XPeng Inc.

I will now turn the call over to your host, Mr. Alex Z, Head of Investor Relations, Capital Markets of the company. Please go ahead, Alex.

Thank you. Hello, everyone, and welcome to Xpeng's fourth quarter and fiscal year 2024 earnings conference call. Our financial and operating results were issued via our newswire services earlier today and available online. You can also view the earnings press release by visiting the IR section of our website at ir.xpeng.com. Participants on today's call from our management will include co-founder, chairman, and CEO, Mr. He Xiaopeng, Vice Chairman and President, Dr. Brian Gu, Vice President of Corporate Finance and BW Projects, Mr. Charles Zhang, Vice President of Finance and Accounting, Mr. James Wu, and myself. Management will begin with prepared remarks and 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 note that today's discussion will contain forward-looking statements made under the safe harbor provisions of the U.S. Private Securities Litigation Reform Act of 1995. Overlooking statements involve inherent risks and uncertainties. As such, the company's results may be materially different from the views expressed today, so the 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 price release and this conference call include the disclosure of unaudited GAAP financial measures as well as unaudited non-GAAP financial measures. EXPON's earnings price 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.

大家好,我是何晓鹏。 在2024年的第四季度,我们达成了过项的关键目标, 交付量达到了91507台的新高。 我们的汽车的毛利率连续六个季度改善, And in the second half of the year, the free-throw limit was exceeded by 4 billion. I believe this is a fully verified example of Xiaopeng's transformation in terms of size, systematization, and strategic driving capabilities. In 2025, we will become the champion of the new four-wheel drive in two consecutive months. It will also become one of the most stable cars. I believe that stability can support us. This year, Xiaopeng will focus on further improving Hello everyone, I'm He Xiaopeng.

In the fourth quarter of 2024, we achieved significant milestones, including a record 91,507 vehicle deliveries in a single quarter, six consecutive quarters of improving vehicle margins, and over 4 billion RMB in free cash flow in the second half of the year. These results highlight our significant progress in scaling, organizational improvement, and strategic execution. As we move into 2025, Xpeng has topped monthly deliveries among China's emerging EV companies for two consecutive months, demonstrating our robust momentum. I believe that the key to achieving long-term goals lies in steady progress. This year, I'll dedicate myself to further enhancing Xpeng's organizational capacity, user-centric product development, AI and automotive technology, commercialization and globalization.

那么在上一个季报跟AI的科技日, 我分享了未来的十年的汽车行业是AI的时代。 In 2025, the AI model will bring a subversive experience for autonomous driving. And today, in the first three months of this year, in China, we have seen breakthrough progress in AI technology. It makes me believe that Xiaopeng's strategy and continued route in the AI model of the entire physical world is correct. We have trained hundreds of billions of parameters to scale the cloud-based machine-made foundation model. We have used hundreds of billions of kilometers of I believe this is a model with no potential in the field of automatic driving in China. In addition, we will enhance the learning, improve the ability of the model, and the ability to deal with the long-term scene. By steaming, cutting, and quantifying technology, we will improve the accuracy of the entire model by more than one times.

In our last earnings call and AI Tech Day, I emphasize that the next decade in the automaking industry will be the era of AI, and by 2025, advanced AI models will revolutionize autonomous driving. The AI advancement led by China in the first quarter of this year has only strengthened my conviction. I'm now more confident than ever in X-PANX strategy and technology roadmap for AI large model in the physical world. We've developed a cloud-based foundational model that boasts tens of billions of parameters, leveraging over 100 million kilometers of real driving data for pre-training. This marks an unprecedented scale of model parameters and data volume in the field of autonomous driving in China. To further enhance the model's generalization capabilities and its ability to manage long-tail scenarios, we employ reinforcement learning techniques. We then implement the model in vehicles applying distillation, pruning, and quantization techniques, and have successfully boosted the model's accuracy by over 100%.

