XPeng Inc.

Hello, ladies and gentlemen. Thank you for standing by for the third quarter 2025 earnings conference call for Xpeng Inc. At this time, all participants are in listen-only mode. After management's remarks, there will be a question and answer session. Today's conference call is being recorded. I will now turn the call over to your host, Mr. Alex Shea, head of investor relations and capital markets of the company. Please go ahead, Alex.

Thank you. Hello, everyone, and welcome to EXPO's third quarter 2025 earnings conference call. Our financial and operating results were issued by Newswire Service earlier today and available online. You can also view the earnings press release by visiting the IR section of our website at ir.xiaopeng.com. Participants on today's call from our management team will include co-founder, chairman, and CEO, Mr. He Xiaopeng, vice chairman and president, Dr. Brian Wu, Vice President of Corporate Finance and VW Projects, Mr. Charles Zhang, Vice President of Finance and Accounting, Mr. James Wu, and myself. Management will begin with prepared remarks. The call will conclude with a Q&A session. A webcast replay of this conference call will be available on the IR section of our website. Before we continue, please know that today's discussion will contain full-looking statements made under the safe harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. Four looking statements involve inherent risks and certainties. As such, the company's results may be materially different from the views expressed today. Further information regarding these and other risks and uncertainties is included in the relevant public filings of the company as filed with the U.S. Securities and Exchange Commission. The company does not assume any obligation to update any forward-looking statements except as required under applicable law. Please also note that EXPON's earnings press release and this conference call includes disclosure of unaudited GAAP financial measures as well as unaudited non-GAAP financial measures. EXPON's earnings press release contains a reconciliation of the unaudited non-GAAP measures to the unaudited GAAP measures. I will now turn the call over to our co-founder, chairman, and CEO, Mr. He Xiaopeng.

Please go ahead. Hello, everyone. Hello, everyone. After the launch of the new Xiaopeng P7, it quickly reached $200,000 to $300,000. Pure electricity is one of the top three, leading us to $40,000 from the beginning of September. In the third quarter, the company's comprehensive profit broke 20% for the first time, and the loss was further shortened. Our goal is to achieve a profit-loss balance in the fourth quarter. In Q3 2025, Xpeng reported record results in key operating metrics with new highs in deliveries, revenue, gross margin, and cash reserves.

Vehicle deliveries for the quarter totaled 116,007 units, a 149% increase year over year. The all-new Xpeng P7, launched recently, quickly became one of the top three BEV sedans priced between 200,000 RMB to 300,000 RMB, boosting monthly deliveries to over 40,000 units starting in September. Additionally, the company's gross margin exceeded 20% for the first time in Q3, and will reduce our net loss further. Our goal is to achieve break-even for the company in the fourth quarter. These continuous operational improvements strengthen our focus on physical AI R&D, supporting the targeted mass production of our VLA 2.0 model, RoboTaxi, and humanoid robots in 2026.

With the deep integration of AI models and the physical world in data, I believe that machines will gradually be able to interact, communicate, and even create in the physical world. I believe this will change the way we travel and live in the future. In the past 11 years, we have been committed to the full-fledged industry. From the development of software-defined vehicles to today's physical AI, we strongly feel that the two biggest applications of physical AI, vehicle and man-made infrastructure, can be based on the same physical world model, chip, and sound system fast reception and evolution. What excites me is that our physical AI system has begun to show a lot of potential. In the next ten years, our vision is to make small and medium-sized automobiles into a global giant smart company. By using physical AI, we will create a powerful product technology system and commercial ecosystem. We not only want to bring AI cars to the global market, but also large-scale deployment of pre-installed and mass-produced Lobotaxi, and to achieve the scale of human-made machines. I believe that the value created by the application of physical AI for humans must be maximized through the opening of the ecosystem. To do this, we will open up our physical world model and cooperate with new platforms to promote the services of Robotastic and open the SDK for human-made machines. These will expand the ecosystem of our physical AI application and accelerate the value creation together with the partners of business and technology.

As AI models advance and become increasingly integrated with real-world data, machines are slowly gaining the ability to interact, communicate, transform, and create within our physical environment. This development is reshaping the future of mobility and daily life. Over the past 11 years, XPeng has dedicated itself to building full-tech technologies in-house, evolving from software-defined vehicles to the emerging realm of physical AI. We understand that vehicles and humanoid robots, the two primarily applications of physical AI, share a homogeneous physical world model, SOCs, and infrastructure, allowing for rapid iteration and evolution. Excitingly, new capabilities are continuously emerging from our physical AI technology spec. Over the next decade, my goal is to make XPeng a leading global company in embodied intelligence. Focused on physical AI applications, We're developing an extensive portfolio of technologies, products, and a supporting business ecosystem. Besides providing AI-powered vehicles to consumers worldwide, we aim to deploy pre-installed mass-produced Robotaxis on a large scale and achieve the mass production of humanoid robots. We believe that an open and dynamic ecosystem is crucial to unlocking the full potential of physical AI for humanity. To achieve this, we plan to open source our physical world model, launch RoboTaxi services in partnership with mobility platforms, and release our humanoid robot SDK. This approach will expand the physical AI application ecosystem through collaborations with business and technology partners and accelerate the value creation process.

我非常开心的是,我们的AI汽车业务正在开启一车双轮的大产品周期。 那么通过更全面的动力和产品的组合布局,会加速扩大规模以及在新能源市场的份额。 那么在11月6号,我们推出行业领先的大电池磷酸铁力5C的真诚车型,也就是小黄F9超级真诚开启预售,综合续航达到了1602公里。 I believe this is the world's longest, with the largest air flow rate, the smallest turning radius, and the most user-friendly flagship model. I believe the super-large scale will help accelerate the transition of a large number of blue-and-yellow users to new energy vehicles. After our super-large scale launch, we see that we have gained unprecedented user attention in the northern region and more inland cities. This has transformed a group of customers that we can't reach by charging. X9 super authentic pre-order order has reached nearly 4 times of the previous pre-order same period data. X9 super authentic will be officially launched on November 20th, and the last delivery, I believe, will be re-released in December at 12th.

