TuSimple Holdings Inc (TSP) 2021 Q2 法說會逐字稿

Welcome to the TuSimple Second Quarter 2021 Earnings Conference Call. My name is Sylvia, and I'll be operator for today's call. (Operator Instructions) Please note that this conference is being recorded.

I will now turn the call over to James Mann. Mr. Mann, you may begin.

Thank you, Sylvia. Good afternoon, everyone, and welcome to our second quarter 2021 earnings call. With us today are TuSimple's President and Chief Executive Officer, Cheng Lu; and Chief Financial Officer, Pat Dillon. Cheng and Pat will review the operating and financial highlights, and then we'll take questions.

Before I turn the call over to Cheng, let me cover a few housekeeping items. As a reminder, TuSimple's shareholder letter and a replay of this call will be available later today on the Investor Relations page of our website. The call is being recorded. If you object in any way, please disconnect now.

Note that TuSimple's shareholder letter, press release and this call all contain forward-looking statements that are subject to risks and uncertainties. These forward-looking statements are only predictions and may differ materially from actual future events or results due to a variety of factors.

We will also discuss non-GAAP financial measures, which are not prepared in accordance with generally accepted accounting principles. Please refer to the safe harbor disclaimer and non-GAAP financial measures presented in our shareholder letter for more details, including a reconciliation of the non-GAAP measures to the comparable GAAP measures.

I will now turn the call over to Cheng to begin. Cheng?

Thank you, James. Hello, everyone. Welcome to our second quarter earnings call. Our mission at TuSimple is to bring the safest, most reliable and low-cost autonomous freight capacity to the market. To achieve this goal, we have a focused, 4-pronged strategy, which includes: one, the development of the most advanced autonomous driving system for trucks; two, first to market with production of purpose-built autonomous trucks at scale with our OEM and Tier 1 partners; three, the expansion of our go-to-market strategy, the Autonomous Freight Network; and lastly, investment in our people and operations. I will discuss our progress across these areas and then turn the call over to Pat to discuss our financial results.

Overall, the second quarter represented an impressive acceleration across all areas of our business. The capital we raised and the recognition we received from the IPO further elevated TuSimple's position as a leader in autonomous trucking. It helped to derisk our commercial execution and accelerated our ability to hire world-class talent.

In terms of people, we continue to invest heavily in building our team of exceptional technical and operational experts to complement our existing team. We hired Robert Rossi to lead our proprietary HD maps ecosystem and further build out that team during the quarter. We've already seen early returns from Robert's leadership as unique map miles increased by 3,500 miles in the quarter for a total of 8,500 miles at the end of June. HD maps are the core infrastructure that enables autonomous freight operations.

In July, we hired Adrian Thompson as our Vice President of Systems Engineering. Adrian brings extensive experience from his prior roles in the aerospace and autonomous driving industries. His systems engineering team is responsible for developing our technical capabilities, tools and processes to ensure the performance and reliability of autonomous driving system. This is critical as we integrate our ADS with purpose-built production vehicles alongside our OEM partners. Adrian's team also played a key part in the development process as we continue to invest in supply chain, including critical Tier 1 components such as braking, steering and onboard compute units.

The supply chain for Level 4 autonomous driving is still far from mature, and we believe our investments and partnerships further enable us to be first to market.

Overall, our R&D team continues to represent approximately 80% of our workforce. This quarter, we added 188 new R&D personnel to bring the R&D team's head count to just over 1,000 employees as of June 30. We're pleased with the quality of talent that's choosing to join TuSimple, and we expect to continue to hire at a similar pace for the foreseeable future.

Moving to the development of our autonomous driving system. We're confident we have the most advanced virtual driver in the industry. In the second quarter, our fleet of 70 autonomous retrofitted vehicles drove over 900,000 miles. Total road miles driven now stand at 4.6 million, which is a 26% quarter-over-quarter growth. Our road miles, conducted in fully autonomous mode and data collection mode via human drivers, provide valuable data to refine our autonomous driving system.

In terms of our IP portfolio, we awarded 38 new patents in the second quarter, a 14% increase from last quarter, for a total of 318 patents granted at the quarter end. Our newly awarded patents include technology advancements in rear-facing perception systems and automatic weighing sensor systems that further distinguish TuSimple's ability to automate our trucks at scale.

As an update on our Driver-Out pilot program, we continue to make significant progress during the second quarter. We're on the third phase of our 4-phase plan, which includes the upfit and autonomy reliability acceptance of our 10 new pilot test trucks, continued to refine of our risk management systems and the launch of TuSimple Connect, our proprietary transportation management and oversight system, swiftly designed to manage autonomous trucks. We have increased the frequency of our test runs on the pilot lane, which is a hub-to-hub route from Tucson to Phoenix, covering close to 100 miles. We feel very confident in the performance of the autonomous driving system. We aim to move to the fourth phase of our development plan in the coming months.