今年1月份的时候,中国的DeepSeq R1 在数字世界的AI大元模型是取得了非常大的成果。 那么很有趣的是,这跟小鹏在自动驾驶的技术路线 多个角度是不谋而合的。 We can see that big models are interlocking in different industries, and children are exploring the boundaries of these technologies with AI technology companies in different fields. This makes me very excited. I believe that AI will drive huge changes in the next few years in the automotive industry, including the acceleration of L3 and even the unmanned driving of L4. At the same time, I also see the development of autonomous vehicles and the integration of giant intelligence. In January 2025, DeepSeek R1 makes significant strides in large language models, advancing AI capabilities in the digital realm.

Much of its technology roadmap resonates with ours. I'm thrilled to see that large models are integrating in various industries, and XPeng is exploring the frontier of technology together with leading AI tech companies from different fields. I truly believe AI will transform the automotive industry in a few years' time, speeding up the development of L3 and L4 autonomous driving. Additionally, I see a growing trend of smart vehicles and embodied AI evolving together and smart vehicles may pave the way for humanoid robots. I'm confident that all future smart cars will be embodied AI vehicles. My goal for this decade is to build Xpeng into an AI-empowered global automotive company and a smart technology brand cherished by users worldwide.

我们正处在从汽车的电动化时代过渡到AI智能化时代的周期。 So if this year is the original year of the application of AI, then the AI vehicles developed by Xiaopeng Automotive and the energy generation, I believe, are the two biggest applications of AI in the physical world. I think the leading AI companies need to have from cloud model to terminal model, to whole vehicle EEA and architecture, to chip, to the new generation of software hardware that data must be returned, and it must be added. So this year, Xiaopeng will scale up the production of Turing chips. The current progress is very smooth. We will also apply it to cars, flying cars, and energy savings. It can also be used to quickly deploy the global scene. So Turing chips, coupled with full-time software support, will allow Xiaopeng to fully play out the advantages of Scanner. We believe that not only will our model structure lead the industry, but also the effective parameter scale of the car end will be ten times higher in the industry. This will make Xiaopeng and his opponents take a significant step forward in the next step. I believe that in the second half of this year, Xiaopeng will first practice the software capability and experience of L3-level autonomous driving in China, and then the L3-level autonomous driving will make users have a stronger demand and a higher age. I believe that when autonomous driving makes a large number of family users trust it, when our parents can also use it at ease and can use it at a high level, it will definitely be the real success of autonomous driving.

We're currently undergoing a transformational shift from the era of vehicle electrification to the era of AI. If 2025 marks the beginning of the AI application, then Xpeng's in-house developed AI vehicles and humanoid robots will be two of the most significant applications of AI in the physical world. I believe that a top AI company should possess cutting-edge full-stack software and hardware capabilities and be able to iterate closed-loop data effectively across large cloud-based AI models in vehicle models, vehicle electronic architecture, or EEA, and system-on-chips, or SOC. We expect to begin mass production of our self-developed Turing chip, which, by the way, is progressing really well for vehicles, flying cars, and humanoid robots this year. This chip is also designed for swift deployment across global markets and diverse scenarios. The Turing chip, Combined with our full-stack in-house developed software and hardware, we'll enable XPeng to fully leverage scaling law of our vehicles. Our model architecture will not only remain a generation ahead of the industry, but the effective parameters used in our vehicles will also be dozens of times larger than those of our competitors, creating a significant gap between us and them. I believe that in the second half of this year, XPeng will be the first in China to offer a smart driving experience that boasts software capabilities and user experience equivalent to L3 autonomous driving. L3 autonomy will generate strong user demand and exceptional user loyalty, marking an iPhone 4 moment for AI-defined cars. Additionally, in 2026, We plan to mass-produce vehicle models that will support L4 autonomous driving in low-speed scenarios. True success will be achieved when smart driving earns the trust of a broad range of family users, so even those from my parents' generation can use it confidently and frequently.