I'm also glad to report that as we introduce the one-vehicle dual energy product cycle for AI vehicles, we'll expand our scale and increase our NEV market share through a wider product range. On November the 6th, we launched pre-sales for the XPeng X9 Super Extended Range EV, an industry frontrunner in extended range vehicles equipped with a 5C rate high-capacity LFP battery and a total range of up to 1,602 kilometers. It is the world's first large seven-seater to offer the longest range, highest AI computing power, smallest turning radius, and most efficient space utilization in its category. We see super extended range EVs as crucial for accelerating the shift from ICE vehicles to NEVs. Since pre-sales began for the X9 Super EREV, We've experienced unprecedented interest, especially in northern regions and inland cities of China, attracting many customers who previously hesitated to switch to BEV models. To date, pre-orders for this model are nearly three times higher than the pre-sale of the previous X9 on a light-for-light basis. The X9 Super EREV will officially launch on November 20th, with deliveries starting immediately afterward. I anticipate reaching a new delivery record in December.

I anticipate reaching a new delivery record in December. We are very happy that we will begin to enter a larger product cycle in 2026. In 2026, Xiaofeng will also launch four new models of one-seater dual-cylinder. These include the first-hand products that we put in the important segment market. This will help us achieve past breakthroughs in these categories, and will create more hot sales in the market, such as Mora M03. 我相信在明年上市的七款具备超级增增配置的一车双轮车型将会极大幅度拓宽我们车型可达市场空间,也就是TAM,为我们带来显著的销量增长。

We plan to introduce three super extended range products in Q1 2026, focusing on alleviating key challenges for EREV users by offering long, pure electric range and quicker 5C supercharging, thereby capturing more of the EREV market. We have put in more R&D expenses in 2025. As a result, in 2026, we'll also launch four new one-vehicle dual energy models, including our first product launch in some key market segments, These innovative products will help us establish a presence in these markets and build leading products like the Mona M03. I'm confident that the seven one-vehicle dual energy models with super extended range technology debuting next year will greatly increase our total addressable market, or TAM, and provide significant sales growth opportunities.

We not only have high sales growth, but also long-term growth through the deployment of localization. In September, Xiaopeng's overseas monthly sales broke through 5,000 for the first time, and grew 79% in total. In the third quarter, we added 56 stores overseas, and the sales service network covers 52 countries and regions around the world. In the third quarter, we held the first localization project in Europe. is officially launched at the McDonald's factory in Gratsch, Austria. The first batch of XiaoPeng's G6 and G9 will be launched smoothly. At the same time, XiaoPeng is also launched at the R&D center in Dehombo-Lihe. We will strengthen the needs of overseas users and accelerate technological innovation and product landing. In 2026, we also plan to launch at least three new models in the overseas market.

On the global business front, we maintain strong sales growth and establish a solid foundation for long-term expansion through our localized approach. In September 2025, our monthly overseas deliveries exceeded 5,000 units for the first time, a 79% increase year-over-year. During the third quarter, we grew our global presence with 56 new overseas stores, expanding our sales and service network to 52 countries and regions worldwide. Additionally, our first European localized production facility at Magnetland in Graz, Austria, officially commenced operations, with the initial batch of XPeng G6s and G9s rolling off the line. Simultaneously, XPeng's R&D center in Munich, Germany, officially began functioning, helping us better understand overseas customer needs and accelerate technological advancement and product launches. In 2026, we plan to introduce three new overseas models, including popular mid-to-small SUVs that meet the diverse preferences of global consumers.

Our AI model We are going to launch Xiaofeng's second generation VLA model. It has increased the number of parameters of the previous generation of end-to-end models by 10 times, and has increased the experience and ability of the entire self-driving system to a new level. In my recent self-driving VLA model, In terms of response, the experience of the entire complex path is extremely exciting. There is no comparison in the current stage. At the end of December, we will open up the co-creation experience of先锋 users. In the first quarter of 2026, we plan to launch the second generation of VIA models by the full capacity of Ultra models. I believe that the mass production of this generation of VIA models will be a huge leap.

Our strong focus on investing in AI large models, computing infrastructure, and data set is driving the continuous emergence of advanced capabilities from our physical world model. Our upcoming VLA 2.0 model, which has 10 times more parameters than its predecessors, will substantially enhance safety and user experience in intelligent driving. From my own reason driving experience during very complicated and complex road conditions, we experienced very impressive and unparalleled driving experience from the intelligent VLA model. So starting from late December, we will initiate a co-creation program with our early adopters. In the early quarter of 2026, we aim to deploy the VLA 2.0 model across the entire Ultra lineup. I see the mass production of VLA 2.0 as a major breakthrough in physical AI models. offering a significant generational leap in user experience and attracting more people to choose XPeng for its leading intelligent driving technology.

Next, the second-generation VR model of Xiaofeng will also open up to business partners in the global market. We hope to bring the first-generation smart driving experience to more users around the world. The public will become the first customers of Xiaofeng's second-generation VR model. At the same time, Xiaofeng's Turing AI chip

Going forward, XPeng will open-source its VLA 2.0 model to global commercial partners, aiming to provide industry-leading advanced driver assistance experience to a wider audience. Volkswagen will be the initial launch customer for the VLA 2.0 model, Additionally, Xpeng's Turing AI SoC has earned a formal sourcing destination from Volkswagen, with co-developed vehicles expected to start mass production early next year.

We will generate authorization for access to foreign technology, and we will continue to develop it. We will invest in the development of virtual chip, VR model, and other types of technology and commercialization. Revenue from licensing our technology to external partnerships will be reinvested into our R&D, mainly to support iteration and upgrades of the Turing SoC and VLA models.

This fosters a positive cycle of innovation and commercialization. We invite more automakers and tier 1 manufacturers to collaborate with us on the Turing SOC and VLA 2.0, working together to promote the adoption of advanced intelligent technologies in both Chinese and global markets.

