管理層發言
Greetings and welcome to the Aurora Third Quarter 2024 Business Review Call. As a reminder, this conference is being recorded. It is now my pleasure to introduce your host, Stacy Feit, Vice President of Investor Relations. Thank you, Stacy. You may begin.
Thanks, Paul. Good afternoon, everyone and welcome to our third quarter 2024 business review call. We announced our results earlier this afternoon. Our shareholder letter and a presentation to accompany this call are available on our Investor Relations website at ir.aurora.tech. The shareholder letter was also furnished with our Form 8-K filed today with the SEC. On the call with me today are Chris Urmson, Co-Founder and CEO; and David Maday, CFO. Chris will provide an update on the progress we have made across the key pillars of our business and David will recap our third quarter financial results. We will then open the call to Q&A. A recording of this conference call will be available on our Investor Relations website at ir.aurora.tech shortly after this call has ended. I’d like to take this opportunity to remind you that during the call, we will be making forward-looking statements. This includes statements relating to the benefits of integrating artificial intelligence into our product, the safety benefits of our technology and product, the achievement of certain milestones around and realization of the potential benefits of the development, manufacturing, scaling and commercialization of the Aurora Driver and related services, including relationships and anticipated benefits with partners and customers and on the timeframe we expect or at all the market opportunity, the expected future market size, our product compatibility with those with partners and customers, our expected market share, the efficiency of our validation process, our remote assistance, efficiency for driverless operations and profitability of our products and services, regulatory tailwinds and framework in which we operate, expected cash runway and overall future prospects.
These statements are subject to known and unknown risks and uncertainties that could cause actual results to differ materially from those projected or implied during this call. In particular, those described in our risk factors included in our annual report on Form 10-K for the year ended December 31, 2023 filed with the SEC. Additional information will also be set forth in our quarterly report on Form 10-Q for the quarter ended September 30, 2024. You should not rely on our forward-looking statements as predictions of future events. All forward-looking statements that we make on this call are based on assumptions and beliefs as of the date hereof and Aurora disclaims any obligation to update any forward-looking statements, except as required by law. Our discussion today may include non-GAAP financial measures. These non-GAAP measures should be considered in addition to and not as a substitute for or in isolation from our GAAP results.
Information regarding our non-GAAP financial results, including a reconciliation of our historical GAAP to non-GAAP results maybe found in our shareholder letter, which was furnished with our Form 8-K filed today with the SEC and may also be found on our Investor Relations website. With that, I’ll now turn the call over to Chris.
Thanks, Stacy. Today, we stand on the brink of a new era in mobility and logistics, which will bring a safer, more efficient, and more accessible future for everyone. With commercial launch now within sight, we are closer than ever to unlocking the benefits of the Aurora Driver for our customers and the motoring public. We are excited that the broader industry shares our vision. Enthusiasm for our technology continues to build among many of the industry’s most respected carriers. This sentiment was particularly evident at our recent partner summit and is further underscored in our commercial contracting progress. With another launch customer signed, our expected launch capacity is now fully contracted, and we are in the final stages of contracting our remaining second half capacity to match our anticipated supply. Importantly, with the support of many of Wall Street’s largest institutional investors, we completed another successful capital raise in August, adding nearly $0.5 billion to our balance sheet.
This incremental capital extends our runway well into 2026, and we expect to fund the initial phases of our scaling strategy. As we press towards commercial launch, we continue to lead the industry with our commitment to safety and autonomy performance transparency. To begin driverless operations, we must close our safety case for the Dallas to Houston launch lane. Our safety case framework is a comprehensive evidence-based approach to confirming that our self-driving vehicles are acceptably safe to operate on public roads. We quantify our progress toward closing our Dallas to Houston launch lane safety case through the Autonomy Readiness Measure, or ARM, which is a weighted measure of completeness across all claims in the safety case for our launch lane. We remain the only company in the industry that has provided this level of transparency. As of the end of October, ARM was 97%. With our most recent software release, we have validated the majority of highway driving.
