WeRide Inc. (WRD) Earnings Call Transcript
November 20, 2026
Earnings Call Speaker Segments
Thanks. Hi, everybody, and thanks for being on WeRide's third quarter earnings call. We had a really strong quarter, coming in ahead of what the market expected, with higher revenue and a smaller net loss. Revenue rose 88% year-over-year to RMB 321.5 million, while gross margin climbed to 43%, mostly on the back of our high-margin overseas business. Versus the prior quarter, our net loss shrank 22.6% to RMB 310 million. So we're expanding quickly while also getting more efficient. Worth pointing out: Robotaxi revenue jumped 204% year-over-year, and 90% of that comes from overseas markets. As it stands, WeRide is Uber's biggest fully driverless Robotaxi partner outside the U.S. On top of that, revenue from our AI infrastructure segment almost doubled, as we ride surging demand for physical AI. Jinshuo, a WeRide subsidiary, now holds several billion RMB of orders under framework agreements. Operationally, our L4 fleet stood at roughly 4,100 vehicles, and our Robotaxi fleet has gone past 2,300 vehicles. In China, we hit positive unit economics per Robotaxi in Guangzhou. In the Middle East, fleet size hit 600 vehicles, a 50% increase. In Europe, Spain granted us its first national operating permit for L4 autonomous passenger vehicles, and we expect commercial operations to begin by year end; that gives us a benchmark for rolling out commercially across Europe. We remain on track toward deploying tens of thousands of Robotaxi in Europe before 2030. Before going deeper, I'd like to touch on a few topics investors care about most. First, losses and cash. Investors have asked whether growth is costing us rising losses. It isn't. Revenue was up 39% quarter-on-quarter, and gross profit rose 59%. Net loss shrank 22.6% quarter-on-quarter to RMB 310 million, ahead of market expectations; we closed the quarter with roughly RMB 4.7 billion in cash reserves, enough to fully fund our plan to reach positive quarterly free cash flow in 2028. Jennifer will take you through our cash position in detail. Second, the quality of our growth. More and more of our revenue base is recurring. Service revenue, which includes Robotaxi operating service fees, climbed 102% year-on-year and now makes up 58% of total revenue. That stream scales with every ride, carries high margins, and needs no vehicle ownership from us. Third, our international business. Overseas revenue increased 171.5% and now represents close to half of total revenue, plus more than 90% of Robotaxi revenue. Our Middle East fleet grew 50% in just one quarter. We started public Robotaxi services in Zurich, and Spain gave us its first national L4 operating permit. Permits take years to obtain and are among the hardest moats to replicate in this business. We work with multiple platforms and fleet partners, among them Uber, Grab, and several local platforms, which broadens our channels and makes growth more resilient. Fourth, on AI infrastructure. A few investors have wondered if Jinshuo, the subsidiary, takes focus away from what we are doing at L4. It does not. What Jinshuo does is turn a profit on the data and the tooling that we had already developed for our own autonomous driving operations. It is an asset-light data powerhouse with strong momentum in growth. Revenue for it almost doubled in the quarter, and it holds several billion RMB of orders under framework agreements. I will come back to each of those points as I go through the quarter. Let us now look at the overall results. We saw considerable growth in both revenue and gross margin. Revenue was up 38.8% quarter-on-quarter, while gross margin widened 5.5 percentage points sequentially. The main growth driver was overseas market. Our revenue there rose 171.5% year-on-year and made up close to half of total revenue. Robotaxi revenue for the quarter came in above RMB 100 million, an increase of 204% year-over-year. Even more important, high-margin overseas market represented more than 90% of that revenue. Inside the AI infrastructure segment, we have assembled a sizable library of high-quality, real-world data set for physical AI, extending our capabilities from autonomous vehicles into robotics. On the ADAS segment, that is our L2++/L3 business, cumulative deliveries passed 100,000 units by the end of the quarter. That means we reached our full-year target early; on top of that, we won 10 new mass-production design wins and established new partnerships with several leading OEMs. Looking ahead, we will push our physical AI strategy forward and speed up L4 fleet scaling and expansion. Let me now turn to our strategy. Investors have also inquired about our long-term strategy; they would like to grasp the strategic positioning of every business and how resources get allocated. I want to stress that WeRide has grown from an autonomous driving company into a physical AI company. WeRide now projects her capability and value in physical AI onto several application dimensions, among them the L4 dimension, that is Robotaxi, the L2/L3++ dimension, that is ADAS application, and the AI infrastructure dimension, that is model services. Jinshuo, our subsidiary, concentrates on real-world data collection, labeling, simulation, and related model services; it now covers multimodal data and complex scenarios; this supplies an underlying data and model engine for physical AI. In summary, Jinshuo is the projection of physical AI company WeRide in AI infrastructure dimension. Today, we take our capabilities and value in physical AI to market chiefly through our L4 business. When WeRide projects her capability and values in physical AI onto autonomous vehicle dimension, a Robotaxi fleet takes shape. Via our Robotaxi application, we have managed to bring foundation models, operating autonomous driving vehicles, data infrastructure, and commercial