The last phase is the verification of the safety case and final acceptance. Safety comes before all else at TuSimple. And this is particularly important for the Driver-Out pilot program. The safety case verification process is multifaceted to mitigate the risk inherent in the system. And this includes significant work in systems engineering, social safety reviews as well as internal audits of our process. It includes proper testing, verification and validation of various software and hardware modules. Given the complexity of an autonomous driving system and this mission, we have to clearly understand what systems are on the truck, that the truck performs the way it's intended to perform at all times and that our operational plan is designed to execute the pilot program safely.

In addition, it includes validation that our risk management system will identify any issues and signs of degradation from both software and hardware and ensure that the system reacts appropriately, including pulling over or stopping lanes safely, what we call enter into a Minimal Risk Condition, or MRC.

And lastly, of course, the safety case includes the final on-road validation of the pilot with safety drivers on board. As you can see, we have a holistic safety case verification process to ensure that we have correctly identified and mitigated risk. Furthermore, this process is critical in advancing our autonomous driving truck production programs and us being first to market. Our target timing for start of the Driver-Out missions is still end of the year. As we move into the fall, we'll continue to provide updates on our safety case verification progress and the overall pilot program status.

We believe -- we are confident that our Driver-Out pilot program will be the first of its kind in the industry. We believe our time line is several years ahead of any competitors and is certainly an indication of the advancement and maturity of our technology.

Further highlighting the importance of safety. We recently kicked off a safety performance study with Geotab, a leader in vehicle tracking and fleet telematics. This study will analyze the comparative safety performance of our Level 4 fully autonomous trucks versus manually driven trucks. The data will be sourced in various conditions to measure industry standard indicators for unsafe driving that can increase the risk of accidents. We expect the study to add to our growing list of data illustrating TuSimple's autonomous driving technology benefits.

Now moving on to the next pillar of our strategy, which is being first to bring reliable, purpose-built autonomous trucks to market at scale. Our production program with Navistar continues to advance. During the quarter, we made progress toward our supplier down select process. Downselect is an important milestone as we move from research and development towards setting the final specifications for components with Navistar required for series production, where we narrow down the list of suppliers for key components over the next several months, especially those with long lead time items, such as braking and steering systems as well as LiDAR and onboard compute units. To our knowledge, we're the only autonomous trucking company approaching this critical milestone.

In addition to the work we're doing with Navistar, our partnership with TRATON continues to progress well. Pinnacle teams from TuSimple and Scania, one of TRATON's sub-brands, are expanding the scope of our concept programs. We expect to continue to increase our development activities in Europe, and we'll be investing in expanding our team there as well.

In China, our team continues to focus on testing on the Donghai Bridge and the Yangshan deepwater port near Shanghai. As we shared with you in the past, this is a 20-mile bridge that connects the mainland to the world's busiest deepwater container port. We believe this project has the most near-term commercial viability in China.

The last pillar of our strategy is our go-to-market for the development of our Autonomous Freight Network. The coverage and density and efficiency of the AFN, we believe, is TuSimple's long-term competitive advantage. The size and coverage of the AFN comes from high-definition map routes and strategic terminal locations. Terminals are the starting and ending point of autonomous operations. Our technology can leverage terminals from our customers, strategic partners and ones that we lease ourselves, thus enabling more coverage, better access to autonomous freight capacity while being very capital efficient.

We announced our Alliance Texas terminal opening in June to support our new freight lanes, providing us prime access to Texas triangle that runs between Dallas, Houston, San Antonio and Austin. The new Texas facility is located near a premier intermodal hub and will serve as a central logistics terminal to the Dallas-Fort Worth area.

In July, we entered into an agreement with Ryder Systems (sic) [Ryder System] that will celebrate our AFN growth. Ryder will supplement our existing network with dedicated terminal space for autonomous trucks, service and maintenance, roadside assistance and integration with Ryder's existing transportation solutions. There's also strong potential for Ryder's leasing and asset management services to better support our customers' needs. The relationship will offer Ryder new and expanding revenue opportunities, including its ability to leverage its financial resources to underwrite vehicle leases for our customers in the years to come. We're excited by the potential of our collaboration with Ryder, and we'll continue with our strategy to expand the AFN ecosystem with best-in-class partners.

Lastly, before handing off to Pat, I want to provide a brief update on the CFIUS review status. In terms of timing, our shareholder, Sina Corp, and TuSimple, submitted a voluntary joint filing to the committee, and that filing is currently in the 45-day review phase. It is not uncommon for these reviews to move into the second phase, particularly for companies engaged in new technology areas like artificial intelligence.

An additional phase, if that was incurred, will not be a definite signal of the ultimate outcome of the process. We were recently informed by CFIUS that the transaction being reviewed is the 2017 acquisition of U.S. assets of TuSimple LLC by TuSimple Cayman Limited. Additional information on the CFIUS process as well as some factual information regarding the 2017 transaction and our ownership is included in the 8-K. At this stage, we're fully participating in the process with CFIUS and sharing information, but we cannot predict the outcome of the review at this time.

We pride ourselves in being transparent and a level of vetting had gone through as part of our private capital raises, and our traditional IPO speaks to that. We will continue to provide record updates, and we look forward to the resolution of this process.