一个非常有趣的数据,我们会看到从今年开始,在中国, The high-level process self-driving will be the first to exceed 10% in China. This means that there will be a turning point in the demand for self-driving. The growth curve will be non-linear. This will give the entire self-driving a huge growth opportunity. Xiaopeng's technology, cost, and vehicle layout are all in the process of quickly expanding the sales scale of AI devices and promoting the preparation of AI self-driving to the global market. From a technical point of view, we believe that Xiaopeng and Tesla are the only two in the world that have the ability to rely on high-definition maps, not rely on high-definition data, and use a set of software to realize AI technology that is easy to use in many models around the world. Next, we look forward to the best AI-based enterprises in China and the global market to compete with Taiwan to promote AI-based acceleration. Xiaopeng started with P7 plus all new products, including the newly launched 2015 G6 and G9, will all be equipped with industry-leading pure visual AI nails, and will be available in all Western brands. No need to worry. Then, in the upcoming Mora M03 Max model, we will also first reduce the level of the dust-resistant nails to 150,000, so that more young users can easily enjoy the entertainment of technology. I believe that through the most advanced AI foundation and the most extreme

An interesting figure to share with you is that starting from 2025, the penetration rate of advanced urban smart driving among new car buyers in China will exceed 10% for the first time. marking a turning point in the demand for smart driving. This growth trend is expected to be non-linear, presenting a significant opportunity for us. We're well prepared in terms of technology, cost, and model lineup to rapidly ramp up the sales of AI vehicles and promote the global adoption of AI-powered smart driving. From a technology standpoint, XPeng and Tesla are the only two companies worldwide that are capable of providing globally reliable AI smart driving experience without relying on high-definition maps or LIDAR and using a single software suite across the vehicle models. Going forward, we look forward to competing alongside the world's leading AI autonomous driving companies in both the Chinese and global markets, jointly accelerating the penetration and adoption of AI-driven autonomous driving technologies. Starting from the P7+, all new X-Ping models, including the recently launched 2025 G6 and G9, will come with industry-leading pure vision-based AI smart driving as a standard feature across all trims. No extra costs or add-ons. Our upcoming Mona M03 Max, set to launch in May, will lower the entry barrier for urban AI smart driving to the 150,000 RMB range for the first time, making X-Ping advanced vehicle technology more accessible to younger consumers. By integrating cutting-edge AI technology with a well-developed and highly efficient platform-based approach, Xpeng is democratizing AI technology across the widest price range, making it affordable and reliable for a broad base of ordinary users.

在整个AI智能化加速的同时,在未来三年,小鹏将 based on the product type of positioning, continue to do gold, do difference, and do globalization. So through these three years of product cycle, achieve high-speed sales growth, the scale is leading, and to achieve sustainable profit. In 2025, we will have new models and remodels in every quarter, including the latest generation of AI technology-in-one-model new models. Then on March 13, The launch of the G6 and G9 of the 2025 model is a product that has been upgraded to be sold by the customer. The order continues to climb beyond our original plan. At the same time, what makes us very happy is the outbreak of the supply chain after the launch of the new model. On the other hand, it began to drive us to a big change in the demand for all-terrain vehicles. We plan to launch a new G6 delivery on March 21, and we are currently working with our supply chain. to fully enhance its capacity, so that more users can access it as soon as possible. Next, the Mola, M03, G7, X9, and others will all receive a new version of Max. I believe that through a series of combinations, including M03, G6, PG+, G7, G9, X9, and others, a strong product combination will completely cover users from young users, family users, and quality-improved users,

As the AI boom accelerates over the next three years, we'll focus on refining, differentiating, and globalizing our well-defined product portfolio. This will enable us to achieve rapid sales growth and scale leadership through our three-year product cycle, while also ensuring sustainable profitability. Starting in 2025, we plan to launch new models or updated versions every quarter, all equipped with Xpeng's latest next-gen AI technologies. On March 13, we launched a new 2025 G6 and G9. Their comprehensive upgrades and top-tier product capabilities have been widely recognized and appreciated by our users. Orders continue to pile up and have exceeded our expectations. Additionally, The surge in store visits following the new model launches has driven an increase in demand for our entire lineup. We plan to begin deliveries of the new G6 on March 21st, and we're currently working closely with our supply chain to ramp up production capacity, ensuring that more customers can receive their vehicles as soon as possible. Next, our Mona M03, G7, and X9 will introduce new MAX trains and launch new models or facelift. I'm confident that our strong product offerings, represented by the M03, G6, G7, P7+, G9, and X9, will effectively cater to the diverse needs of young consumers, families, and those seeking premium upgrades, ultimately helping us increase our market share across diverse market segments.