I think in the past, the large-scale models, they were mostly made into the best L2 models. But today we are very happy to see that the physical world model makes the era of unmanned driving really speed up. I believe that only the pre-assembled mass production and the large-scale popularization of RoboTaxi, which is powerful in the field of human resources, and the commercial model are maintained at the same time. In the year of 2026, we will launch three types of RoboTaxi. Our RoboTaxi technical route is not to use high-end maps, not to use digital data, but can solve the RoboTaxi industry very well. high cost network, especially deployment cost network, travel, more restrictions, weak emulation capacity, and other chain problems. So support will be in the global range of high-speed, efficient, and scaled deployment. We plan to open a private operation of Xiaomeng Robotics in China in 2026, and further polish the software hardware and operation ecology of Robotics. China can only form an eco-friendly ecosystem with a comprehensive industrial chain. . . . . . .

Traditionally, end-to-end models were able to maybe reach advanced level two at its best. However, the rise of physical world model is speeding up the arrival of true autonomous driving. I believe that only pre-installed mass-produced robo-taxis with a strong ability to generalize can achieve widespread adoption and create a sustainable business model. In 2026, XPeng plans to launch three robo-taxi models, Our technology set for RoboTaxi does not depend on high-definition maps or LiDAR. This approach enables us to address current industry's challenges, including high costs, operational limitations, and poor generalization, allowing for an efficient and scalable deployment worldwide. We intend to begin pilot operations of XPeng RoboTaxi in China in 2026, continuously improving both software and hardware of RoboTaxi while building an operational ecosystem. I believe that a collaborative ecosystem where all industry stakeholders benefit is key to scaling rapidly. Therefore, we plan to open our SDK to our partners, and AMAP will be the first ecosystem partner for XPeng RoboTaxi.

We also invite more companies in the mobility sector to explore RoboTaxi collaboration opportunities with us.

Xiaopeng's energy savings have been used up.

in the same technical route driven by the physical world model. Then, with the full support of the whole car, the power, and many other teams, this time, Xiaofeng showed the new generation of Aeron engine in the Science and Technology Day, and the shape of the engine, the elegant posture, finally brought a huge surprise and emotion to many friends, and also produced a huge cross-circuit effect. This makes us see the very big commercial potential of the future of the energy-powered engine. In fact, this time, IRON has only shown a very small part of its capabilities. In the second quarter of 2026, we hope to realize the full-fledged and rich energy inheritance through cross-border integration and innovation. This will definitely bring a far greater effect and experience than the current market. At the end of 2026, we look forward to the first large-scale production of high-end energy inheritance. Then we will first enter the commercial scene, to provide services, including islanding, island shopping, and island groups. I hope that in the next year, we will be able to see Erlang become our new classmate in Xiaopeng's doorstep, office area, and factory. At the same time, in the next step, Xiaopeng will also welcome the global developer, Open SDK, to develop together on Erlang to make Erlang more diverse and long-term. have a stronger ability to create more extensive applications. From a long-term perspective, I think the market potential of man-made robots is greater than that of cars. Once the robot has surpassed the commercialization and electrification concepts like the Chinese EV industry, I believe it will cause a great surge. I expect that in 2030, Xiaohong's annual sales will exceed 1 million units a year.

Our humanoid robots adopt a technology roadmap driven by physical world model. With full support from our vehicle and power chain R&D teams, we unveiled our next generation Iron robot at the latest Xpeng Tech Day. The Iron's human-like posture and agile gait surprised and deeply moved many Xpeng fans and also highlighted the great commercial potential of humanoid robots. Currently, Iron demonstrates only a very small fraction of its capabilities. In Q2 2026, we plan to achieve full capability integration through cross-domain innovation, aiming for performance and user experience for far surpassed current market offerings. Our target is to begin mass production of advanced humanoid robots by the end of 2026. Once produced, IRON will be first deployed in commercial scenarios, providing services like tour guiding, retail assistance, and patrols. By the end of next year, I hope Iron will be working alongside us at XPang stores, campuses, and factories as our new team members. Additionally, XPang Robotics will open its SDK to global developers, inviting partners from various industries to collaborate on secondary development. This will enable Iron to be trained and to evolve across diverse and long-tail real-world scenarios, unlocking broader application possibilities From a long-term perspective, I believe the market potential for humanoid robots will exceed that of automobiles. Once a new generation of robots reaches the inflection point, just as China's EV industry did with electrification, we expect explosive growth ahead.

I envision that by 2030, Xpeng robots could sell over 1 million units annually.

With the opening of the product cycle of one car and two people, I expect the total turnover of the fourth quarter to be 12.5 million to 13.2 million units, which is 36% to 44% in the same ratio. The income of the fourth quarter is expected to be 21.5 billion yuan to 23 billion yuan, which is 33% to 42% in the same ratio. So Xiaopeng AI's scale and market share is at a high speed, expanding to the early stage. I believe that NovoTest and Renyin Jicheng are accelerating the scale of production. I believe that Xiaopeng will become a leader in China, even in the world of physical AI, creating greater value for global customers and shareholders.

With the launch of our one vehicle dual energy product cycle, I expect total deliveries in the fourth quarter to reach between 125,000 and 132,000 units, reflecting a year-over-year growth of 36.6% to 44.3%. We project fourth quarter revenue to be roughly between 21.5 billion RMB to 23 billion RMB. up 33.5% to 42.8% from the previous year. XPeng's AI-driven vehicle business is in the early stages of rapid expansion in terms of scale and market shares, while Robotaxi and humanoid robot programs are swiftly moving forward and towards mass production. I'm confident that XPeng will establish itself as a leader in physical AI, both in China and globally.

delivering greater value for our customers and shareholders worldwide.

Thank you, everyone.

With that, I'll now turn the call over to our VP of Finance, Mr. James, who will discuss our financial performance for the third quarter of 2025.

Thank you, Xiaopeng.