With that, I'll now hand over the call to Pat to discuss our financial results and guidance.

Thank you, Cheng. I'll start with our reservation program. We ended the quarter with 6,775 reservations, an increase of 1,000 reservations quarter-over-quarter. We also started our Voice of the Customer Program for our existing reservation holders. This program allows our reservation holders to engage in 2-way communication with us and Navistar on critical vehicle specifications and other operating requirements. This is extremely valuable feedback as we make design decisions and set our bill of materials for our production vehicle.

Also, we have now started more fulsome engagement regarding potential reservations with the next group of customers beyond our first wave. We are encouraged by the responses we have received to date, including a verbal agreement for a 100-truck reservation from a significant Tier 1 logistics operator, both domestically and globally. We are also in discussions with this customer regarding hauling freight beginning in the fall.

Now shifting gears to our financial results for the second quarter. We reported $1.5 million of revenue in the quarter, which is a 5x increase versus the same period last year and a 57% increase quarter-over-quarter. While we have maintained a flat number of TuSimple-owned trucks in our fleet, we continue to increase asset utilization and add to our AFN partner fleet capacity. We do expect to expand our fleet once we receive new trucks from Navistar in the second half of the year.

In the quarter, we spent $76 million on total R&D, including $25 million of stock-based compensation. This compares to $22 million of total R&D expenses in the same period last year. We spent $42 million on G&A during the period, including $27 million of stock-based compensation. This compares to $5 million of total G&A spending in the same period last year.

Our stock-based compensation expense for the quarter was $53 million, which includes $43 million related to employee equity awards that had previously satisfied time-based vesting requirements but had an ideal liquidity condition for full vesting. Upon successful completion of our IPO in April, the instruments were vested triggering the $43 million charge.

This quarter, we also began disclosing adjusted EBITDA, which we believe is a helpful business performance metric that is complementary to our GAAP results. This quarter, our adjusted EBITDA was negative $66 million. This compares to the same period last year of negative $26 million. The increase reflects our accelerating investments in technology and commercial development. We invested $5 million in capital expenditures during the quarter, driven by equipment purchases and facility costs. We ended the quarter with a cash balance of $1.5 billion.

Now to provide an update to our 2021 guidance for the full year. As we move towards our Driver-Out pilot and full commercialization, we have identified key areas of incremental investment to accelerate and derisk technology and commercial development, and we have updated our guidance accordingly.

First, as Cheng mentioned, we are making selected strategic investments to accelerate and enhance our high-definition maps as well as to build out our systems engineering efforts. The leaders of these 2 groups, Robert Rossi and Adrian Thompson, are world-class technical experts and experienced managers of large teams.

Secondly, we have also identified selected areas within our core algorithm, hardware and software teams that will allow us to increase our vertical integration of the autonomous driving system. These core groups are led by some of our most senior executives, including our Co-Founder and CTO, and the incremental investment will enhance our flexibility to work with multiple OEMs as we are doing today with Navistar and Scania.

Based on these items, we are increasing our R&D expense guidance to $200 million to $220 million, excluding stock-based compensation. We are also increasing our guidance for G&A expenses to $50 million to $60 million, excluding stock-based compensation. The increase is driven by incremental investment in critical functions, such as IT and HR, which scale with the broader business.

Beginning this quarter, we are introducing guidance for stock-based compensation in the range of $130 million to $150 million for the full year. We are also introducing guidance for adjusted EBITDA in the range of negative $260 million to $280 million for the year.

We are increasing our capital expenditure guidance to $14 million to $18 million for the year. This increase is comprised, in large part, by 2 items: first, our decision to purchase 25 Navistar trucks rather than lease; and two, incremental investments in new terminal facility build-outs as we expand our AFN ahead of schedule.

There is no change to our previously communicated revenue guidance of $5 million to $7 million for the year. We expect to end 2021 with a cash position in excess of $1.25 billion for the year.

I'll now hand it back to Cheng for a few last remarks.

Thank you, Pat. To summarize, we have a clear and focused strategy to be the first to bring the most reliable autonomous freight capacity to the market. We continue to prove our leadership capabilities in technology, hardware partnerships and go to market. Since 2018, we have demonstrated our capabilities to operate on surface streets, on-ramps and highways on Level 4 autonomy, and we continue to make significant advancements. We believe we're years ahead of our autonomous driving system development, and we intend to extend our leadership position.

Our hardware partnerships with Navistar and TRATON are checking major milestones to deliver our scalable purpose-built solutions. We're expanding our network of terminals in autonomous-enabled lanes to support our current and future operations. We know we still have a lot more to accomplish, and we work very hard at it every day. And we certainly can't do it without our amazing employees and business partners globally. So a big thank you to them.

With that, we're ready to start our Q&A session.

(Operator Instructions) Our first question comes from Ravi Shanker from Morgan Stanley.

Thanks for the detail on the Driver-Out test. A couple of questions there. It sounds like you still have a few Is to dot and Ts to cross on the technology side and the prep side to get that done. Are you confident? And kind of do you have at a full line of sight that you will get that done before the end of the year?