With the rise of new models and remodeled models, I believe Xiaopeng's sales will be stable in the second quarter, beyond the period of product change, continue to grow and re-create the new stage. In the second half of this year, we also plan to deliver many new products, including pure electric and Kunpeng super electric products. Most of our models will support one car, two people. Through super electric products, we can expand our market size. By the end of 2026, Xiaopeng will be in line at a price range of 100,000 to 500,000 yuan, from compact to large, and have a more complete product layout than the mainstream market. I also want to report a good news to everyone. Our pre-sale noise control center in Shanghai has been put into use and is expanding significantly. Xiaopeng's left-handed art and right-handed technology combination is in a bigger crisis. I believe Xiaopeng's new model including in terms of internal and external shape and texture. In the future, it will give users a bigger surprise. In addition, our new development will insist on extreme platforming, platforming various bottom-up technologies such as AI frame, AI workbench, electrical appliances, architecture, AI power, and so on, in order to achieve more efficient development, delivery, quality, and more powerful scale effect.

With the deliveries of our new models and updated versions ramping up, I expect ourselves to steadily navigate a transitional period of product generational change, continue to grow, and reach new highs in the second quarter. In the second half of the year, we plan to deliver several new products, including battery electric vehicles, or BEVs, and quim-pum super electric vehicles. Most of our models will feature dual energy or power, significantly broadening our addressable market through super electric models. By the end of 2026, we'll have a comprehensive product lineup across the 100,000 to 500,000 RMB price range, covering all mainstream segments from compact to large models. We are also thrilled to report that our design center in Shanghai is now operational, and the team is growing rapidly. The integration of art and technology is gradually unleashing its full potential. I believe our new models will impress users with their aesthetics and refined perception. Additionally, the development of our new models will adhere to a thorough platform-based approach, sharing a unified set of underlying technologies, including AI-driven smart driving, AI Cockpit, EEA, and AI Power Powertrain. This will enable efficient R&D advancement and strong economies of scale.

Our growth in the overseas market is also strong. And we see that the potential of electrification and intelligentization has a huge space. In 2024, Xiaopeng's overseas sales broke 20,000. It is the first in China's new car market. It is also the first in China's high-end pure electricity export. I am very grateful to friends around the world for their support and love for Xiaopeng. Then in 2025, our overseas business goal will be realized. is to achieve a sales growth of more than half, and to establish more than 300 sales and service points worldwide. Compared to the end of 2024, it will double. We not only want to sell cars to the world, but more importantly, we want to be a representative of high-tech and high-quality. In 2025, Xiaopeng will establish a new R&D center overseas. We will conduct R&D and testing overseas with the help of Tulin AI. We hope that in the case of local French permission,

Our growth momentum in overseas markets is also strong, and there is still significant potential for both electrification and smart technologies. In 2024, XPeng's overseas vehicle sales exceeded 20,000 units. ranking first in export volume among China's emerging EV brands and also leading in the export of China's new to high-end BEV. I want to express my heartfelt gratitude to all Xpeng owners around the world for their unwavering support and passion for our brand. Looking ahead to 2025, our overseas business milestones include doubling our sales and establishing over 300 sales and service stores worldwide, double the number we had at the end of 2024. Beyond expanding our vehicle sales globally, we're committed to establishing ourselves as a leader in advanced technology and high quality. In 2025, we'll set up new overseas R&D center and commence our overseas research and testing for our Turing AI power smart driving. Our goal is to bring XPeng's AI smart driving capabilities to more countries and users around the world in 2026 and beyond, also ensuring compliance with local regulations.