Now let me provide a brief overview of our financial results for the third quarter of 2025. I'll reference RMB only in my discussion today, unless otherwise stated. Our total revenues were $20.38 billion for the third quarter of 2025, an increase of 101.8% year-over-year, and an increase of 11.5% quarter-over-quarter. Revenues from vehicle sales were 18.05 billion for the third quarter of 2025, an increase of 105.3% year-over-year and an increase of 6.9% quarter-over-quarter. The year-over-year and quarter-over-quarter increases were mainly attributable to higher deliveries from newly launched vehicle models. Revenues from services and others were 2.33 billion. for the third quarter of 2025, representing an increase of 78.1% year-over-year and an increase of 67.3% quarter-over-quarter. The year-over-year and quarter-over-quarter increases were primarily attributable to the increased revenues from after-sales services and technical R&D services rendered to the Volkswagen Group due to the successful achievement of certain key milestones in the current quarter. Gross margin was 20.1% for the third quarter of 2025, compared with 15.3% for the same period of 2024 and 17.3% for the second quarter of 2025. Vehicle margin was 13.1% for the third quarter of 2025, compared with 8.6% for the same period of 2024 and 14.3% for the second quarter of 2025. The year-over-year increase was primarily attributable to the ongoing cost reduction, while the quarter-over-quarter decrease was due to targeted promotion to clear outgoing inventory during product transition. R&D expenses were $2.43 billion for the third quarter of 2025, representing an increase of 48.7% year-over-year and an increase of 10.1% quarter-over-quarter. The year-over-year and quarter-over-quarter increases were mainly due to higher expenses related to the development of new vehicle models and technologies as the company expanded its product portfolio to support future growth. SG&A expenses were $2.49 billion for the third quarter of 2025, representing an increase of 52.6% year-over-year and an increase of 15% quarter-over-quarter. The year-over-year and quarter-over-quarter increases were primarily due to higher commission to the franchise stores driven by higher sales volume, as well as higher marketing and advertising expenses. As a result of the foregoing, loss from operations was $0.75 billion for the third quarter of 2025, compared with $1.85 billion year-over-year and $0.93 billion quarter-over-quarter. Net loss was $0.38 billion for the third quarter of 2025, compared with $1.81 billion year-over-year and $0.48 billion quarter-over-quarter. As of September 30, 2025, our company had cash and cash equivalents, restricted cash, short-term investments, and time deposits in total of $48.33 billion. To be mindful of the length of the earnings call, I would encourage listeners to refer to our earnings press release for more details on our third quarter 2025 financial results. This concludes our prepared remarks. We'll now open the call to questions. Operator, please go ahead.

Thank you. If you wish to ask a question, please press star then one on your telephone and wait for your name to be announced. If you wish to cancel your request, please press star then two. If you were on a speakerphone, please pick up the handset to ask your question. For the benefit of all participants on today's call, if you wish to ask your question to management in Chinese, please immediately repeat your question in English. For the sake of clarity and order, please ask one question at a time. Management will respond and then feel free to follow up with your next question. The first question today comes from Tim Tsao with Morgan Stanley. Please go ahead.

Hi, I'm Tim from Morgan Stanley.

Thank you for answering my question. And congratulations to the Group for achieving a new high in this quarter. I have two questions. The first question is mainly focused on the physical AI development that Mr. Xiaobang mentioned just now. Because we have seen that the competitive advantage of the past car companies is mainly reflected in the cost, brand, channel, and so on. So my first question is about the physical AI, because in the past, the competitive advantages of other companies were reflected in several aspects like cost, brand, and channels. Just wondering if the management could elaborate a bit more about in what aspects experts' long-term competitive advantage in physical AI will be demonstrated, and how will the company continuously enhance its strength in these areas?

That's my first question.

I think this is a very big question.

I think it's actually two different genes. In the past, I think that most of the production of a car is positioning, adding layers, positioning where the user group is, how to carry out the T01's ability to carry out efficient and quality layers. However, if it is combined with AI and physics, its entire logic is completely different. It will use the perspective of AI to define how the future technology will change. Through full-time self-reliance plus cross-border integration, to carry out a combination of different product capabilities. For example, like Xiaofeng's Ailong-Lanyin machine, in fact, in many fields, it has to be developed to be integrated. So you will see that these are two different genes, producing two different products. I believe that a four-wheeled car will be the next generation of robot in five to ten years. At this time, we will see the logic theory of the ultimate level and the logic theory of integrated innovation will produce different products at the end. So at this time, the software development system is different. The logic of the blockchain is different. The logic of the whole sound law is different. The logic of data avoidance is different. So the final product is just a scale. Another one will also consider the ability of the Internet platform or the Internet ecosystem. I think this is AI will definitely occupy 50% of a car or a machine in a few years.

This time will come soon.

I think this is definitely a big question.

The traditional way for automakers to make money is completely different from the new physical AI model generated kind of business format. They come from different DNAs. Traditionally, older traditional automakers focus on their own positioning and also about how they target their user segments. And then everything boils down to their integration of Tier 1 suppliers and all the other different parts of the supply chain. However, when it comes to a physical AI-generated model, the definition is different. Everything boils down to the definition of the future tech. It involves full-stack technology capability and also cross-domain integration. For example, the launch of our IRON robot is a great example of that. So that's why different DNA is going to generate different products and different growth momentum. In the future, I believe that cars will be a new format of robotics, and it's going to actually come to the real life in the coming 5 to 10 years as the next generation of robotics in our lives. So traditionally, the integration of supply chain is completely different from what we are looking at right now, which is the physical AI technology integration across different domains. It involves software, hardware, and infrastructure upgrades, which will lead to a completely new set of products. As a result, traditionally, software were only a small percentage of traditional car development, Whereas right now, it takes up a large part of new product development. And I believe that when you look at our future development, we are actually going to see more and more physical AI components in the future for car development, over 50%.

And we are going to see that very, very soon. Thank you.

Thank you, Director Xiao.

My second question is about revenue from the collaboration with Volkswagen. So first of all, congratulations on the project wings of Turing Chips at the Volkswagen. So where I know from which quarters the related revenue will start to kick in, and how should we think about the trend of the revenue contribution from the collaboration with Volkswagen in December quarter and the full year 2026? That's my second question. Thank you.