And second, are you simultaneously working on the operational things you need to do, which is the -- either the permission from the police, sort of state authorities to get that done, so when you're finished with the technology side, you will be ready to execute the test?

Ravi, that's right, yes. So we are still targeting end of the year. We're very confident in our technology. As I mentioned in the prepared remarks, when we think about what we're doing now, it's akin to effectively like launching a rocket, a satellite or a new aerospace prototype. There's certainly a lot of steps we had to take in order to identify and mitigate the risk in the system. And so we're working towards that. And in terms of operational side of things, yes, we are in a constant dialogue with the Arizona DOT as well as Department of Public Safety. And we do not believe that to be a real block.

Great. And maybe as a follow-up question, Pat, kind of on the higher R&D costs. It sounds like you're bringing stuff forward rather than seeing a higher level of cost relative to what you envisioned during the IPO. Is that accurate? And kind of maybe if you can give us like 1 or 2 catalysts for what exactly is driving that, kind of what you are bringing forward. And kind of does that mean you kind of get to your destination quicker? Or kind of just how do we think about that over the next 6, 12 months?

Yes. Ravi, I think it's accurate to say that, to some degree, it's an acceleration of costs. Certainly, the -- some of the key teams where we're building out, like maps and systems engineering, are areas that we always intended to build out. I would categorize us as being opportunistic as we've had the opportunity to bring on top talent, particularly at the leadership level, and build out the teams beyond them.

So to the broader question, we don't see this as an indicator of accelerating our expanded costs over the next 2 or 3 years as we approach full commercialization, but rather an opportunity to bring forward some of these team members. The net result being that, I think, both an acceleration of our technology, but also a derisking because we have more time prior to launch to build out some of these critical functions. In some areas that are acceleration, you've obviously seen it in our ability to build out our unique map miles to expand to the East Coast, which some of our customers have been asking us to do, so we're already seeing the fruits of some of these investments.

And the next question comes from Chris Wetherbee from Citigroup.

It's [James] on for Chris. Just wanted to actually ask about the hires you're making and basically what is the geographic location of them. Are there any sort of cost advantages if they're not actually in the U.S.? And just sort of getting your thinking about like essentially where the head count sits in the development sort of geographically.

Chris (sic) [James], geographically, the vast majority of our hires are in the U.S. I would say, close to probably 80%. And we also are having teams building out in Europe and then expansion of our technical teams in China as well, but primarily in the U.S. Now with COVID and what we learned from the pandemic, I think we are, of course, more open to having remote work. So that has been helpful in terms of attracting talent. At the same time, our recruiting team, actually the size has expanded. So all these things also contribute to the fact that we're adding on quite a bit of talent this year.

Got it. And those numbers you gave, I think they were for the company broadly. Is it true for the increment as well, with the 200 you plan on bringing on roughly?

That's right. That's right. Yes.

Our next question comes from Brian Ossenbeck from JPMorgan.

I wanted to ask about the Ryder partnership. Can you give a little more color around that? It sounds like there's a bunch of opportunities for just kind of like the nuts and bolts with maintenance in fueling. Will you be able to launch cargo from these areas? And maybe you can just give a sense as to what the possibility is to maybe lower CapEx, if you can scale with that sort of network? And on the finance side, is this going to function similar to what Penske could possibly do as well?

Brian, I hope you're well. And yes, as mentioned in the prepared remarks, our strategy is to expand the AFN ecosystem with best-in-class partners. And honestly, as it relates to maintenance, to leasing, to a network of real estate, to offering customized logistics solutions, Ryder is certainly among the best. And so to answer your question, really, it covers all the nuts and bolts that, as we think about launching autonomous freight operations from point A, point B, right? So it's a lot more than just the technology side of things.

This really will be beneficial to our capital expenditures. Our goal, our strategy, we believe, given that our technology has ability to drive both on surface streets and highways, is we can connect to more existing real estate locations. It doesn't make sense for us to spend money on all of them. We -- our goal is to provide customers, in the long run, the most reliable and easy access to autonomous freight capacity at low cost. And so certainly, this is a partnership that has a lot of potential.

We are working with Ryder, the collaboration to identify places that we can use as terminals. And we're also having a collaboration to go through the maintenance and other parts of the business or the services.

Okay. Great. As a follow-up, can you just maybe talk more -- a little bit more about the regulatory environment? I know in the back of your slide deck here, you talked about getting a couple of more states on board for full commercial testing. Are there any restrictions with those in terms of size or timing?

And then one more for -- just for, I guess, normal testing or availability. So can you give us an update on the regulatory side with adding a few more states and what else you might see developing in that area for the rest of the year?

Thanks. Yes, I mean, as our shareholder letter pointed out, 2 more states have had a positive regulatory -- regulation to allow for Level 4 autonomous operations of commercial vehicles. So that brings to a total of 26 states. And just a reminder, there are no states currently today that have any regulation that prohibits the testing of autonomous commercial vehicles.