deployment together, creating a closed-loop evolving system that underpins the scaling and evolution of the physical AI capability of WeRide. When we project our physical AI capability onto mass-production passenger vehicle, we have our ADAS business line, which are growing at a speed of 10 times per year. Let me now go through the progress made across each of our business lines in the quarter. First, on L4 and Robotaxi. I will begin with our Robotaxi business line. We’ve hit a turning point in scaling Robotaxi, backed by positive UE in Guangzhou and the momentum from our overseas expansion. Robotaxi revenue for the quarter exceeded RMB 100 million. Service fees based on mileage now account for a larger portion of revenue, which is pushing up overall Robotaxi gross margin. In the Middle East, we sped up our asset-light expansion. Our fleet is at least around 600 vehicles, which is a 50% increase from the end of the prior quarter. That gives us a wider base for scaling commercial operations throughout the region. At the same time, we now cover key core areas of Abu Dhabi and Dubai, and we’re aiming for full core city coverage by the end of the year. Our fleet has expanded, our technological capabilities and operational efficiency have gotten better, and our service footprint has grown. As a result, average daily rides, mileage per ride, and operating hours all increased, which drove improvements in our unit economics. In this market, we work with an e-hailing platform and local taxi operators, including Uber and TXAI. We adapt our autonomous driving solutions to local markets while staying compliant with regional safety and traffic regulations. Partners that own assets have incentives to allocate more rides to us. Local fleet operators handle day-to-day tasks such as cleaning, which lets us focus on strengthening autonomous driving capabilities and operational efficiency.
Going forward, we anticipate that our Middle East fleet will almost double in the coming year, and we also anticipate attaining citywide driverless operations across a few major cities in that region. In Europe, we use the same asset-light partnership approach that we deployed in the Middle East; we obtain Spain's first national L4 passenger autonomous vehicle permit in Madrid. Preparations for operations are in progress for a commercial launch by year end. In Switzerland, drawing on the driverless permit obtained last year, we introduce public Robotaxi services in Zurich alongside partners SBB and VBZ. In Slovakia, backed by the national Elevate initiative, road testing of the first batch of vehicles is in progress. Singapore has seen cumulative deliveries of over 20,000 rides, which lays the groundwork for a commercial driverless rollout down the road. Our driverless Robotaxi operations in China cover Guangzhou’s Tianhe, Haizhu, and Huangpu districts at present. By the end of this year, we intend to extend into additional core districts. On operating metrics, both fleet size and service footprint are still expanding. Average daily rides per vehicle in Guangzhou held steady at 21 versus the prior quarter, with a peak of 33 in a single day, which underpins positive unit economics. WeRide Go registered users climbed 143.5% quarter-over-quarter. Also, in late October we officially launched our global insurance partnership, with Allianz in Singapore and Ping An Property and Casualty Insurance joining as our first partners. Together, we're building customized insurance products that span the entire life cycle of L4 autonomous driving. These products satisfy the regulatory requirements of local markets in China, Southeast Asia, the Middle East, and the European markets we're entering. They'll support the worldwide commercial rollout of Robotaxi. Our Robobus relies on the same underlying autonomous driving technology stack as Robotaxi. At Zurich Airport, we've achieved fully driverless Robobus operations. In Belgium, our Robobus public transit route continues to run commercially on a stable basis. We're also advancing autonomous driving partnerships with public transportation operators, including Keolis in France; in Slovakia, Robobus is part of the country's national plan for multiple autonomous driving products. The fleet has arrived and begun testing. Additionally, our Robosweeper expanded its footprint in Singapore. Working alongside Chye Thian Maintenance and RE Sustainability, we provide services at major sites throughout Singapore. Furthermore, I take great pride in sharing that we recently earned our eighth championship at the China Intelligent Driving Competition. Across the latest two contests, our ADAS system, WRD 3.0, outperformed rival offerings that had significantly higher computational power. Through our L2++/L3 business, we generate cash, develop our mass-production capabilities, grow our driving data sets, and boost consumer recognition of the WeRide brand. Lastly, allow me to present Jinshuo, our subsidiary focused on AI infrastructure. Drawing on closed-loop data capabilities we built up through our autonomous driving work, this offering provides pre-training data, environments for post-training, and trusted verification spanning the entire journey from pre-training through continue learning. Through automated 3D reconstruction, multimodal alignment, and granular labeling, we have built a substantial library of data sets covering home, retail, and industrial settings; at present we serve dozens of foundation model, autonomous driving, embodied AI, and robotics companies worldwide, delivering standard and customized data sets and skill sets, licensed data processing software, and post-training environments. Orders under framework agreements have now reached several billion, and market demand for physical AI data keeps growing rapidly. Let me also walk through the product stack. Jinshuo consists of three components: Airgoc, SkillForge, and Crucible. Airgoc is a wearable device for multimodal data collection. At the hardware level, it maintains synchronization across multiple data streams. SkillForge turns first-person task data into skill sets. Training, testing, and licensing are among the uses these assets can support. Crucible serves as an environment hub. It converts tasks into interactive environments in which models can operate. Results can be stored separately, and environments are able to connect to a post-training system. Our strategy is unchanged: a single shared physical AI foundation that supports multiple businesses. Those businesses rely on the same core technology and reinforce one another. During this quarter, the overseas asset-light Robotaxi approach has taken firm root in the Middle East and keeps expanding throughout Europe; within China, Guangzhou achieved positive UE. Taken together, these achievements mark a new phase—an inflection point for scaling our L4 business. Over the next 12 months, we've laid out specific milestones that investors can measure us against: our inaugural large-scale commercial rollout in Europe, in Madrid; complete coverage of the core urban areas of Abu Dhabi and Dubai, plus a Middle East fleet that nearly doubles in the coming year; driverless service across more core districts of Guangzhou and additional cities in China; and continued conversion of Jinshuo's framework orders into revenue. Self-driving systems have to react in milliseconds and operate with complete accuracy. That demands specially designed models running live inside the car, trained and validated using enormous volumes of real-world driving data. In the end, who wins in autonomous driving and physical AI is determined by how strong the physical AI foundation is and by the capacity to keep learning from data and folding it back in. Durable advantages stem from technology that is safe and dependable, an expanding store of high-quality data fueling continual enhancement, and a sound, profitable business model; with all of those elements in place, we are confident about succeeding on the road ahead. With that, I will hand the call over to Jennifer, our CFO and head of international. She’s going to take you through the financial results for our third quarter. Thanks.
Thanks, Tony, and good morning to everybody. I'll take you through the financial and operational results we posted for the third quarter of 2026. Let me open with three key points. First, scale. Revenue climbed 88% compared with a year earlier, hitting a record RMB 321.5 million, comfortably above what the Street had expected. Second, quality. Gross margin came in at 43%, and overseas markets account for almost half of total revenue now. That's high-quality revenue: recurring, contracted, and designed to compound as our fleet expands. Third, discipline. The net loss came down 22.6% quarter over quarter to RMB 310 million, and that figure also came in ahead of what the Street had been expecting. We're holding RMB 4.7 billion in cash reserves, which means we have enough funding to carry us through to positive free cash flow. On revenue and gross margin: total revenue reached a record RMB 321.5 million, driven mainly by overseas markets. Revenue from overseas climbed 171.5% to RMB 146.9 million, accounting for close to half of our total revenue. Each of the three businesses contributed. L4 revenue increased 47.2% year over year to RMB 168.8 million, L2++/L3 revenue was up nearly five times, and AI infrastructure revenue about doubled. Gross margin for the quarter came in at 43%. That marks a rise of 10.1 percentage points compared with a year earlier, along with 5.5 percentage points sequentially, bringing gross margin to 39.6% for the first nine months. A more favorable revenue mix drove that improvement. Service revenue climbed 102% versus the prior year and now makes up 58% of total revenue. Two factors are particularly notable. To begin with, Robotaxi service fees from overseas—recorded by us on a net basis—accounted for a larger portion of revenue. In addition, L2++, offered through a licensing model, delivers a gross margin in the high double digits. It's worth noting that this quarter's gross margin also benefited from a favorable revenue mix, which could vary from one quarter to the next; nonetheless, the broader shift toward stronger margins is unmistakable. Turning to how the businesses performed: L4 revenue increased 34.8% compared with the previous quarter. Within L4, Robotaxi revenue tripled — a 204% year-over-year jump to over RMB 100 million — and it now accounts for about 60% of L4 revenue. More than 90% of Robotaxi revenue was generated overseas, with the Middle East leading the way. For our margin, the key question is how we generate this revenue. During the quarter, our Middle East fleet grew 50%, and as operating hours and the service area expanded, our mileage-based revenue sharing climbed faster than total Robotaxi revenue. That revenue is recurring and asset-light, and it increases with each ride. Turning to L2++ and L3: by the close of the quarter, cumulative L2++ deliveries had passed 100,000 units, which is ahead of what we told you to expect at the beginning of the year. By the end of next year, we expect to reach 500,000 units. Because we get paid through NRE and royalty, this business carries a gross margin in the high double digits and produces operating cash flow at low marginal cost. As it scales, it also directly strengthens our operating leverage. AI infrastructure revenue nearly doubled compared with a year earlier, and Jinshuo now holds several billion RMB in orders under framework agreements, which gives us strong visibility into future growth. Demand is accelerating. A number of lead customers that used to purchase millions of hours of data are now looking at tens of millions of hours across the coming two years. This