通过推出更多竞争力领先的产品和全球扩散, 我相信小公司在2025的总销量会相比2024实现翻倍以上的增长, 并且大幅度扩展我们在中国跟全球 in the smart electric vehicle industry. It has become the leading brand of AI vehicles widely recognized by consumers. At the same time, through the improvement of the overall business capability of the past, present and future, continuous improvement of continuity, and a larger scale response, we expect that the interest rate of vehicles will continue to improve this year. We are very confident to believe that at the same time as we insist on big investment in AI research,

With the launch of more competitive products and ongoing global expansion, I'm confident that XPeng's total sales in 2025 will more than double that of 2024. This will significantly increase our market share in both the Chinese and global smart EV sectors and establish us as a leading AI automotive brand widely recognized by consumers. Alongside our growth, Ongoing technology-driven cost reductions and economies of scale will help further improve our vehicle margins in 2025. This allows us to continue our investment in AI research and development while moving towards profitability in the fourth quarter.

从长点来看,我认为AI将会从助手变成社会的一个基础设施,成为我们生活中间不可或缺的一个部分。 我期待在未来城市里面知道无论驾驶、 and cross-city transportation to use flying vehicles, and in factories and small areas, to enter the era of man-made robots, to build a future-oriented ecosystem with AI as the core. So in my plan, Xiaofeng has three growth curves. The first is AI plus vehicles. The second is to go from China to the world. The third is to create a deeper fusion of man-made robots and the automotive industry. I am confident in leading our team to work hard to cross these three growth curves,

In the long run, I believe AI will evolve from being just an assistant to becoming a crucial part of our social infrastructure, woven into the fabric of our daily lives. When I think about the future of an AI-driven mobility ecosystem, I envision driverless vehicles cruising through cities, flying cars for inner-city travel, and humanoid robots integrated into our factories and communities. Regarding my strategic plan for XPeng, I have identified three primary growth trajectories. First, we're working on AI-powered vehicles, equipping our cars with cutting-edge AI technology. We're focusing on global market expansion, actively broadening our presence beyond China and into international markets. Lastly, we're dedicated to the development of humanoid robots that can seamlessly integrate with the automotive industry. Our amazing team is committed to excelling in each of these areas, putting in the hard work to deliver the best AI-driven products to consumers around the world and to provide greater value overall.

Thank you. 在2025年的第一季度,我们预期总交付量约为91,000至93,000两, 同比上升317%至326%, 而收入预计约为人民币150至157亿元, 同比上升129%至139%。 We anticipate that our total deliveries for the first quarter of 2025 will be between...

91,000 and 93,000 units, reflecting a year-over-year increase of 317% to 326.2%. At the same time, we forecast our total revenue to fall between 15 billion and 15.7 billion RMB, which represents a year-over-year growth of 129% to 139%. Thank you, everyone.

Next, we have our Finance VP, James. 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 fourth quarter of 2024.

Thank you, Xiaopeng. Now let me provide a brief overview of our financial results for the fourth quarter of 2024. I'll reference RMB only in my discussion today, unless otherwise stated. Our total revenues were $16.11 billion for the fourth quarter of 2024, an increase of 23.4% year-over-year and an increase of 59.4% quarter-over-quarter. Revenues from vehicle sales were $14.67 billion for the fourth quarter of 2024, an increase of 20% year-over-year and an increase of 66.8% quarter-over-quarter. The year-over-year and quarter-over-quarter increases were mainly attributable to our higher deliveries. Revenues from services and others were 1.43 billion for the fourth quarter of 2024, representing an increase of 74.4% year-over-year and an increase of 9.7% quarter-over-quarter. The year-over-year increase was mainly attributable to the increased revenue from technical R&D services related to the Volkswagen Group The quarter-over-quarter increase was mainly attributable to the increased revenues of maintenance services and auto financing services. Gross margin was 14.4% for the fourth quarter of 2024, compared with 6.2% for the same period of 2023 and 15.3% for the third quarter of 2024. Vehicle margin was 10% for the fourth quarter of 2024, compared with 4.1% for the same period of 2023 and 8.6% for the third quarter of 2024. The year-over-year and quarter-over-quarter increases were mainly attributable to the cost reduction. R&D expenses were $2.01 billion for the fourth quarter of 2024, representing an increase of 53.4% year-over-year and an increase of 22.9% 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 as the company expanded its product portfolio to support future growth. Past G&A expenses were $2.28 billion for the fourth quarter of 2024, representing an increase of 17.5% year-over-year and an increase of 39.3% quarter-over-quarter. The year-over-year and quarter-over-quarter increases were mainly attributable to the higher commissions paid to the franchise stores driven by our higher sales volume. As a result of the foregoing, loss from operations was $1.56 billion for the fourth quarter of 2024, compared with $2.05 billion year-over-year and $1.85 billion quarter-over-quarter. Net loss was $1.33 billion for the fourth quarter of 2024, compared with $1.35 billion year-over-year and $1.81 billion quarter-over-quarter. As of December 31, 2024, our company had cash and cash equivalents, restricted cash, short-term investments, and time deposits in total of $42 billion. To be mindful of the length of our earnings call, I would encourage listeners to refer to our earnings press release for more details on our fourth quarter earnings and full year 2024 financial results. This concludes our prepared remarks. We'll now open the call to questions. Operator, please go ahead.