I think in general, we continue to work closely with the regulators, both on the federal and different states. We continue to be very transparent with them. I think the regulatory environment is positive today. And a lot of that is regarding, of course, the safety benefits that autonomy can bring as well as -- and I think you're very familiar. I mean the driver shortage where the tightness in the supply chain today is probably one of the all times based on kind of what we're hearing and seeing. So I think these drivers will continue. And hopefully, that has a continued positive impact to regulation.

Next question comes from Ken Hoexter from Bank of America.

Cheng, Plus, I guess, just recently did an autonomous Driver-Out test. It seems to be pressing ahead and noted plans to be operational maybe a year earlier in their commercial deployment. Is there a difference from your perspective in the tech or capabilities of what some others are pressing ahead? Just want to understand the competitive differences of what we see out in the market.

Okay. Yes, I mean, we saw kind of what you saw this morning or recently. It's a short video, and there's some media reports. [RSN] is actually in some of the media reports already came out of China about a month or 2 ago for the same thing. RSN is also -- the media report confirmed that it was at a closed highway, the highway hasn't opened to public yet. So I think -- I mean it's clear, I think this is a very different type of test, if you will, from our Driver-Out pilot program, which is on open highways, on commercial operations, on long stretches of road, both on surface streets and highways in the U.S. So I think that's really not an apples-to-apples comparison, at least from what we can see.

And I think in terms of commercialization, I'm not quite sure -- I mean, my understanding of Plus is that they're talking about basically a Level 2 solution, right? It's not a Driver-Out solution. So it's a retrofitted kit that can make -- basically -- they can make the driver assistance. So as far as, -- again, from what we can tell, it's playing actually a different category. So I think -- I don't know if that's helpful, but that's how we -- effectively, that's what they're saying.

No, perfect. And then for my follow-up. You talked about the R&D. You talked about kind of where you're focused on some of the capital. You also mentioned, I think, somebody threw in there that there's a project that [Zaudy] is working on. Is there something specific that you're scaling on the R&D that we should look for?

I mean, it's all part of our overall development of our autonomous driving system. It's, on one hand, it's, of course, the system, the software itself, the virtual driver. As our production process with Navistar gets into more of the later gates, there's a lot more work that gets kicked off and more engineering work streams have to come together. So there's more resources that require significantly more resources.

We talked about continued investment in the supply chain. That is something that we'll continue to invest more capital. And that's something we talked about at IPO, too. With more capital, there are investments and partnerships we can do that can make sure that the Tier 1 components are on the same time line as our production vehicle. And that includes steering, braking the compute units, autonomous domain controller and of course, some of the sensors.

So I mean those are the uses that we have. But I wouldn't -- I mean I wouldn't point out to 1 specific project per se at this point.

Perfect. And just to clarify, if I can. Pat, you said there were no more orders, right, in the quarter, than the 1,000 were booked last quarter. But you said that there are more that are coming going forward, right? I just want to clarify what you had said.

Yes, Ken. So our reservations, we added 1,000 in the quarter, which is what we had actually previously disclosed on the first quarter call. And then we have had a verbal indication for another 100 trucks, but the reservation agreement has not been signed for that yet. But we are optimistic with the customer that made the verbal indication as well as several others. It's progressing nicely. So we're encouraged by the early outreach to this more expanded group of potential customers.

Great. Congrats on the progress.

On that part, I think, as Pat noted, I mean, we measure progress certainly on the number of reservations, but also on the depth of the collaboration. And I think that's maybe hopefully, it came out in terms of conducting 3-way Voice of the Customers. We have multiple sessions with majority of the reservation customers.

And this is something that we have said all along, is to -- it's really to get our large strategic customers involved in the development and in adoption of autonomy. And given that the reality is everyone's autonomous system will be a little bit different, right, assuming that people can also build this. We need people to understand how to use autonomous in the supply chain, how to use TuSimple's autonomous driving systems. So there's a lot of work behind the scenes, and we're actually very happy about the level of collaboration that we're seeing across the existing customers we have.

Our next question with Brian Ossenbeck from JPMorgan.

Cheng, you mentioned, I think in the last call, about the supply chain issues, and clearly, they're quite disruptive, as we all know, right now. You mentioned getting some trucks in the back half of the year for Navistar. Can you just update us on how many of those would be? And just what you're seeing from supply chain and sourcing some of your critical components and if that's affected development or assembly yet?

Yes. Just related to the Navistar trucks purchases. We are -- we have an order for 25 new trucks. I think the original time line is for all to be delivered in the July time frame. I think right now, we're expecting to receive our first 1 -- first 2, I think, sometime next week or in the coming weeks. And we're hoping to receive the rest before the year ends.

So again, as you know, supply chain is really affecting all the OEMs and the equipment makers. I mean, Navistar is giving us special, really great, great level of service with regard to this, but it's only behind schedule.

In addition to sort of sourcing of critical components, we're also seeing some lack of resources in terms of a supplier being able to, I guess, assist with the debugging. And actually -- so to answer Ravi's question about it. One, the timing risk for our Driver-Out pilot program is on the supply chain. But we do have, as an example, right now, we're dealing with 1 key component where a portion of the components, a section of the booked component are not giving the same level of performance reliability that we thought it should have. And we had to have some time to work this out with the supplier. We now have engineers there and they have engineers here, but that took a little bit of time. And so we hope to work through this problem. But so those are things that we're seeing.