business is asset-light and generates cash, and it also reinforces our own physical AI foundation platform. Now on expenses: R&D expense climbed 47.1% year-over-year, yet just 7.1% quarter-over-quarter. R&D expenses as a share of revenue dropped 40.3 percentage points year-over-year and 42.7 percentage points quarter-over-quarter. These are planned investments, chiefly for developing our next-generation Robotaxi, which once launched we expect to cut BOM costs further on a pre-installed basis. R&D expenses this quarter also take in costs tied to summer campus recruiting and first-half performance bonuses. Our R&D organization is largely built out at this point, and we're not planning any significant growth in R&D staffing. Going forward, the emphasis is on efficiency and output. Selling expense rose 39.2% compared with a year earlier, while falling 9.3% from the prior quarter. As a share of total revenue, it was just 8.2%, which is 2.9 percentage points lower year-over-year and 4.4 percentage points lower quarter-over-quarter. That reflects how efficiently we're expanding the market as well as the benefit of operating leverage. Administrative expense for the quarter totaled RMB 13.9 million. That figure includes a non-cash reversal of about RMB 49 million related to share-based compensation expenses associated with employee departures. Overall, our operating loss improved 12.3% year-over-year and 21.1% quarter-over-quarter, coming in at RMB 332.9 million. Our net loss came in at RMB 310 million, down 22.6% from the prior quarter and roughly level with a year earlier. After backing out share-based compensation and fair value changes, the adjusted net loss stood at RMB 336.5 million, about the same as last quarter, and that held even as we continued putting money into our next-generation Robotaxi. Since revenue continues to grow while our cost base stays steady, we anticipate the adjusted loss will shrink over time. Let me now address the three subjects investors have raised: how we used cash this quarter, how strong our cash position is, and the possibility of equity dilution. We ended the quarter with RMB 4.7 billion in cash reserve. Cash and cash equivalents went down by RMB 1.67 billion, though roughly RMB 1.07 billion of that simply moved out of cash and into short-term investments like time deposits and treasury management products. All told, our cash reserve was lower by about RMB 700 million. This was largely a matter of timing rather than run rate burn. Ahead of the big overseas Robotaxi delivery in the fourth quarter, we purchased vehicles and equipment, and certain sales were still sitting in receivables when the quarter closed; on top of that, we prepaid for our next-generation Robotaxi and paid out first-half-year bonuses. That working capital is now flowing back in. During Q4, we're making solid progress on collecting receivables, and we also anticipate getting prepayments from customers within the quarter. As a result, we foresee our overall cash reserves holding steady through Q4. Going forward, we've made our spending plan tighter and set priorities for our investments. Based on where we presently expect revenue and expenses to land next year, we project that total cash burn will drop to approximately RMB 1 billion in 2027 and continue decreasing in 2028. Achieving our first quarter of positive free cash flow in 2028 is still on schedule, as is positive free cash flow for the entire year in 2029, and by then we anticipate retaining over RMB 1 billion in cash. In short, the cash we have on hand is enough to cover the entire plan through to self-sustaining cash flow. Each of the three businesses is playing a part in getting us there: the overseas operation produces recurring operating cash inflow, L2++ brings in high-margin royalties, and Jinshuo is turning its expanding framework orders into cash inflows too. Finally, regarding equity dilution, I want to be direct: we think our share price fails to reflect what our business is worth, and we have taken action based on that view. Starting in March, we have bought back more than $95 million U.S. We bought back a dollar's worth of our stock, which cut our share count substantially. Since our plans are fully funded, we don't see any reason to issue new shares at a valuation we believe fails to reflect this company's value. Regarding the recent lock-up expiry, under Chapter 18C rules of the Hong Kong Stock Exchange, three senior executives had a 12-month lock-up after our Hong Kong listing last year. That lock-up concluded on November 5th. The expiry applies to shares held by three executives: Dr. Tony Han, our founder and CEO; Dr. Yan Li, our co-founder and CTO; and Dr. Hua Zhong, our senior vice president. Not one of these executives has ever sold shares in the company. Dr. Tony Han’s shares are still entirely locked up through October 2028 as part of his voluntary three-year commitment. The management team has complete confidence in what lies ahead for WeRide. With the strong results we posted this quarter, we now anticipate 2026 revenue of no less than RMB 1.1 billion—a gain of over 60% versus 2025 and above what the Street is forecasting. As we look toward 2027, the drivers of our growth are evident: an overseas fleet we expect to double, expansion across Europe, the Middle East, and APAC; L2++ deliveries climbing toward 500,000 units; and Jinshuo turning framework orders into revenue. We intend to achieve this growth while reducing cash burn to roughly RMB 1 billion in the coming year. Longer term, we anticipate our commercial fleet expanding by approximately ten times every three years. Accelerated expansion, improved margins, and reduced cash burn — those are the means by which we reach self-sustaining cash flow and deliver long-term value to shareholders. That wraps up my financial overview. We're ready to take your questions now. Thank you.