We are now primarily focused on final behavior refinement and validation for components of surface streets that we sequenced later in our work plan, some rare construction elements, and closing a small number of vehicle claims, specifically related to redundant systems. Validation of the Aurora Driver’s perception system is virtually complete, with over 400 modes validated and just 6 remaining. The perception capabilities are truly outstanding, not just in everyday conditions, but also in just about any condition the world throws at it. On Page 5 of our presentation, we have shared a collection of our challenging scenarios that, while rare, do occur. In one case, the Aurora Driver detects a motorcycle speeding down the road at an astonishing 150 miles per hour. In another, it detects someone lying under a car on the freeway from over 150 meters away. The Aurora Driver also skillfully tracks a work convoy blocking multiple lanes.
In one of the most unpredictable and dangerous scenarios, it detects a jaywalking pedestrian suddenly emerging from behind trees on a dark, rainy night when visibility on the highway is very limited. Each of these situations poses extreme danger, whether due to a small, high-speed actor, low profile individuals whom a human driver would likely not notice, in-motion work crews, or poor visibility and erratic pedestrian movement. Yet, in every case, the Aurora Driver consistently perceives the scenario. These perception capabilities enable the Aurora Driver to navigate the complex environments it encounters. In the video on Page 6 of the presentation, you can see the Aurora Driver’s superhuman perception capabilities on the road. The Aurora Driver is navigating heavy traffic on I-45 just outside of Houston while a Texas motorcycle 100 meters behind is illegally lane splitting through a narrow space between trucks.
The Aurora Driver quickly identifies the motorcycle as what we refer to in the self-driving industry as a vulnerable road user and adjusts to create as much space as possible. Unlike the human driver who would likely focus on the vehicle directly ahead and not anticipate this behavior from behind, the Aurora Driver has 360-degree awareness and can proactively respond to mitigate danger. Similarly, in the video on Page 7 of our presentation, we see the Aurora Driver traveling through Huntsville, Texas on a dark night and at speed from 230 meters away, detects a person attempting to run across the highway and immediately takes action, slowing down while changing lanes away from the pedestrian’s path to create additional safe space. Once it has safely passed, the Aurora Driver merges back into its preferred travel lane, successfully navigating this complex, very high-risk scenario with precision and care.
With further confidence in the Aurora Driver’s performance, we are transitioning to a single vehicle operator for some of our commercial loads as we approach driverless operations. This operating mode also provides the discipline to fully implement our remote assistance capabilities and supports our analysis for remote assistance efficiency for driverless operations. Based on our current modeling, we expect to be able to operate at least 10 trucks per remote assistance specialist by the end of 2025, which is a meaningful threshold that supports our path to achieving positive gross profit. We expect to continue increasing this ratio throughout 2026 and beyond. Another key metric we use to assess the Aurora Driver’s performance and commercial readiness is the autonomy performance indicator, or API. The indicator penalizes the use of onsite support, which will be the most expensive support to enable the Aurora Driver.
We are focused on driving up the percentage of commercial loads that do not require any form of onsite support, which we refer to as 100% API loads. As a reminder, we do not anticipate that aggregate API will ever reach 100%, even at launch, because certain situations will always require onsite support. However, we believe the percentage of 100% API loads is a strong indicator of our progress toward commercial launch, and we expect this metric to reach approximately 90% by that time. During the third quarter, 80% of the commercial loads on the Dallas to Houston launch lane had 100% API, which is a 5 percentage point increase from last quarter and consistent with the performance we saw in the June stable software release. To supplement our internal ARM and API metrics, during the third quarter, we invited leading authorities in commercial driver licensing and evaluation, J.J. Keller and Road Master, to assess the Aurora Driver’s proficiency.
They evaluated our system as they would assess a traditional truck driver and found that the Aurora Driver performed exceptionally well, meeting or surpassing the standards expected of a professional driver. We feel good about our progress and are confident in our ability to close the safety case for driverless operations on our launch lane. With additional visibility on the time needed to complete the aforementioned remaining validation, we now expect to launch commercially in April 2025. While this is modestly later than we had intended, this timing remains within the margin of error we have anticipated and conveyed throughout 2024. With our intention to introduce the Aurora Driver with a crawl, walk, run approach, this shift to our timeline will have a negligible financial impact and does not affect our scaling efforts on our path to self-funding. During launch, we expect to deploy up to 10 driverless trucks in commercial operations, starting with one driverless truck and transitioning the balance to driverless.