But look, I know everyone is kind of putting all the problems in supply chain. So we don't want to do that for sure, but it is a real problem that we're facing across the upfit, across even just basic equipment that we need to procure as a company.

Okay. And then just going back to what I think you said earlier about pulling forward some work to maybe work with OEMs -- more OEMs in the future. Obviously, that's a selling point of one of your competitors, and it's still a fragmented truck market and fleets have their preference. So I was assuming you're going that route anyway. But maybe you can just talk about what you're pulling forward and why? And if this was part of the road map to begin with?

Sure. I mean I think what we're seeing is more on helping the supply chain mature as well as having a more general platform. I think the way we work with the OEM still has to be very close-knit collaborative process just because, as you see in the trucks, I mean, everybody has very different placement of sensors, compute requirements, power requirements. And of course, these trucks have to be integrated, certified by the OEM because that's what the customers care about, right? Who provides a warranty? Who provides the aftermarket support? And these trucks have to drive a million miles. Those things don't change.

And so what we're saying is there are additional investments, partnerships we have on the supply chain that can move forward the industry. And so I would say contrasts very differently with saying we have an aftermarket solution that can fit in all OEMs. I think that idea, maybe that idea -- the reality is it's not practical because in an industry like this, where the supply chain, the technology is still in the sort of developing stages, there is no common protocols, right?

Today, for instance, a customer can ask for a [Cummins] engine or they ask for maybe a proprietary engine, but both engines, wherever they choose, the OEM certifies that truck. But that wouldn't be a case today if you put in aftermarket autonomous driving kit. And so until that industry becomes very mature, we don't see that being a viable solution anytime soon.

Our next question comes from Colin Rusch from Oppenheimer.

This is [Brendan] on for Colin. First one, can you guys just delineate in the quarter how many of the miles were conducted in Europe?

Sure. We actually don't count the miles that's in Europe. That is actually -- because the trucks there are actually owned by Scania. So -- and so we're not disclosing the miles in Europe given there's some obviously sensitive NDAs involved, but it's a decent amount. So that makes sense. So we only disclose the miles that are run by trucks owned by TuSimple.

Great. And then just as a follow-up. Would you mind just giving us any update in terms of qualifying sensors, sensor functionality and then vehicle design ahead of the upcoming -- or ahead of the 2024 launch?

I might -- and perhaps we can get back to you on that one. Again, I have to make sure that we're at a position to publicly talk about this information. But we are making some design -- definitely making a lot of design choices as we speak. But again, I will quickly come back to you afterwards to ensure that this is public information that we can disclose.

Our next question comes from Denis Pyatchanin from Needham & Company.

Denis on for Raji tonight. I had -- I want to ask you a question about your Driver-Out program. So I think you had mentioned that you're kind of on the third phase out of 4. Can you talk in a little bit more detail about kind of what you're ironing out right now? And what needs to be done before you guys are comfortable going Driver-Out for maybe that demo at the end of the year?

Sure. It will be a long answer. But I think to keep it -- try to be concise. So there is -- so we have a new -- I mean, from a hardware standpoint, we have built or designed and updated these redundant trucks or basically trucks with redundant sensors and compute units. These are important for being able to remove the driver, right? Because we don't have that last line of defense that is the driver. And we need more than just 1 because we have to be able to validate these trucks with enough miles in order to ensure does the safety work to mitigate the risk associated with it.

And so Phase 3, a large part is that. And this is not only just simply to sort of put the mechanical parts onto these trucks, but also to ensure the acceptance of the autonomous performance of these trucks. And so these are retrofits. So there are some visibilities or things we have to tweak or iron out.

We also mentioned, for instance, we have components because we're working with some Tier 1 prototype components, as we speak today. And across 10 components, 5 of them might work 1 way and then 5 might not have the same performance that we need. So these are all the sort of the nuts and bolts that we have to work out this phase.

In the last phase, again, I mean, if you think about launching, let's say, aerospace prototype, right, this is not about -- this is not a PR. It's not a marketing event. This is -- we wanted this to be very successful. We want it to be very safe. We need to understand the risk associated with the thing, not simply by just running a bunch of miles, but having a more of a deep dive on all the checklist around the safety case verification.

And I kind of talked about in our -- in my prepared remarks, there's a few things to oversimplify. But one is you built this very complex system, we have to know what's in the system, right? And from a functional safety, from a engineering standpoint. Two is we have to be sure, to be very sure or highly confident that the truck performs and these functions perform the way it's meant to perform at all times, right? And the third thing is we have to design and validate the operational environment or the operations to safely conduct this test. And lastly, we have a validation process with actual driving, but still with the safety driver behind the wheel.

So I'm not sure I -- no sure that I provided you enough color, but it's a complex engineering problem. We are, as mentioned, very confident with the functionality of our autonomous driving system. If -- I think for those that have been to Tucson, I think that's evident. But to make it safe and really mitigate the risk and add reliability, I mean, that's where you need to really take out the driver at scale.