If you wish to ask a question, please press star 1 on your telephone and wait for your name to be announced. If you wish to cancel your request, please press star 2. If you're on a speakerphone, please pick up the handset to ask a 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. Your first question comes from Tim Hsiao with Morgan Stanley.

Hi, I'm Morgan Stanley.

Thank you for your question. I have two questions. The first question is about the price. As Mr. Xiaofeng mentioned, since the beginning of the pandemic, we have seen domestic car companies pushing the price. What do you think of the price equality? According to the price calculation software notification, My first question is about the ATOM driving. So with the AD ATOM driving or autopilot functions become commodity after recent AI democratization in China, And has upcoming upgrade to level three at Tom's driving system can really help to widen the gaps between the leaders and the laggers, or will it remain a tight race over time?

That's my first question. Thank you.

Yes, thank you.

I believe in the past few months, we have seen that almost all of the car companies in China are announcing their prices. I think this is a very So I believe the first one is that all users will be able to understand more about their own capabilities. I believe this is a huge change for the entire industry. So in such an ability, in fact, in the past, I believe that self-driving is only valuable to users in high-speed and motor vehicles, but it is not really in the full range of point-to-point, car-to-car. So this is a huge value. So last year, Xiaopeng launched the concept of AI Cheat. standard hardware, including software. We believe that this series will allow future AI vehicles to form more unique capabilities, including AI Agents. So how do we see L3 as a guideline? I believe L3 as a guideline, one is the improvement of technology, including hardware capabilities and software capabilities, including model capabilities. This is, I believe, better than the current high-end L2. I believe it will increase the capability of 10 times to tens of times. This is a very obvious thing to see. The second thing I believe to see, what I just said is that when the high-level value is formed in the whole market at the 10% point, you will see that all users will follow up and choose a car from the top 10 to the top 5 to even the top 3 or even the top 2. So Xiaopeng will be with other car companies First of all, thank you so much for your question.

Indeed, we have seen for the past several months that basically all OEMs in China have made huge announcements of entering the realm of smart driving, which is really exciting because when you look back at 2020 or 2021, People were still talking about the future being just new energy vehicle instead of traditional ICE. And right now, we are actually in a new generation of technology advancement. And I believe that this is going to trigger a wider adoption of smart driving for the whole society. All people, all consumers will become more and more interested interested in understanding what smart driving is and what ADAS really is and what kind of experience it's going to generate and bring forth to our society. And actually, during the past period of time, I think that a lot of the so-called smart driving capability was limited to highway driving and also auto parking, which already achieved a lot of progress. However, since last year, our launch of the so-called AI plus vehicle or AI empowered vehicle, which really allow AI capability to be embodied in the vehicle in the physical world. And it comes as a standard feature in our lineup to have this high-level advanced ADAS capability, not only in the software, but also in the hardware as well. And I believe this series of changes is going to allow AI vehicles to actually become AI agents And that will also bring forth a lot of unique capabilities that people have longed for or have never envisioned to be able to have. And as regards to the second part of your question, which is how do we view L3 capability as a turning point? First of all, I believe that first of all, it represents an advancement in technology, not only in the software, but also in the hardware and the integration of both as well. And also the model capability, large model capability, I believe will be way higher than even the advanced level of L2 right now, maybe by 10 times or even more. And that is definitely very clear and foreseeable. And also another point that I want to mention is that when we look at, as I mentioned in my prepared remarks, that when the whole adoption rate or penetration rate of high-level autonomous driving in the market reaching 10%, it's going to generate a lot of user following up and understanding of, advanced capability of smart driving. And when it comes to making the purchase decision, people are going to more and more prioritize having this advanced level of ADAS capability. I'm going to pause in here, and I'll continue to answer your question.