Hi, Tim. This is Charles. So in Q3, we delivered a few key development milestones on time. So you probably have seen that the revenue from the technology collaboration increased significantly quarter over quarter. And we continue to see that there are a few key development milestones to be delivered in Q4. So we believe that the revenue from technical collaboration in Q4 will be expected at the comparable level we see in Q3 2025. And then regarding your question on the Touring SOC, yes, our Touring SOC was selected by Volkswagen for the two B-class vehicles we're joint developing. And we have already started to supply the Touring SOC to some of our partners' pre-production and verification vehicles. So therefore, the revenue, we would expect that the revenue from Touring SoC will start to be recognized in Q4, and probably in a small amount. But however, as our jointly developed vehicle SOP from early next year, we would expect that the revenue from the Touring SoC will ramp up with the sales volume of the two vehicles we jointly developed. In terms of the revenue from the technical collaboration in 2026, and we expect that as long as we can deliver the key milestones that are scheduled in 2026, we would expect that the revenue from the technical collaboration for the full year 2026 would be comparable to that of the revenue we recognized in 2025. So I think looking back, we have demonstrated that we delivered the revenue from commercialization of technology for seven consecutive quarters. And I think we believe that there are still opportunities we would like to explore to commercialize our technology and also, as our CEO Xiaofeng mentioned, and we will reinvest such revenue from the licensing or technical collaboration back into our R&D. Thank you, Tim.

That's great. Thanks for sharing all the details and congratulations again on all the project practice.

Thank you.

Next question comes from Nick Lang with J.P.

Morgan. Please go ahead.

Okay, thank you for letting me speak. I'm Nick from J.P. Morgan. I'm also very excited. Mr. Peng mentioned the goal of 1 million robots in 2030. 我有兩個簡單的問題想再展開一下 對機器人這個業務跟中長期的發展 第一個就是在科技日的時候 阿倫的這個機器人的行走姿勢跟人非常類似 然後想討論一下 小鵬機器人這個技術路線 跟現在市場上的人型機器人 有哪一些差異 可以跟我們深入討論一下 然後小鵬在國內外的機器人的行業 非常多的競品公司 我們的競爭優勢體現在哪裡 My first question is, my two questions actually related to humanoid robot strategy and vision in the long-term. At recent technology day, Shabong demonstrated our first humanoid robot, Iron, which worked really like a human. And can you talk about our technology roadmap and compare it with the comparable peers? And where is our comparative advantage comparing with the peers in the medium and long-term? That's my first question. Thank you.

Because there are a lot of technology companies in this market. To be honest, Xiaofeng's technology and product lines are moving forward according to our expectations. We don't pay attention to the differences in the technology of other companies. I can only talk about Xiaofeng. For example, our product concept is very promising. You will see this time, We have muscles and skin. In fact, we found a very interesting point, that is, when an inheritance has muscles and skin, everyone dares to embrace it. In the past, when we made an inheritance, no one dared to embrace it. So it can really be integrated into our future life and future work in many systems. So the inheritance of Xiaopeng in this generation is the seventh generation we have launched. We will have a real mass production in the eighth generation. In fact, I see that some of the robots in the entire industry are still in the third to fifth generation of children. It is the ability of joints to drive the entire software and hardware, mainly in operation control. So you will see that the positions of many robots are very consistent. I think these robots it is very difficult to enter into commercialization. So, in the few generations behind us, we all think about how to design all-in-one AI to drive all-in-one hardware. So, this time we are really talking about all-in-one and cross-border integration. I think Xiaopeng's car has a lot of advantages in the industry. For example, the support of our physical AI and our AI vehicles have the same effect. For example, we think more about the future. It should be the car rules, or even beyond the car rules. For example, we consider how to carry out business and how to carry out mass production. This is because we have the accumulation of vehicles. Because, for example, our future sales, after-sales services, and generalization have a large effect. I believe Thank you.

Because there are so many robotics companies in the market, to be honest, the technological and product development roadmap and strategy of Xpeng's robotics is moving forward as we expect. According to our own plan, we have paid really little attention to any other differences in the robotics industry to other companies before we launch our own products. Now, when we look at XPeng, for example, our product philosophy is highly theoretical. You can actually, well, it's highly human-like. That is the goal of developing our own humanoid robot. What's interesting about our product is that we realized that when we incorporate muscles and very bionic skin onto our robots, we actually attracted a lot of people to dare to hug him. And this is very, very exciting because traditional robots really were not that attractive and appealing for human beings to give them a hug. In addition to that, we also would like to mention that in the future, I believe that across many aspects of life and work, we are going to see more and more robots that are working alongside us. So for the current generation of Xpeng robots, last time that we launched it, it was actually the seventh generation, and we are going to begin mass production of the eighth generation of our humanoid robots. In fact, when we look at some of the available robotics in the market, I believe that a lot of them are between generation 3 and 5, which is mainly being driven by joints and all the operation of different hardware. And when you look at the operation of hardware and software, you can see that the available products in the market look very similar in the way that they walk and they move. And these kind of robots, I believe, are very, very hard or difficult to commercialize in the end. So in the future generations of our robots, we actually have been thinking about what kind of technological route we should be using. And we have fully integrated, actually, hardware and software driven by integrated AI. So this time... You can see that the robot that we show to the market is based on our full-stack R&D capability and cross-domain integration. I believe that XPeng Motors has many advantages when it comes to our robotics and humanoid robot development. For example, our physical AI resources have a synergy effect with our AI cars. For example, we actually are considering maybe producing higher than car grade performance for our humanoid robots. And also, our thinking logic on how to conduct business and mass production of our humanoid robots is largely driven by our knowledge and industry know-how in the EV industry. For example, when we built the future sales and marketing layout and globalization, there's a lot of synergistic effects that we can enjoy from the existing layout with our car sales. So I believe when it comes to the future robotics development, some companies will still Some of the players will come from automaking industry, and I believe that XPAC will definitely have a first mover advantage in this regard because of the data, the SOCs, and the capability that we have.

Thank you.