Appreciate it, Tony and Jennifer, for taking my question. There are two things I'd like to ask. First, on scaling the Robotaxi fleet in a way that's sustainable, how should we think about how that compares with global leaders such as Waymo and Tesla? Then my second question concerns competition in Europe. More and more Chinese Robotaxi companies are entering the Europe market, it seems. How should we view the competition that may be rising in that market? Those are my two questions. Thank you.
I’ll take the first two questions. The first one is about scalability relative to Waymo and Tesla—essentially, how do we respond to competition from Waymo and Tesla? The second is more specific to Europe: as everyone knows, Waymo is also coming into the European market, so what does the competitive landscape in Europe look like, and how will WeRide compete with Waymo there? I think in any field—any interesting field or profitable field—there will be plenty of competition. We respect Waymo and Tesla a great deal; they are tech giants and two great companies. We are honored to compete against them. The thing is, I think we have our own competitive edge. I believe very few companies worldwide can manage the entire process, from driverless operation and local compliance through to commercial delivery. Of course, Waymo and Tesla have already achieved certain results, but I believe the market is large enough for multiple players, and WeRide is definitely among one of the major players in the global market. Let’s talk about our competitive edge, okay, compared with Waymo and Tesla. For the global market, there’s one thing I want to mention, and that is our technology, okay? Our L4 system combines a one-stage end-to-end model with dedicated safety architecture that, to my best knowledge, is superior to other companies’ technology. We have built-in redundancy and an independent rule-based safety layer. Basically, if you talk about Tesla, Tesla is basically an end-to-end model, and Waymo, according to our best knowledge, is still based on the rules, but WeRide’s approach is a kind of marriage of the two approaches, and we can adapt quickly to new environments, and the safety architecture meets strict standards like European standard. This is our one technology advance. Second is about scalability. If you look at the efficiency and hardware costs, you will find out, especially for international deployment, we are in a much better position. Our vehicles are cheaper, and our hardware is more cost-efficient, and we can leverage the superior supply chain of the China auto industry. Regarding the European market, we have another advantage, which is early entry; we actually entered the China and Middle East markets very early, and now we entered the Europe market very early. We got Spain’s first national L4 operating permit; with all of this first-mover advantage, I think we expect to remain in our leading position in the European market. Last but not the least, that is our capital efficiency. We work with partners like Uber and Grab under an asset-light model, and we don’t need to build our own customer acquisition, dispatch service, and insurance system from scratch. This is a great advantage over Tesla and Waymo; they all build up their things on their own. Waymo and Tesla, they are kind of asset-intensive model; with all of this kind of aforementioned factors, I think we are really in a leading position. Once more, as you've noticed, WeRide is a startup that was actually founded in 2017, so we're a much younger company than the other two. Overall, if you look at our deployment footprint and our scalability, we're comparable to those other two companies. We have confidence in our final results, and we're confident about our final position in this very competitive field. That's my answer.
Thank you very much. That's very helpful.
Hello, Tony and Jennifer, and thanks for taking my question. I've got just one question. As you said, you've shown positive unit economics overseas and validated a business model that seems to work for both WeRide and local partners. I'm wondering when you expect the scaling to pick up speed from here, and what has to happen for deployment volumes to accelerate in a meaningful way over the next few years? Thank you.
Thanks, Tim. I'll handle that one. Our Robotaxi operations abroad are in very good shape, and they generate profit both for us and for the partner. On our side, the business breaks into two pieces: at the point of initial delivery, we sell the Robotaxi, and that comes bundled with roughly six years of recurring revenue sharing based on mileage. Every vehicle brings us upfront revenue along with a long-term recurring income stream. It's also asset-light from our perspective. Our platform partners and the local partners who own the vehicles have a very strong incentive to get utilization as high as possible. We concentrate on the technology and the capabilities that make the vehicle genuinely driverless, plus, say, remote assistance, and because of that we need less capital. We're actually able to move very quickly. On scaling, the same principle applies; in each market, going from testing with a safety driver to fully driverless operation typically takes at least, say, a year in major cities. That's the time required just to finish all the mandated testing procedures and build trust with regulators, particularly in Europe. At this point, we've also noticed that since we began earning regulatory confidence across multiple markets, entering a new market now actually goes a bit faster than the one before. The trend is clear. To address your question about scaling up, this year the count of vehicles we've deployed overseas has already surpassed our own expectations. Say the Middle East fleet rose 50% in a single quarter to roughly 600 vehicles, and we expect it to hit somewhere around 800 to 1,000 by year end, then double next year; in Europe, we'll have more than 100 L4 vehicles by the end of the year, which will be the biggest L4 fleet anywhere in Europe. Our main constraint isn't really the tech, it's really just permit timing. Demand is already locked in. Here's a very important number: we hold binding contracts for about 3,000 Robotaxis globally as of today, signed with multiple partners for delivery in the near term, as long as we can secure the permit. Once the permit comes through, those will translate directly into revenues for our overseas business. The demand is there. We're working very hard to earn regulatory trust as fast as we can so it can scale up. As of now, Tony also just mentioned, we're already Uber's largest driverless partner outside the U.S; we're also the largest driverless partner for multiple platforms; we can already see the positive trends. What matters more, our industry-leading safety record overseas and also the largest overseas fleet, all of that speaks for itself. Those track records are our long-term moat, and our reputation, our safety procedure, and long-term partnership across the local ecosystem are very hard to replicate, and they're what lets us scale truly globally. Yeah. I hope that answers your question, Tim.