We are deliberately starting this way as our early efforts will be focused on exercising the full product suite to ensure a seamless product launch while demonstrating the value proposition for our customers and continuing to build trust with all of our stakeholders. In the second half of 2025, our focus will be on expanding our product capabilities, adding new lanes, and increasing capacity to tens of trucks by the end of that year. To ready our customers for driverless operations, in September, we hosted our annual partner summit where we were joined by more than 20 of the nation’s largest and most sophisticated carriers. In aggregate, these businesses operate well over 100,000 Class 8 trucks. We brought these customers together with industry safety experts, regulators, first responders, and law enforcement for collaborative conversations on driverless operations. We also announced and soft launched our partner success program in which customers have the opportunity to more deeply evaluate and assess the Aurora Driver’s performance as a final step to move forward with driverless operations.
The enthusiasm was palpable. Some of the participants on average rated the ride experience 4.7 out of 5 across a set of 20 criteria. While all eyes are first on the Dallas to Houston launch lane, our customers have also been keenly interested in when the Aurora Driver will begin to operate beyond Texas. We regularly gather feedback from many of the industry’s leading carriers regarding where the Aurora Driver can add the most value. At the summit, we announced that in 2025, we plan to extend our Fort Worth to El Paso Lane, on which we are autonomously hauling loads daily to Phoenix, which is one of our customers' most frequently requested lanes. The Fort Worth to Phoenix lane spans over 1,000 miles and takes at least 15 hours to complete, making it particularly compelling for autonomy, since the Aurora Driver isn’t subject to hours of service limitations and can operate nearly 24/7. We expect to begin commercial pilots for customers between Fort Worth and Phoenix in the first half of 2025 with the intent to go driverless on that route later in the year.
With the promise of how the Aurora Driver can benefit not just operations between Dallas and Houston, but also their network more broadly, Schneider recently executed their contract for 2025 volume and joins our commercial launch cohort. As we prepare for commercial launch, we continue to autonomously haul freight for all our pilot customers, including FedEx, Werner, Schneider, Hirschbach, Uber Freight, and others. We are scheduling nearly 160 commercial loads per week, or more than double the commercial volume we were executing a year ago. Cumulatively to date, we have autonomously delivered under the supervision of vehicle operators more than 8,200 loads, driving over 2.2 million commercial miles with nearly 100% on-time performance for our pilot customers. As our customers gain confidence through hands-on experience, we are also seeing growing optimism around autonomous technology from state governments, including a request from the California DMV for informal public input on draft regulatory language regarding the testing and deployment of autonomous trucks.
While preliminary, we view this as an encouraging step toward a process that would explicitly allow the deployment of self-driving trucks in the state. To demonstrate the safety benefits the Aurora Driver would bring to California’s roadways, which tragically see more than 400 fatal collisions involving trucks each year, we examined data from 14 fatal crashes on I-5 in Sacramento County that occurred between 2018 and 2022. We recreated these collisions in simulation to understand how the Aurora Driver would have acted if placed in these situations. Our analysis found that the Aurora Driver would not have caused any of the fatal collisions. We included one of these simulations on Page 14 of the presentation. In the original incident on I-5, a three-way collision occurred due to a wrong way driver on the highway. We created a variation of the original incident in simulation by placing the Aurora Driver in the position to avoid the wrong way driver while accounting for a passenger vehicle alongside it in the adjacent lane.
In the simulated scenario, a proprietary long-range FirstLight Lidar recognizes the wrong way driver far in advance, allowing the Aurora Driver to safely complete a contested lane change to move out of the way, preventing the collision. More than 5,000 fatal crashes involving large trucks occur on our nation’s highways annually. We applaud the California DMV for undertaking the robust stakeholder-driven process as it considers the development of a regulatory framework that would enable the safe deployment of autonomous trucks in California to fully realize the safety and economic benefits of this technology. We submitted public comments, which were due to the DMV earlier this month, and we look forward to continuing our efforts to help move these regulations forward. In the meantime, under existing law and regulation, autonomous trucks can be deployed today in the vast majority of states in the U.S. All of the work we are doing to launch driverless operations in Texas is supporting the development of our playbook for rapid lane expansion to capitalize on the significant autonomous trucking opportunity.