好的,谢谢。 我第二个问题也是想在G7这边再深入讨论一下。 那从现在这个时间点来看,Iron从现在到的量产,还有哪个比较重要的关键任务需要achieve? My second question is also related to humanoid robot long-term strategy and operation. And from here to commercialization, What are the key critical milestones that we should be mindful? And from now toward the end of 2026, can you remind us what's the capacity and expected scale of our human robot operations? And also in terms of use case, by say 2030, you mentioned that 2030 we target to deliver one million units. Can we also talk about the use case in the longer term? Thank you.

To be honest, Alan's production, I believe, is the most difficult car I've ever produced. If you compare it to a car, there are a lot of challenges. Alan hopes to be a machine that can communicate with language and do things. He hopes to be a machine that can simply teach him, train him, and he can work. So, in fact, we have a lot of There is no final level of ability. For example, how to make it very stable. The movement or running in various positions is safe and reliable. This requires joints. This requires a lot of control. This requires a lot of speed. For example, how to make it relatively China China China China China China Thank you.

To be honest, irons mass production is probably the most challenging kind of vehicle or product I've ever worked on at Xpeng Motors if I have to make a comparison between mass-producing irons and other cars. because there's still a lot of challenges. For example, our ultimate goal is for it to be easily trained with human language so that it can really help us in various ways. And there's a lot of room for improvement there when it comes to capability integration. For example, if this robot can walk or run in various safe postures, that requires a lot of integration of capability as well. For example, it needs to have all the joints embedded in management and also the full coupling of different wiring, etc. Also, if we need to allow it to have more generalized kind of dexterous hand movements, it will also require a lot of hand-based VLA, which we believe by the beginning of next year will be integrated. We also need to allow it to have that kind of communication and language-based communication capability between the robots and humans. So that also will come from, for example, a lot of VLM and VLT, which is the small brain and large brain kind of modeling capability. But what I'm really excited to share here is that we will start entering the 1.0 stage of our new generation of mass-produced models next month. I believe that in the next 10 months, we'll be able to actually promote the robot development in an orderly manner during mass production.

And I think... That's the first part of my answer.

Thank you. Thank you. including the commercialization scene of the front desk plus the island patrol. In 2026, we hope to start testing in Xiaofeng's own store area, and we will do the whole SDK and teaching training work well. In this way, our cooperative partners can use their commercialization in a very simple way after buying our machine. If you want to say which scene will be used in five years, I think it will be more extensive than we imagined. For example, in the future, our commercial robots may be able to enter the industry by exchanging two hands. So when will the robot be able to enter the family? I think it will definitely be within five years. I think there is a chance. We also hope that through SDK, we can let more long-term scene needs partners be able to come with us to make such a robot ecosystem better. Thank you.

I think the ramp up in robot production capacity is much simpler compared to cars. However, the commercialization of robots is indeed very, very challenging. It requires us to look for really new heights of technology and ultimately achieving more capabilities. Therefore, we hope to initially implement in several commercial scenarios included tour guiding, shopping or retail assistance, etc., In 2026, we hope that we actually can see a lot of our own robots working alongside us at our XPeng stores, campuses for the first stage of field testing. At the same time, we are also opening our SDK to more of our partners so that our partners can easily and simply buy our robots and train them for commercialization purposes. If your question is about future possibilities of scenario application, I think it's going to be even more than you think. For example, for commercialized robots, maybe you can switch their arms and allow them to go into the industrial production scenarios. And when will the robots go into our household setting? I think maybe five years' time. We still have a big chance of achieving that. And I hope that through opening our SDK, we can allow more kind of partners to help us tackle those diverse and long-term scenarios of application so that we can all enjoy a better robotic future and build a better ecosystem.

Thank you.

The next question comes from Ming from Bank of America.

Please go ahead.

谢谢给我提问的机会。 我是Ming的Ming。 然后也是两个问题。 第一个问题,公司为什么选择在2026年选择推出Robotexy? 能否分享一下背后的技术拐点或者是成本的有效降本? 那面向無人駕駛小公司的技術路線跟商業模式與行業中其他L4的友商相比有哪些優勢? 那我翻譯一下這個問題。 Why does XPON choose to launch robot taxi service in 2026? Could you share... your technology inflection point or how fast you lower your cost. And compared to other robot taxi companies in China, what is Xpeng's technology path or business model? What is your advantage?

谢谢,这是我第一个问题。 谢谢明。 在小鹏的研发策略里面有其中的两条, including full-time self-employed and cross-border integration. I believe that in 2026, the launch of RoboTaxi means that Xiaofeng has appeared in multiple self-employed systems. For example, we have increased the configuration of Robo in the current model. For us, this development has reached the point. For example, our VOM can form a new ability for future cars to become robots. For example, our second generation VLA can use a model and different logic to train our four-wheel drive Ultra, or our Lobotaxi, and even our MAX version in the future. So from another perspective, our cross-border integrated Lobotaxi can solve the current Lobotaxi industry. For example, the car is expensive. For example, you can't travel to many places. For example, small roads, small districts, parking lots, For example, there is a series of problems such as the need to scan and deploy radar in advance. So in 2016, Xiaofeng, we hope to launch the L4 commercial model, Lobotaxi's fully shared model, and the robot passenger L4 model, two methods are launched together. I believe that in the future, it will prove faster that Xiaofeng has a huge advantage in this business logic than other L4 companies. I believe. Thank you.

Thank you, Min, for your question. I think that within our R&D strategy, there are two key aspects, which are full-stack self-development and also cross-domain integration. I believe that in 2026, we'll be actually seeing a collection of inflection points within our own development system. For example, we are going to be able to launch our current models into the RoboTaxi configuration of fleets, which by that time, we believe that the inflection point will arrive. At the same time, our VLM models will continue to offer new capabilities for our future vehicles to be more robotic-like. In addition to that, our current second-generation VLA can actually train our intelligent driving alter cars and also in the future maybe also train our mass version of cars using the same kind of large model too. In other words, we have our cross-domain capability based on our robotic development which really can solve a lot of Robotexy current limitations. For example, the high cost of production and also the limitation of the mobility destinations For example, current mobile taxi now cannot really handle very complicated and complex road conditions and also in residential areas that has a lot of an unpredicted and also a lot of them currently require LiDAR for their perception capability and so on. So in 2026, we hope that by commercializing fully shared L4 capability in our RoboTaxi, we actually can have the dual development of the driverless L4 model together with an assisted driving L4 model. With the launch of both method or roadmap, In the future, I think very soon it will be proven that Xpeng has actually a better commercial logic thinking compared to other Robotaxi companies, and that will give us a great competitive advantage.