Got it. And then as a quick follow-up, I think you might have alluded to this. Is the chip shortage currently impacting kind of your ability to procure the trucks from Navistar? I mean I think you might have mentioned some kind of debugging delays or something like that. Could you just speak a little bit about that, please?

Yes. Yes, there are 2 different things, but the chip shortage is impacting our ability to procure more trucks from Navistar on time.

And the next question comes from Ben Kallo from Baird.

First of all, can you just talk about the Geotab relationship a bit? And is it basically allowing you to compare your technology and show the benefits of it? Is that how you do it? And then go deeper into that, is that for regulators or who is that -- or customers or both? Maybe just some more detail there.

Yes, great question. It's for both. I mean it's for everyone that's involved in the ecosystem. So with the telematics device on the truck, the steel tabs and the data that they have versus the data they have from just manual truck driving, like, of course, they have a lot of data there, we're testing for basically harsh driving events. And I think the industry definition of harsh driving events are harsh acceleration, harsh braking and harsh cornering. So basically kind of where the g-force kind of hits the truck. It's given the height of the truck, the center of gravity, right, it's easy for a truck to roll. So those kind of 3 things are considered harsh driving events. And it's pretty common that these events do impact the frequency of accidents. And so that would be where we look at the data in conjunction with Geotab.

So the early -- the preliminary data that we've seen is very promising. And this goes in line with our independent study, a year-long study with UCSD on fuel savings of the TuSimple autonomous driving system. Empirical data, we have our partners, we've done a tire study. So I mean, the biggest thing about autonomy, of course, is the safety, the efficiency gains, but we want to have empirical data to highlight and quantify the benefits that our autonomous driving system can provide, and this is a part of that.

Got it. And then just on CFIUS, I know you can't comment too much. But the 45 days, when does that end? If you can give us some information on that. And then, Pat, it looks like there's about $100 million of cash burn in the second half. Is that the right number?

Sure. On CFIUS, while we submitted the review, the filing in June, so we can't -- we're not giving the exact date, but that's kind of the timing. I think just kind of relating to CFIUS, obviously, I think it's a sensitive word, but we continue to be very transparent. And I think we try to lay out the facts in the 8-K. And I think we look forward to resolution in due time.

Ben, it's Pat. On the cash burn question, I think it's somewhere between $150 million to $200 million of cash burn on the year, just looking at some of the -- our cost of goods sold, our R&D, our G&A and our CapEx, just if you interpolate between what we've achieved for the first half and our guidance for the year. And that puts us comfortably above the target that we have, which is $1.25 billion. So we're not providing anything more specific than that, but I think if we can go through in more detail in a smaller session, if that's helpful.

And our next question comes from Alex Potter from Piper Sandler.

First question, can appreciate maybe you can't talk to specifically around sensors and suppliers, brakes, things of this nature. But when do you think you would be, I guess, maybe estimating that the bill of materials for the production truck will be more or less set in stone?

Yes. We are targeting by early next year. And there's, of course, different stages, but that's the target that we have alongside Navistar as part of the production program.

Okay. Perfect. And then regarding the order book, I also can appreciate that you're being very selective, looking for deep relationships to -- in order to sign up a fleet, and that demand outstrips your capacity to serve that demand. To what extent do you think you have capacity to entertain incremental orders? Are you basically -- my historical impression was that the order book wasn't necessarily closed, but it was more or less closed because you're trying to focus on those really, really high-quality customers. But do you anticipate adding many more customers and many more trucks to the order book between now and 2024?

Yes. I think it's a good question. I do think it will be a little bit of a stair-step approach. In the initial period for the rest of this year, we are really focused on the highest quality of the reservation book. So it is around those folks that have the seriousness of intent to be able to dedicate resources because it really is a mutual dedication of resources to be able to put an L4 technology into our carrier-owned capacity customers' networks. So from that perspective, we do expect to add additional customers, but we will be selective. And it's important to note that our standards remain the same. There is a financial commitment involved with each one of the reservation partners. As we've indicated before, it's either a reservation that goes into escrow and cash or for those that have made a pre-IPO investment in the company. So I would say, without being too specific, that we do expect to expand it, but we're certainly being selective about how we do it.

As we move closer to the start of production in 2024, then I would expect to see that this book -- folks that are placing reservations and eventually firm orders will start to expand. And that's only natural. We do see this as an extraordinary vote of confidence from our customers that we don't take lightly for them to make a financial commitment to engage with us this far ahead of the actual delivery of the trucks. So we try and take that very seriously and be very selective to make sure that we can provide the level of service and engagement that's required to have a great relationship with them.

Our next question comes from Joseph Spak from RBC Capital Markets.

Just a first question. As you expand the map, and I don't even mean what you're doing now, I just mean sort of, over time, it expands to a lot of different landscapes and geographies and environments, is -- does the ODD differ by region? Or is there some like minimum standard that has to be met so that these trucks can operate fairly seamlessly as they travel across the country?

