned and operated by Aurora, we'll pull loads on behalf of the launch customers. We'll get paid as Aurora, the full, you know, the full freight, whatever they would pay another shipper to do that. Out of that, we'll pay for the hardware, we'll pay for the insurance, we'll pay for the truck at that point, the fuel, et cetera, et cetera, et cetera. We expect to be paid per mile, the Aurora Driver drives a mile, we get paid per mile. We expect the sort of the amount of money that we would charge, we'll be competing against human drivers for, you know, the foreseeable future because of how this technology will be introduced into the world.
we think our pricing will be tagged against, what a human driver gets paid.
Mm-hmm.
We also think that there's a ton of value that's gonna be come. We talked about some of the reasons why this is really gonna address the pain points of potential customers in the freight market. We believe there's a lot of value that can accrue from those benefits to the customer as well. Therefore, it makes sense that we would get paid around what a human driver gets paid. As you transition into the Driver as a Service model, some of the costs kind of fall away, and what we're left with is we get that revenue per mile on a subscription basis. We then have to pay for the hardware, which again, will come down in cost over time because it's, as it scales, it's just...
There's no unobtainium in the hardware, it's just you need to make more of it, and that will come down the regular cost curves. We'll have revenue, we'll have hardware to pay for, we will have insurance to pay for, which again, expensive at first because the insurance companies aren't used to it. Will come down significantly over time, we expect, as we prove that we're safer. You know, some connectivity costs. The truck drives itself based on the computer on the truck, but it does have to be kind of connected to base for monitoring and things like that.
Mm.
There's some trivial costs on that front. And then there's kind of, you know, you call them kind of like, remote assistance, rescue-.
Teleops. Yeah.
Yeah, that kind of thing. You know, if a truck, if a tire blows...
Yep
... the truck needs to be rescued, things like that.
Perfect. Richard, there's a couple questions. Your excitement over FirstLight is I guess ringing true. I'm gonna combine a couple of Lidar questions.
Okay.
Over the course of the last two days, you know, we had a several Lidar providers. Aeva.
Yep
... FMCW, similar to your technology, which, you know, I think essentially came from, you know, from Blackmore and the acquisition. We had Luminar, which does 1550, and we had Innoviz, right? Who does 905. With that in mind, one of the questions is, for your FMCW Lidar, is it also cost-effective, cost comparative for some of these other Lidars, which also have relatively long range? You mentioned obviously FMCW ideal for the velocity of a vector.
Yeah.
We heard that from Aeva. We also know Mobileye. I think some of the most competitive Lidars out there that were thrown out over the last, let's call it, you know, two days, was something like $500-$800 for, you know, 905. I think that's the lowest number we heard from Innoviz. FMCW can be cheap because the parts are available, but we just don't scale.
Right.
A long way of saying, is this still a pretty cost-effective way to go to market, given the technology seems extremely attractive for all the reasons that you highlighted?
Yeah. I mean, I think that the technology is the right technology to kind of get comfortable that you can operate a Class 8 truck on the highway. That's kind of the first thing. Now, relative to some of those other kind of numbers-
Mm-hmm
... for some of that more traditional, you know, easier, like-
Yep
... Lidar, obviously we sit at a much more expensive price point than that. We see the line of sight for that cost coming down. The Continental deal now kind of locks us in. If you took our original model when we went public, there was a line in there for hardware costs. There's some sense. That some sense is now what's locked in with FMCW technology.
Yep
... with the Continental deal. Sure, it's gonna be expensive at first, it's gonna be more expensive, you know.
Yep
... paying than some of those things, but we genuinely believe that it's the right technology to have.
Yeah
... it'll come down in price. Like I said, if somebody came up with something that was significantly, you know, cheaper.
Right
... and was as performant as our system was, then, you know, we'd certainly look at that. Right now, we get the benefit of not having any other customers other than ourselves. You know, we've got a hardware problem or a software problem.
Yep
we can solve that rather than waiting for the next generation of something that a X number of other customers-
Yeah
... might be asking for something different.
In terms of the BOM, also, the way I imagine is you sort of now it's locked in with Continental.
Mm-hmm.
Continental's job is to deliver that.
Yeah
... and I'm sure there's some form of price downs with units. Basically, Continental has the role of a traditional Tier 1.
Yep
... to deliver that cost down, which, you know, essentially puts one.
Yep
off your plate.
Yeah, with the advantage that we only pay them per mile.
Yep
... we get revenue per mile. it's doubly even better for us.
It's variable. Fantastic. I wanna, you know, we've been talking about this, you know, ad nauseam for two days. We all obviously, it's kind of fitting, you know, NVIDIA is up 25%-30% on sort of the topic today. Everything Generative AI, ChatGPT, Predictive AI.
Yep
... models, we use deep learning in Aurora, right?
Yep.
I, if we had Chris on-
Yep
... you know, he would put on a master class about teaching us, I'm sure, about all machine learning. How much from your seat do you see it being implemented, even in the back end, on processes to reduce development times, to reduce cost? You just any way that this last six months where things seem to be moving, you know, so rapidly, you know, how is that playing out at Aurora?
Yeah, sure. Yeah, yeah, I think you're right. You know, yeah, these Large Language Models and things have been getting tremendous attention over the past, you know, six months or whatever. At Aurora, you know, the six months hasn't sort of changed, you know, what we've been doing, what our development has looked like. We've been leveraging those kind of advanced machine learning capabilities for years. One of the sort of the reasons why Chris and the other founders started the company six years ago, was that they were confident that, you know, there was a path to developing autonomous vehicles that people were going down. You know, brute force, drive a lot of miles, that's the answer. Lots of data, that's the answer.
They realized that, like, with those revolutions in machine learning that were happening even, you know, those six years ago, that that was actually gonna be an unlock to achieve the necessary breakthrough to commercialize self-driving technology. All of that kinda, like, the AI stuff, the machine learning stuff, that's all embedded in our development and has been from the start. That insight, you know, it helped to drive and set our clean sheet approach to building the Aurora Driver, which leveraged those enhancements, which we see pay off now as we move towards commercialization next year.
You know, as we enter this final stage of refining the Aurora Driver's performance. Then we absolutely expect these key technologies to continue to enable our pace towards the next key milestone, getting Aurora Driver Ready at the end of the year and then commercial launch. You know, the more people use these things, and the more the AI kind of develops and gets better, like, absolutely we'll take advantage of that, and, you know, we'll see what the future holds.
Exciting times. A great way to end on it. Richard, thank you so much. Really appreciate. For those on the line, if you want to learn more about Aurora, happy to put you in contact with the company. Stacy, thank you so much for helping us organize this as well. Also, feel free if you wanna learn about some of the concepts and the industry structure that Richard discussed, I'm happy to send over, you know, our general thoughts on autonomous and the potential upside for more autonomous trucking.
Yeah.
With that, Richard, we really appreciate it. We look forward to the progress, honestly, over the next six months as we go to Driver Ready. It really looks like it's gonna be an exciting 18 months at Aurora.
Yeah. No, it's awesome. It's like, as Chris says, he's been in this, like, 10 years, right? Or well, probably 20 years, you know. Crazy amount of time, but he's like, "You know, regardless of what happens in the press, whatever the sentiment is, like, we're closer than ever," right?
All the time.
You see Cruise driving, Waymo driving. We're gonna get this truck and product launched hopefully end of next year, and we'll be great. Thanks very much for having us. Really appreciate it, and thanks, everyone, for tuning in.
Thank you. Thanks, everyone, for joining.