Last month, McKinsey & Company projected that the U.S. will have the fastest autonomous truck adoption rate globally, with autonomous heavy-duty trucks accounting for 13% of trucks on the road by 2035. Given the self-similarity of the U.S. Interstate Highway system and the power of our verifiable AI technology, we expect the Aurora Driver to capture considerable share of this opportunity. We anticipate the Aurora Driver’s capabilities to transfer across lanes with the opening of new lanes requiring limited development just for incremental features. In fact, our team recently completed bidirectional mapping of the El Paso to Phoenix route, 450 miles each way, in just two weeks. The Aurora Driver navigated a vast majority of the lane autonomously in its first round trip. Autonomy software capabilities transferred seamlessly to the Phoenix Lane as we anticipated, with vehicle operator interventions for the Inland Border Patrol station and some complex construction.
We also expect the efficiency of our validation process to support more expeditious safety case closure for future lanes as the subsequent turns of the crank will naturally be faster. These factors give us confidence in our plan to rapidly expand driverless operations to the Fort Worth to El Paso Lane and then further to Phoenix before unlocking additional lanes across the Sunbelt. To commercialize autonomous trucking across these lanes at scale, deep integration with OEMs on scalable autonomy level truck platforms is needed. We continue to make good progress with our OEM partners on our vehicle programs. During the third quarter, we integrated several new Aurora Driver equipped Volvo VNL autonomous trucks into our fleet, and they are now operating autonomously on the road alongside our Peterbilt 579 trucks. Our exclusive partnership with Continental is designed to support the scaling of the Aurora Driver hardware for high-volume line-side installation at our OEMs.
This month, Aurora and Continental reached another partnership milestone. We finalized the detailed system and component-level architecture and hardware selections for the scalable hardware as a service generation of the Aurora Driver. Our teams are preparing to start initial testing in the first half of 2025 as we continue to progress towards the start of production planned for 2027. At Aurora, we are driven by a mission to deliver the benefits of self-driving technology safely, quickly, and broadly. That mission has led us to a point where the Aurora Driver is on the cusp of making self-driving trucks the new standard for safety, efficiency, and sustainability in the logistics industry. Tremendous opportunity lies ahead, and we are working tirelessly to capitalize on it. With that, I’ll now pass it over to Dave who will review our financial results.
Thank you, Chris. Let’s discuss our financial results. We have provided a summary on Page 18 of the slide deck for reference. During the third quarter of 2024, we continued to demonstrate strong fiscal discipline. Third quarter 2024 operating expenses, including stock-based compensation, totaled $196 million. Excluding stock-based compensation, operating expenses totaled $161 million. Within operating expenses, our R&D expenses, excluding stock-based compensation totaled $139 million. This amount reflects $834,000 in pilot revenue, which is up 75% year-over-year and which we record as a contra R&D expense. After our commercial launch, pilot revenue will be recognized as standard revenue rather than as a contra R&D expense. SG&A expenses, excluding stock-based compensation, were $22 million. We used approximately $143 million in operating cash during the third quarter of 2024. Capital expenditures totaled $7 million.
This cash spend was below our externally communicated target, reflecting our continued commitment to fiscal prudence. For the fourth quarter of 2024, we expect cash use to be within the $175 million to $185 million quarterly average range. We expect our 2025 quarterly average cash use to be in this range as well. During the third quarter, we opportunistically raised $483 million in gross proceeds from a public offering of our Class A common stock. Net proceeds totaled $466 million. We ended the third quarter with a very strong balance sheet, including approximately $1.4 billion in cash and short-term and long-term investments. We expect this liquidity to support our planned commercial launch and fund our operations well into 2026. With that, we will now open the call to Q&A.
分析師問答
Thank you. Our first question is from George Gianarikas with Canaccord Genuity. Please proceed with your question.
Hi, good afternoon and thank you for taking my questions. If you don’t mind, could you please give us a little bit more detail on the remaining validation required to launch commercially in April of next year? And also, given that the timeline has been somewhat extended, can you help us understand your conviction that this is the last target and that it’s not going to move from April to another date after that in the future? Thank you very much.