Joe, good question. I mean, certainly, we'll start with a minimal viable product ODD. For us, that's a hub-to-hub operations. There will be some [geo-fence] around that. We don't think we'll be able to have, for instance, snow in the first [MVPs] because the winter testing that's required will add more time, another year, 1.5 years to the development process and it's better to put it into the next generation. And of course, these things can be updated OTA.

So I think if you kind of go down 1 level deeper, I mean, there are even more specific things in the ODD that maybe some areas have and some do not. Generally speaking, though, I think our experiences on a highway, the driving in the United States is quite similar, very different than, say, urban city areas. If you compare L.A. and New York, it's a very different sort of behaviors between the drivers. So that's kind of what we see. I'm not sure that was sort of what you're asking.

Yes. No, that's perfect. I appreciate that. The second question is just, again, somewhat of a, I guess, a sensitive topic, but I think one that comes up with investors a lot, which is clearly a lot more just consternation in U.S.-China government relations. You obviously -- you're dealing with AI. You have plans for both -- to operate in both U.S. and China. There's R&D centers in both regions. There's been back and forth over export control. So how do investors get comfortable that one of the opportunities might not be hampered just by geopolitics?

No, good question. I mean look, as -- so any multinational technology company to have operations in these areas, it's certainly a dynamic time. We're very sensitive to all the new regulations that come out from both sides. And we've done a lot of things ahead of time to account for that, right? So the majority, as we talked about, the core IP innovation happens in the U.S. for us. China is our R&D center that's focused primarily on the China market as well as helping with the European market. So I think there is some -- there's actually quite a bit of bifurcation as we speak today. So I think we do have sort of internal processes in place as we see sort of a, call it, a worsening of export controls or geopolitics.

But I do think that -- how do we get investors comfortable? I mean, end of the day, the -- both markets are massive. I think our ability to honestly to just serve either one is a very big win. So I think this is more, I think, I'll say, offshore (inaudible). I mean, maybe that's not the right word to say, but certainly, I think we're playing the 2 biggest markets. And so far, we have been able to navigate. But of course, we're very sensitive to just the changing environment.

Our next question comes from Dan Levy from Credit Suisse.

First one, the color on the head count expansion is helpful. Obviously, we keep on hearing about this arms race for talent. Maybe you could just give us a sense of how your talent acquisition has fared versus others, how much more aggressive you can get. And I guess, maybe more importantly, how much of a constraint is talent in achieving the task of removing the driver and scaling on operations? Or I mean, maybe said differently, does accelerating head count help you to accelerate your path to commercialization?

Dan, I mean, it's yes, in the sense that there is -- I think there's a minimum number of head count. I mean, as we talked about, if you look at TuSimple, we're -- to get to commercial operations at scale, I mean, we're effectively a artificial intelligence company, we're a software company, we're an automotive company, and we're a logistics company. So to accomplish this, whether it's us or anybody else, I mean, there's a minimum headcount number that you can do with, let's say, 100 people, right? I mean this is -- will be in the several thousands, just given the complexity of this. And again, because of that, the barriers to entry in this market will only get higher and higher certainly over the next few years.

I don't think -- the acceleration is not linear. Certainly, the way we prioritize our time. I think with that, I think one of the things that we're most proud of and I think why we've been able to attract a lot of good talent, especially post-IPO, is that our employees understand that we have a very focused strategy. It's not only technology leadership, but we have a very clear commercial rollout. And even engineers, even folks that are deep, deep in tech, appreciate this point.

And so yes, there's no sort of right answer to your question, but I think you have to have certainly scale in terms of talent and focus.

Got it. And then second question. I want to talk to you -- just talk to your ecosystem of partnerships. And I know that none of these partnerships really have exclusivity per se. But is there any stickiness factor in your relationship, of all the commercial partners that you've listed out, that puts you in the front seat for deeper commercial agreements with them as you scale up or that makes it more difficult for them to switch to another tech provider?

Yes, I think there is sort of natural switching barriers because -- so for instance, our network, how we design the network for kind of the right on top of where our current customers are, that provides some switching barrier, right? Because for them to switch technology, you'd assume that the coverage of their maps and everything fits with their supply chain, right, with our customer supply chain. There's things like -- small things. I mean, right now, the Voice of Customer, kind of how the handoff works, how do you plug into a TMS system, what kind of equipment that's on the truck, I mean, all these things do play into -- and different fleets have different ways of operating, and they don't like to change if they can help it.

And then, of course, kind of starting today, be able to plug into carriers and shippers, TMS, transition management system, and the way they dispatch trucks, that -- most of the systems on the market today are quite customized. These are probably old systems that have been very customized carrier-to-carrier over time. And so it's actually a long integration process. So I mean those are the things that I think do have natural switching barriers.

But look, end of the day, we have to deliver. We have to deliver the most efficient, most accessible autonomous freight capacity and enable customers to use it. So I think the customers are with us today is because they see our leadership in technology and where we're going.

We have no further questions at this time. Thank you, ladies and gentlemen. This concludes today's conference. Thank you for participating. You may now disconnect.

Thank you, everyone.