Thank you.
Yes, thanks Tony and Jennifer for taking my question. I've got one question. You just said that Level 4 Robotaxi operations in one Chinese city have become profitable. What are your plans when it comes to expanding into other cities across China, and could you also share your global Robotaxi expansion plans for year-end and next year? Thank you.
I'll take that question. During this quarter, Guangzhou reached positive UE at the city level for an entire quarter for the first time ever. I think that's a terrific milestone; it demonstrates that our operational model works, right? The cost structure and dispatch system are able to support a profitable business. When it comes to China expansion, I believe the policy environment is clearly heading in a positive direction. Just last week, oh, sorry, this week, I think, at the APEC CEO Summit in Shenzhen, President Xi emphasized autonomous driving as among the most frontier technology in China, and strong support will come from the central government. With national policy becoming more supportive and standardized, operators possessing mature technology and solid safety records—we will certainly take our model to the remainder of the city. At present, our priority in Guangzhou is to keep growing the fleet, broaden service areas, and boost utilization there. As policy opens up for additional cities, we will just replicate this model and expect to achieve positive unit economics across multiple cities. That is our plan. Essentially, keep honing our technology, the models, and the efficiencies in Guangzhou, since it is already profitable, and then as soon as other cities open up the policy, we simply copy quickly to those cities. Once we hit positive unit economics in China, the goal is to scale quickly into the rest of the world as well. In the Middle East, we'll continue growing the fleet and reinforcing our position in that market. Over in Europe, we're standing up a complete commercial operation in Madrid, which ranks among Europe's three biggest ride-hailing cities. Looking ahead, our current work is focused on Switzerland, Denmark, and Slovakia. We expect the Middle East to keep expanding next year, and Europe to turn into another key growth engine as we take the Madrid model into other European markets that have supportive policies. Broadly speaking, our aim is for the operating Robotaxi fleet to grow by about ten times every three years. What matters most to us is having unit economics that are positive, cash flow that is healthy, and operations that are sustainable. That is our central focus, and we have strong confidence in our plan. Thank you. That is what I have to say.
That's very clear. Thank you.
Hello Tony and Jennifer, I have a couple of questions. First, what would you say about the strategic partnership with Uber? What do you intend to do with that partnership? Thanks.
Tony, do you want to answer?
Oh, yeah, I forgot to turn my microphone on. I'll take that question. Uber is certainly our main strategic partner and a significant shareholder. I think WeRide and Uber are actually complementary in their strengths. WeRide contributes the L4 technology, the safety system, and a track record that regulators have seen proven, while Uber contributes its worldwide ride-based platform and its resources. Together, I believe we make an alliance that has no equal, and I don't see any other alliance or any competitors being able to compete with the WeRide and Uber alliance. The Middle East has demonstrated that model. As of today, we run Uber's biggest driverless fleet outside the U.S., and we expect our fleet in the Middle East to reach thousands of vehicles before long. Europe represents a very large opportunity, and we launched Europe's first driverless commercial Robobus service on public roads in France, but Madrid is what we are more excited about going forward. With Uber, we are progressing well in Madrid, and we expect to launch in several additional European cities next year. We look forward to sharing the relevant information once we have announced it. Our model also works with multiple platforms and local partners, and we are able to enter each market alongside the right partners. Also, I want to stress that we have an asset-light model, which has already been proven profitable and workable. At present our focus is on high utilization and more efficient operation; that's my answer to your question. He has a second question, right? Jen, you... Oh, that's my answer. I see. For the second question, I think Jen may answer. Okay.
I think Ming has not asked the second question yet. Ming, please go ahead.
Jen.