I think Xiaopeng will make a bigger difference in the next stage and in the industry. I think the first one is in the whole area. This is the ability of the entire self-propelled machine. If our previous self-propelled machine was 100% high, then our new large-model self-propelled machine may be 30,000%. It needs to be able to form a closed loop from the cloud, from the car, from the data, to the chip, to the cost, to the project. This ability allows Xiaopeng to invest more resources, ability, time, and have a greater amount of data to attract circulation. I think the second one is that we are seeing the balance of self-driving, seat, chassis, and power in the car. This is a huge value. This is a formula that needs to be done more than one word, and it can be done better. I think the third one is a very big value, which is that Xiaopeng achieved all the capabilities in the future because he used one platform in the AI car. I believe I believe I believe I believe I believe I believe

And I think in the foreseeable future, you're going to see that XPeng is going to create bigger and bigger gaps between ourselves and our competitors from multiple perspectives. First one, we have the capability to do full stack self-development and we are able to leverage all resources across different domain and different scenarios. If When we look at that, when we claim to have the full stack self-development capability, if it was 100 points, right now we're looking at 30,000 points The reason for that is because we are looking at huge data and also large model. And that kind of smart driving capability is going to require end-to-end closed-loop capability across different domains from the cloud to in-vehicle to data to system-on-chip, etc., and also your engineering and your hardware. And because of our early vision and also early investment and our resources and time and capability that we invested way, I mean, way ahead of our competitors, we're able to create a very significant huge, I mean, first mover advantage. Another big advantage that we're going to have is our capability of doing smart driving combined with our smart cockpit, our platform, our powertrains. And also we're able to integrate not just vertically but horizontally as well, which can also create a lot of value because it requires a lot of cross-domain self-development capabilities. Another area that I want to highlight which is advantageous to ourselves is that when we look at AI plus vehicles, we actually leverage one platform to achieve all the capabilities I just mentioned, which allow us to have higher efficiency of iterating our data and capability in the future when data explode. And another aspect that I want to mention is that when we first envision AI plus vehicle, we already have included not only the Chinese market, but the global market. So when it comes to the deployment of our capability, the efficiency, and the implementation and execution, we're going to have very significant first mover advantage in terms of globalization as well. But the list goes on and on and on. I mean, in summary, I just want to highlight that when it comes to advanced ADAS capability, we are going to create a huge gap between ourselves and our competitors.

Thank you. Thank you, Mr. Xiaofeng, for your sharing. The second question is about the application of flying vehicles. Because of the investment of Xiaofeng in AI applications, in addition to home appliances, I also want to know about the application of flying vehicles. For example, Mr. Xiaofeng just mentioned the humanoid robot and EVOTO. Can flying vehicles land and contribute to revenue? In addition, what are the co-effects of the flying vehicle project and the middle application of the car section? So my second question is about the non-vehicle applications. Are we going to see any meaningful synergy across external investments in AI and applications, i.e. autonomous vehicles, humanoid, and eVolto? And when are we going to see value contribution from non-vehicle products? and whether those non-vehicle projects would lead to a more meaningful increase in R&D and other operating expenses in the following quarters.

That's my second question.

Thank you. Yes, thank you.