Thank you.

那我第二個問題還是跟Robotaxi有關。 那管理層如何思考Robotaxi之後的商業化,然後包括... 我们未来有哪些里程碑 比如说我们的fee number 或者是我们大概的一个出海的时间点 那公司已经宣布跟高德的合作 那可不可以多讲一下之后如何展开 那后续是否计划跟更多的伙伴展开合作 那我也翻译一下第二个问题 So how does the management team think about the commercialization of your robot taxi? business, especially in the future, what is your planned milestone, for example, like in terms of the number of fleet, or when will you plan to roll out in different cities or overseas market? And also, currently, you already have a cooperation with GoldenMap, and could you elaborate more about your cooperation? And in the future, Do you plan to cooperate with more partners like other ride-hailing companies? Thank you.

明年小鸿交会推出三款不同价位Lobotaxi的车型 支持不同类型的出行需求 实际上在下个阶段 我相信在监管许可的前提下 那么我们首先要跑通Lobotaxi的 . . . . . We will quickly establish more robotic ecosystems. This is one of our thoughts.

Thank you. Actually, next year, Xpeng is going to launch three different types of robotaxi models at different price points to support different mobility purposes and demands. And the next phase of development, I believe, with the premise of regulatory approval, our priority is to really get everything running smoothly. when it comes to the whole technological and operation and business model. So in that scenario, we hope to work with more and more business partners in the ecosystem. For example, AMAP will be a great partner. they are going to give us more development support when it comes to traffic and also payment and operation and services, etc. That really set us apart from a lot of the autonomous driving OEMs. And I believe that in the future, for different countries and regions and different steps of development, We are going to actually launch more partnerships with different service providers across different links. And for XPang, what we need to do is that we are building our toolbox really well, and we are opening up our interface capabilities so that we can work more with our ecosystem partners in the future across different countries and cities. And so once we really get everything up and running commercially in different environments, we can then quickly build our ecosystem.

This is one of our considerations. Thank you.

Thank you. That's all my questions. The next question comes from Tina Hu with Goldman Sachs. Please go ahead.

Thank you for accepting my question.

My first question is that we want to ask So I want to see if there is a general income scale. 然後包括就是人心機器人這邊小彭總講到2030年的這個100萬台, 想要看一下有沒有就是有哪幾類就是應用場景, 然後每種應用場景這100萬台大致怎麼分配, 然後到那個時候人心機器人咱們認為的這個售價以及成本大概是什麼樣子的? Let me translate my first question. So first, I would like to understand, over the next one to three years, do we have a rough revenue estimate or breakdown for our new businesses, including Robotaxi, Humanoid, Robot, as well as eVTOL? Thank you.

Hey Tina, it's Brian. First of all, I would say that for these future development areas, we do not provide any numerical guidance at the moment. Clearly, all those three areas, we anticipate volume, scale, level production, and operation in the next four months. For example, the land aircraft carrier from our flying car company is aimed to be delivered to end customers before the end of next year will be in a volume also scale, which I would say in the thousands of range. But the other two, for example, the humanoid robot as well as autonomous driving robot taxi, as we just discussed earlier, next year will be actually a year we'll see a lot of operational testing as well as scaling up a process to make them ready for large quantity production and use. So I would say the contribution from next year will probably be limited, but I think the volume will expect to ramp up rapidly once the model and the stability of these products is proven in the use consumer end as well as the application end. So the long-term goal of having one million per year human and robot sort of sales by 2030 is our long-term goal. And that is something that we have good confidence given. We see the quick ramp up in terms of technology as well as multiple application areas in home, in offices, in factory settings. So with all these future areas, we believe the potential is immense. So at this moment, unfortunately, I cannot give you the exact breakdown as well as precise cost estimates because these are still, I would say, evolving. But I think the overall trend is very exciting for us.

Thank you. So my second question is regarding our passenger vehicles. So wondering if we can get more details on the newly launched new models, their segment as well as price segment. both in the domestic market as well as overseas, and also do we have a volume target for 2026? Thank you.

Hi, Tina. It's Charles here. I think we believe that one chassis to powertrain vehicles present very attractive opportunities. It is also one of our strategic initiatives to expand the volume and the time of each of our vehicles. So I think on November 20th, we are launching the X9 with pricing. That will be our first, we call it the super UREV product to be launched. And then you probably also have noticed that we already have three existing vehicles, super electric model already registered with the regulators. And we plan to launch those three products in early 2026. As Xiaopeng also mentioned that we have four vehicles, four new vehicles when we launch. It will be equipped with both BEV as well as the EREV power chain options and those four new vehicles are positioned in the different various segment, various pricing segment we're in. And we believe that that will continue to enhance our product portfolio in each of the price segment we're targeting. So in terms of the growth into next year, and we believe that the one chassis dual powertrain vehicle models The seven models will significantly drive our growth next year. And also, another growth driver we have seen is that the international market will continue to be a major growth driver for us. With our current products available in the international market, we have already hit 5,000 per month for September and also October, the two consecutive months already. And of the seven new vehicles we're launching next year, three of them, at least three of them will go to international market. And so we're confident that the international market volume will continue to be a very important growth drive for us into 2026.

Okay, thank you. Thank you very much. The next question comes from . Please go ahead. Hello, everyone.