Thank you. Lately, progress at the frontier of AI has prompted debate over whether general-purpose AI models might one day solve autonomous driving. What is your view on how foundation models will shape the Robotaxi industry over the long term, and as AI capability keeps evolving, where will durable competitive advantages actually sit?
Tony, you're on mute.
Should you field that question, or should I? I'll field it.
It's on frontier LLM. I think it's better you take it.
I’ll take this one, yeah. Honestly, because of the jet lag I’m feeling a bit tired, but I think I’ll go ahead and answer it. This is a very heated discussion and debate. Thank you for such a great question. I think it’s really a great question; broadly speaking, you could say AI can replace everything. These days, whatever the case, we can’t safely claim that anything is immune to replacement by AI. If you look closely, okay, we can see that general-purpose models are very strong at high-level tasks such as reasoning and coding, but controlling a car is a completely different problem; it’s just like how someone might believe a humanoid robot could replace every top-level surgeon to perform surgery on patients, okay? That’s still something from a sci-fi movie. In a recent public test, even the most advanced general-purpose model could barely finish a single lap of an empty parking lot, okay? Driving is a matter of life and death. The system must respond within milliseconds with absolute precision, and mistakes are not acceptable. Let me emphasize, mistakes are not acceptable, and auto-grade chips with high power limits. This kind of thing really requires a lot of technology to truly make it work. All of the hardware for the autonomous driving vehicle is purpose-built. I don’t think a general AI model can run on this kind of platform within the next three or four years. These are just some general discussion points. People talk about it in a general way, but when it comes to the field, we have studied large language models very carefully. By the way, I think I’m in a position to say so because my background is in vision and AI. I think, for the general-purpose foundation model, they currently don’t have a chance, and because we have to have a very compact model on it, and precision is key. Over time, foundation models may gradually improve, but there’s a long way to go. I think in the near future, within the five-year framework, I don’t see any chance for this generic-built purpose AI to write an autonomous driving algorithm and replace the current algorithm. Within five years, if you look at our plan, we have already deployed our Robotaxi fleet, covered the global market, and we will also keep an eye on this general-purpose algorithm, and as I mentioned, we have built up all the physical AI relevant infrastructure. If the general-purpose LLM can work, I think we are among the first group of companies to adopt this technology. Overall, we embrace this technology, and we are confident this company will be the first pioneer to adopt this technology, and we will remain in a leading position. As a first-mover company, we will remain at our current position and enjoy the consequential profit and revenue; that’s my answer to this question. Thank you. Great question, very insightful.
Let me add a bit more on that question. A better model by itself won't get a driverless car onto the road. What we do very intensively is hold discussions with the regulator, and the regulator has to approve and validate the system in each specific market. It isn't the model that does this, and the real heavy lifting lies in building trust, integrating the software and hardware, testing across cities and countries, and proving safety over millions of kilometers, and after that we must operate at scale on an everyday basis. That's the reason WeRide has spent the years building, and at this point we hold permits in nine countries. We're running in 13 countries, in 60 cities; that is actually a key moat we have built over the years. Thank you.
Thank you, Tony and Jennifer.
Thank you, team. This is Yuqian speaking. My question actually has two parts, and it concerns the China Level 2+ market. Competition has clearly been heating up, OEMs are having a hard time, and there is likely added pressure when it comes to pricing. Going forward, how would management assess the competitive landscape? And could you also tell us about your order growth in terms of both volume and pricing? Thank you.
I'll take that question. I have to say competition is pretty fierce out there, and carmakers watch costs very closely, and they genuinely want to make smart driving accessible to more customers. We see a big opportunity here. Urban navigation on autopilot still has low penetration at the moment, and we see a very large market, a major opportunity. There's plenty of room and long-term growth ahead. I think our goal is to bring smart driving, urban navigation included, to everyone, both in China and in overseas markets. We can do that because our cost is significantly lower than our peers. Our one-stage end-to-end solution actually just shares the technology we developed with our L4 driverless operation. The record speaks for itself: in China's intelligent driving competition, we took home eight championships — which essentially shows that our ADAS system is just better. It outperforms our rivals by a wide margin, we've secured over 20 mass-production design wins, and additional volume is on the way. Frankly, we're quite confident about growth going forward. Broadly speaking, L2++ has a very distinct part to play in our strategy — it strengthens our brand with consumers and expands our driving data — and essentially, our L2++ application and our L4 Robotaxi operation reinforce one another, sharing data and helping each other refine the algorithms. Looking ahead, achieving L2++ demands increasingly powerful compute platforms capable of incorporating the more sophisticated L4 technology WeRide has built up over the last decade, and it ranks among the top companies in L4. On the whole, I believe WeRide will hold onto its leadership in ADAS, and a growing number of OEMs will embrace our system. I anticipate capturing an even bigger share of the ADAS market, and as I said earlier, NOA penetration will continue rising—I think to 100%—since most people would rather not drive themselves and simply want to depend on the system to handle driving. All things considered, thanks to the technology we have, we're sure we'll keep the top spot we hold today. Thanks for asking.