Yes. Thank you, George. I appreciate the question. First, we are really very excited about the progress we have been making. If you look at some of the performance we are highlighting in this call, and obviously what we see regarding the quality of driving and the safety of what we are putting on the road, we are becoming more and more proud. My conviction regarding our ability to cross the finish line and get this launched is only increasing with every step. Internally, we had a recent milestone where we assessed the progress we have been making, and we concluded that just given the work that we had scheduled in front of us, we weren’t going to be able to finish that by the end of the year. Hence, we thought it prudent to share that as early as we could. This is that moment. When I look at the confidence we had regarding that date for the last 1.5 years or so, we have been discussing the error bars being months.
We see this modest delay as altogether consistent with that. At this point, we see the error bars in our estimate being weeks. So, we are very excited about where we stand here. In terms of the work that’s to be done, it’s primarily in the areas of some elements of surface street driving and some elements of construction that we encounter on the freeway. An important point to make here is that the API number that we shared, the 100% API number which is currently at 80% of trips without the need for driver intervention encompasses driving through construction and on the service streets between our terminals. So, in practice, our system handles these scenarios extremely well today. We just want to ensure that we have extremely high confidence in the system as we move forward and the work we had scheduled is just taking a little bit longer.
So, this validation sounds like a software and training issue. It’s nothing that has to do with your original equipment manufacturer partners or their commitment to the product or other partners' commitment to the launch state?
No, this is our internal execution. We continue to have a very strong working relationship with our OEM partners. The work we are doing with the vehicles is primarily focused on validating the redundant systems in the vehicles.
And as a follow-up, to the extent there are commercial loads you are committed to carrying in 2025 and now it looks like the timeline has been extended by a few months. Is there any impact financially? Are there any penalties you have to pay or any other issues with regards to those contracts? Thank you.
No, there aren’t. We will continue to build our load volume over this time. We will operate as we are today, in a pilot mode, with an operator onboard. We will have an increasing number of our operations operating with a single vehicle rather than two vehicle operators as we go forward as well.
Thank you. Our next question is from David Vernon with Bernstein. Please proceed with your question.
Hey. Chris and Dave, thanks for hosting the call today. I just wanted to ask you if you could help me understand the commentary around the technology, specifically regarding the adoption of Phoenix. I am just observing you guys working on the autonomous readiness measurement index and aiming to build expertise in the Dallas to Houston lane. Why wouldn’t there be additional time required to extend to new lanes as you progress, to ensure that the system would be ready and capable of adapting to that new operating environment?
Yes. We do anticipate there will be a small amount of work, but a core thesis that we have is that driving on a freeway, whether you are in Texas, California, or Montana, is basically the same. We believe this and are starting to see it empirically. That’s part of what we wanted to validate. We built out the lane, know that members of the investment community sometimes have questions about mapping at an efficient rate, but the answer is yes, we can. This is before we have optimized the mapping process, and what we observe is that the execution quality on that is consistent with our expectations. We have a strong relationship with the Border Patrol team, and the majority of our engagement is with the El Paso inland station. In maintaining the quality of that partnership, we disengaged while passing through the Border Patrol station between El Paso and Phoenix. We also mentioned that some work is required to finalize validation on construction zones, which presents complicated configurations we must navigate.
I am trying to think of the sample size as well. The number of hours in your denominator for your indices in Dallas to Houston is a lot higher than for Phoenix. So, as you consider that, what is the scalability from lane to lane? That’s what I’m trying to understand.
Yes. We expect it to be relatively rapid and to get faster over time. Part of how we have structured our validation involves what we see as three buckets: frequent occurrences, rare occurrences, and exceptional occurrences. That provides us with a granular view of the capabilities of the Aurora Driver. When we open a new lane, we evaluate whether new features or capabilities are required that do not align with our validation already completed. Generally, freeways look very similar nationwide, and thus, not much needs to be added. As we add new lanes, the second lane will require somewhat more work than the first, but much like a human driver, we expect the Aurora Driver to navigate most places without retraining after a few iterations.
Awesome. Thanks so much.
Thanks for the question.
Our next question is from Mark Delaney with Goldman Sachs. Please proceed with your question.
Good afternoon and thanks very much for taking my questions. First, I was just hoping you could share an update on operational readiness of the supply chain, including working with parts suppliers in areas like sensors and compute truck OEM partners. You mentioned Volvo and PACCAR, and then also some of your contract manufacturing partners.