Thank you for your question. I have two questions. The first one is about the car model of Zengcheng. The Zengcheng model of Zengcheng will be launched very quickly from next year. Mr. He also mentioned that in some northern markets, we have seen the development and comparison structure of Zengcheng. 我想请教的第一个问题就是说我们可能看明年来说我们怎么看这个增长车型对于这个小鹏本品这边的一个销售的增长一个带动这是第一个这个问题。 My second question is still about the robot. Because the scale effect on the car is very significant. In fact, the car may have been from 300,000 to 500,000 in the beginning. After all this scale, it has achieved a price of 100,000 to 200,000, even lower. In fact, we also used a lot of new technology on our robot. These new technologies may be a small amount at the beginning, but the cost is relatively high. I would like to ask, for example, at what scale can we see a scale discount? Or in terms of affordability, is it possible for our iron robot to achieve a price that can be accepted at the level of a 2C family? For example, the price may be around 200,000 yuan, or there is still a chance to do it lower. Thank you. I have two questions. My first question is about the new EREV model. What do we think about the growth potential of our new EREV models in 2026? My second question is about the humanoid robots. How do we think about the fuel economy of the humanoid robots? I think we have implemented some new technologies, for example, the solid state batteries and et cetera. And in terms of affordability, will every robot be affordable for a family, say, like 200 kRMB or even less?

Thank you.

Thank you.

The first question is about the super-thin layer. From the data of our current X9, in fact, there is a big difference between the pure electricity and the thin layer. So we are still quite happy. We believe that when X9 is delivered in the super-thin layer, it will grow several times more than the original pure electricity. Then these user groups come from different user groups in different areas. They have different scenarios. This is a very interesting phenomenon. We believe that when we look at other companies, they also have similar situations in terms of pure electricity plus hybrid or pure electricity plus true power. This is what I want to share. But in the case of different cars, I think the proportion of pure electricity and true power will be different. I think the proportion of true power in large cars will be higher, and in small cars, especially in A-class cars, I think it will be relatively high. First of all, regarding the first question, I think what's interesting that we discovered from the sales figures that we gathered since the launch of X9 was that the

Targeted customers and also the actual users of BEV and EREV are quite different. So we believe that we can expect to actually see several times of quarter-over-quarter growth when the new version of X9 actually gets delivered. And actually different customer groups, when they purchase BEV versus EREV, they are using the cars across different scenarios as well. Specifically, what I want to share is that obviously BEV and EREV users in different sizes or scale of cars are also different. In larger vehicles, the percentage of ER EV adoption is higher, whereas for Class A vehicles, especially smaller passenger vehicles, BEV ratio is actually higher. So I think we'll have to wait for more numbers to show, maybe by Q4 and also Q1 next year, before we actually can give you a more concrete answer.

Thank you. The second question, I think Xiaopeng is exploring this.

Last year and this year, Xiaopeng's BOM in the 6th and 7th generations is still very high. But after we started preparing the real mass production in the first half of this year, we think that the cost of this machine will eventually be close to the BOM of the car. So I also think that In the future, the price of robots and cars will be similar. Secondly, I think there is a big difference. In the field of cars, sometimes it is based on the weight. For example, how much iron is in the ground. But I think there is a big difference in the field of robots. In the past, cars may only account for 10% to 20%, but robots may account for 50% at the beginning. So, in another perspective, the cost of software training, or the investment of software's overall storage capabilities. For example, how do we put four SoCs into a super storage to make it as light, small, and cheap as possible? These are all huge challenges. So, from a different perspective, we are very hopeful that in the future, And the second part of your question, regarding the pricing affordability of robotics

I think, first of all, the pricing logic is very different between cars and robots. When we look at the BOM costs of our Gen 6 and Gen 7 robots, they remained very high last year. But by first half of this year, when we were preparing for true mass production, we actually have enough reasons for us to actually believe that the future retail sales price of the robots can be very similar to car prices. And the second point that I want to mention here is that the traditional way of pricing a car is weight-based. It involves how many kind of iron and lithium and all kinds of elements included and components included in making a car, whereas robots, it's very different because robots The percentage of software in a robot is over 50% since day one, whereas the number is only 10% to 20% for a lot of cars. In other ways, you have to put in a lot of cost to train the software and the model, and you need to have the overall capability to do a lot of integration and also domain controller as well. For example, you need to be able to combine all four SOCs into a super complex domain controllers so that you can make them as light as possible and as affordable as possible. These remain very challenging for many industry players. In other words, we really have high hopes for our future when it comes to robotics development. Hopefully, we expect to handle a limited amount of SKU integration, not as many SKU as when you're making a cart. And we also will try our best to make the pricing of robots as affordable as possible so it really can truly help and empower thousands of households in the future.

Thank you. The next question comes from Xiao Yue with Jeffrey.

Please go ahead.

Thank you for taking my question. I have just one question. Could you please provide an update on the progress of our overseas localized production for next year? And additionally, how do we plan to leverage our smart driving capabilities to drive the sales growth in international markets?

Hi, it's Brian again. Just to address your question on overseas plan for next year, you're right. We actually initiated our local production. This happened. second half of this year with first factory in Indonesia and also another factory production facility with partnership with Magna in Austria. Those, I think, is slowly ramping up the capacity, so we anticipate the volume for next year's production in these two plants will continue to rise and support our overall overseas growth. I think in Europe, we are looking at the tens of thousands in terms of numbers of vehicles locally produced there. And in Indonesia, I think probably a smaller but also a sizable number, high thousands, is something that we want to achieve. Looking beyond those two plants, we continue to look at additional opportunities to have local capabilities in other markets. as well as building local supply chain capabilities to support the localization in these key regions. So we will be increasing our local content, increasing our local source materials, and also looking for a further localization strategy to be implemented. So that's something I think is ongoing. I think it's a must-do for a company that has global ambitions Looking at the global product sales next year, I think as Charles mentioned, we're looking for higher growth in the international markets compared to our domestic market. We're also looking for higher contribution economically from those markets. So I would say next year or the year beyond, we're looking at a faster growing, higher growth profit contribution for our international businesses.

Thank you. That's very helpful. Since there are no further questions, I'd like to turn the call back over to the company for any closing remarks.

Thank you once again for joining us today. If you have further questions, please feel free to contact Expense Investor Relations through the contact information provided on our website or the PF and Financial Communications.

This concludes today's conference call. You may now disconnect your line. Thank you.