Thank you, Tony.
Hello there, and good evening to everyone. My question concerns unit economics. During this quarter, WeRide reached positive unit economics in Guangzhou. Could you explain the way you define UE and which costs get included, along with roughly how those costs break down? Thank you.
I’ll handle that question. Getting to unit economic breakeven is genuinely a major milestone for us. Let me be quite specific about what that figure covers. On the revenue side, we include net ride-hailing revenue after refunds. On the cost side, we count every major cost line: vehicle depreciation, covering the battery through to autonomous driving hardware, plus charging, salaries for remote staff and ground-support staff, and also cleaning, maintenance, insurance, parking, network, and telecom fees. All of it is in there. A bit under half of total cost is tied to vehicle depreciation, with the remainder made up of the other items; that gives you a very thorough accounting of all cost components. In Guangzhou, we turned positive UE for the first time at an average of 21 orders per vehicle per day. Three factors are really driving that: higher utilization, longer average trip distance, and a more efficient staff-to-vehicle ratio for both remote assistants and ground support. We’ll keep improving those ratios so per-vehicle labor cost stays as low as possible. And as the fleet expands, we want to ensure per-vehicle labor cost can remain really low. If, say, we can already hit positive UE in Guangzhou, then overseas the structure becomes even more attractive. With our asset-light business model, partners own the vehicles, so we don’t bear the CapEx for them. Beyond that, we take what we’ve learned in China into our overseas markets. At present, we’ve already lifted remote assistant efficiency in the Middle East to better than one to 10. We’re also refining dispatch systems, the maintenance process, and fleet management playbooks; put simply, China is among the most demanding operating environments anywhere. Bringing that experience abroad helps us and our partners too, streamlining cost and launching new markets faster. This matters even more in places such as Europe, where at initial launch, say, labor and energy costs can run very high. Reaching UE breakeven in China is a key milestone for us. Yeah.
Thank you.
Actually, I believe time is nearly up. Perhaps we could take just one additional question—one final question.
Appreciate it. Hello, management—given the time limit, I’ll keep this short. Congrats on such a strong quarter. I have two questions: first, is our cash sufficient to get to breakeven in 2029? Which costs are temporary and which will be recurring? Second, under the current share price, will the company issue new shares or do a placement? Thanks very much.
I'll keep this short. On your first question, the answer is yes — our cash is sufficient to get us to breakeven in 2029. As we said earlier on this call, we closed the quarter with RMB 4.7 billion in cash reserves, and even more significant, during the fourth quarter the collection of receivables plus the prepayment ought to hold our cash reserves approximately flat versus where they stand now. We also expect cash burn to drop to roughly RMB 1 billion in 2027, then fall further still in 2028. In 2028, that is the point at which we anticipate our first-ever quarter of positive free cash flow. On that basis, assuming everything goes to plan, in 2029, when we expect positive free cash flow for the full year, we would still hold more than 1 billion in cash. That is the runway as we see it. You also raised the temporary items: the one-offs will be, let's say, for the last quarter, the first-half performance bonus and the upfront payment for our next-generation Robotaxi; those are the one-offs. On stage-specific investment, let's say the computing cluster, we will still build some, particularly in Europe and the Middle East, and that actually follows the deployment schedule quite closely, but those costs get spread across a much larger revenue base. As for what the recurring cost will be, let's say the R&D staff and the model training, those are the ongoing recurring items. Our view is that we validate unit economics in each market before we truly scale in that market, so that keeps our cash burn under control. In short, we have full confidence about reaching breakeven in 2029, and revenue growth is genuinely continuous. On the last question, I believe you also asked about issuing new shares: yes, as we said just now, we feel the current valuation does not reflect the company's true value, and we will not carry out equity issuance or placement at the current valuation. We won't do it. Our priority is to genuinely improve cash flows through the business itself: scaling overseas by using our asset-light L4 business, our growing L2++ production deliveries, and also Jinshuo's data set services; those will help us move steadily towards positive cash flow. Thank you.
Thank you so much.
Thank you very much.
Thank you.
Read the full transcript via the API
You're viewing the first half of this call. Get the complete WeRide Inc. transcript - plus 132,000+ transcripts from 7,000+ companies, speaker segments and full-text search - through the EarningsAPI REST API or hosted MCP server.
Get an API key View API docs →For developers and AI pipelines
Programmatic access to WeRide Inc. earnings transcripts and 132,000+ others is available through the
EarningsAPI REST API and the hosted MCP server.
Quarterly plans from $162 - full transcripts, speaker segments, full-text search,
and the /api/v1/transcripts/recent polling endpoint for ETL pipelines.