Absolutely. We see no parts constraints or blockers in getting to commercial launch; we are in a good position there. We believe we have pre-purchased everything we need for those launch vehicles. We do not have concerns regarding that. We will continue to build relationships because we do not want to be sprinting and then have to wait before hitting the growth curve. You have heard us talk about engagement with Fibernet, which will allow us to bridge to thousands of vehicles, and our longer-term partnership with Continental that will help us reach automotive scales of tens of thousands plus. We are making good progress with both of those, and we have a strong partnership with Continental and a great relationship with Fibernet. We are already beginning to bring up hardware for the mid-generation, and we discussed the progress with Continental and the Hardware as a Service platform.
Thanks for that, Chris. My second question was around the contracting. You mentioned progress that occurred in the last quarter and doing well with the 2025 bookings in total. I believe the view from Investor Day was that pricing would be relatively in line with market rates. I hope to better understand if that is still materializing as you have gone further into the bookings process.
Hey Mark, it’s Dave. Thanks for the question. Without getting into specifics, because we don’t release the individual contracts, the market itself is down a little bit this year on pricing. We generally set up our pricing consistent with market rates, with adjustments needed for our launch partners and so on. I would say, generally speaking, we are in line with what we expected and what the market rates indicate. It’s really important for us to build a base of business and establish footholds with all our partners as we move forward. Overall, I think the market would like to see a bit more upward price pressure, and we will benefit from that as well.
Thanks Dave. One more financial question if I could please. You mentioned in your prepared remarks that once the company begins commercial operations, loads that are hauled even with a human driver in the truck will move from contra R&D and be considered revenue. Can you give us a sense of the run rate of what’s contra R&D at this point, so we can better understand this as we build 2025 and beyond estimates? Thanks.
Yes. This past quarter, we were a little under $1 million for contra R&D based on our pilot revenue. Essentially, every load that we carry for a customer, we bill out. You will continue to see that build up over time, so if you are doing modeling, it’s just building off of that.
Thank you. Our next question is from Jeff Osborne with TD Cowen. Please proceed with your question.
Great. Thank you. Just a couple quick ones on my side; I was wondering if I missed it, but did you quantify the launch capacity that you have contracted in the duration of the contracts that you signed?
No. What we talked about in the remarks is that we are going to begin with one truck, and then we are going to scale that up to 10. We have not shared the duration of the contracts, but they are in line with industry practices.
Just to add to that, the market generally contracts on an annual basis, and we tend to follow that practice in our contracts. We have been working with these partners for a long time. We have automatic renewals and set up. We will start up again with like, up to 10, and then at the end of 2025, we will be into the tens of trucks, right? So you will see us focus on execution in the first half of the year and then start scaling the business in the second half.
It makes sense. And just to be clear, were there any contracts signed before that you might have termination clauses or delays? Does the four-month delay lead to any hardship for the counterparty?
Not at all. As a matter of fact, I think George mentioned this as well. The last thing we want to do is incentivize getting something out before we get it right. All of our partners, our customers, share the same perception of what this can provide to us. The number one element is safety, and it’s much more important to ensure that we get it right than to attempt to meet an arbitrary date. None of our contracts have penalties, restrictions, or anything like that for not launching within a certain timeframe.
Thanks.
As you can see, I think that’s consistent with the way we think about the partnerships we are trying to build. We are looking for customers who see this as a strategic long-term opportunity and want to prepare themselves for this next generation of full logistics business.
Thanks, David. Maybe the last one for you, Chris, is just could you be more granular? I think when you have done the writing drives with investors and analysts, we have gone through construction zones. Could you just give us a few examples of edge cases that you are still overcoming as it relates to the interaction with construction zones?
Yes. It’s not strictly about edge cases. It’s about ensuring that we have validated across the variability of how construction sites can be configured. It turns out when you drive repeatedly between Dallas and Houston, the construction looks the same every day. So there is limited exposure to different ways that construction can close off lanes. It’s not something new or surprising. It’s just that it’s been in this particular place in the schedule of work, and it’s just taken us a bit longer to get through some of the stuff we had intended. I think it’s important to note that when we examine the actual on-road performance of the trucks, we actually haven’t had a disengagement caused by construction zones since May on the Dallas to Houston lane.
That’s good to hear. Thank you. That’s all I had.
Thank you for the questions.
Thank you. There are no further questions at this time. This does conclude today’s conference call. You may disconnect your lines at this time. Thank you for your participation.