You can see that Xiao Peng's long-term vision is to be an explorer in the future. What's more profound is that we hope to bring AI, new energy sources, and the entire movement to explore a new type of travel environment. In the past 10 years, in the field of cars, in the field of flying cars, and in the past 5 years, from the perspective of the development of the energy generation, in fact, in the next 10 to 20 years, maybe the real smart car company, the company of AI cars, it will be a new type of travel vehicle with a huge amount of natural ability, including So you can see that in the entire field, whether it's from . . . . .

thank you indeed as we mentioned our long-term vision for the company is to become an explorer of future mobility and on a deeper level we are actually exploring the future ecosystem of mobility which means that we want to empower new energy together with ai um so that we can create a new way of travel. And that would include not only vehicles but also flying vehicles or humanoid robots. And for the past 10 years, we have been invested in not only smart driving vehicles but also humanoid robots. And in the coming decade or two, we believe that a true intelligent or AI-powered vehicle manufacturer will definitely be a producer or a manufacturer of embodied intelligence, new mobility or vehicle manufacturer. And that includes vehicles, flying cars, or humanoid robots. And in this area, if you have a very early preparation, you can definitely create a lot of synergies across different domains. For example, what we are developing right now is the L3 capability for autonomous driving vehicles And this physical world AI model is definitely compatible with the humanoid robot's L3 capability. The only difference is the data that they collect to train a model, which is different. But across many other domains and aspects, there's a lot of capabilities that can actually be shared. And also when you look across other domains, you know, from other aspects of development from end-to-end AI development or global sales network and distribution network, and even when we look at the manufacturing capability and also capacity network, there's also a lot of synergy that you can have, you know, across different types of mobility. And I believe that by 2026, you are going to see excellent outcome, you know, from our synergy as we continue mentioned in the preparer remark previously, we believe that by 2026, we can actually mass produce our flying cars. And by that time, you'll be able to see that Xpeng will be, you know, the first in China, if not globally, to actually mass produce flying cars. Thank you.

Thank you very much.

Your next question comes from Ning Sunli with Bank of America.

Good evening, everyone. First of all, congratulations on this year's performance. My first question here is still related to self-driving, because I have seen more and more cars investing in self-driving, including more and more people will slowly start to use high-end models, even world models, or VLM, etc. In the current environment, because of AI or deep-seek, will it speed up the training process? Is it possible that the first tier companies and those who are chasing after them, after they have these tools and the first tier technology, can be a little bit more efficient? I also want to ask about the Turing chip. So my first question is still related to autonomous driving, especially right now we are seeing more and more auto companies, they are using an end-to-end model or the world model to train their autonomous driving system. So is it possible that the leading AD companies and also the companies who are behind, is it possible that the technology gap to be narrow? And also for Xpeng, because you will start to use your Turing chip this year, so besides the cost control, how can this chip help you to advance your AD capability?

Thank you. Thank you. Your first question is a bit similar to the first one. I often combine these two questions. I think... On the other hand, after the appearance of high-level autonomous vehicles, I think that the entire car industry, especially the car industry with full-time self-driving capability, will be further and further apart from the previous car industry, including Tier 1 or the initial self-driving capability car industry. The ability to return the data, the ability to develop, the ability to use capital, that is, profit, plus the ability to develop costs, for the ability of the whole station's chip to speak and use. These abilities will lead to the new high-level automation manufacturers to have the ability to recover, rather than breaking through with a single point. It is similar to the digital world. By opening up through deepseek, it can subvert a lot of the original large-scale companies. There is a huge difference. Then at the same time, you just asked . . . . . . . . . Thank you.

Thank you. Actually, the first part of your question sounded similar to some of the previous questions. I'm going to respond to both of your questions at once. First of all, after the emergence of high-level autonomous driving or smart driving, I believe that for any players in the automotive industry, you will have to have the so-called comprehensive full-sex self-development capability. What I mean is that you not only need to have the software capability but also hardware and also your supply chain capability that allow you to work with the top tier suppliers and also have more and more resources. And having that comprehensive capability is going to draw a bigger advantage for you and create bigger gap between ourselves and our competitors. not to mention we have this very, very powerful data closed-loop iteration capability and efficiency. Together with our profitability that we are generating right now and also our economies of scale, I believe it's going to allow us to be even more competitive than ever before. In addition to that, we also have our self-developed chip and the capability to actually deploy it across multiple scenarios in different markets. That is why we are very, very confident in our competitive leadership because newcomers or existing incumbents are not easily surpassing us just because they may do better in one aspect or two. which really is very different than the digital competitive landscape, you know, as you mentioned in your question, such as DeepSeek and other AI software companies. The second part of your question, this is a very interesting question, you know, how the self-developed Turing chip really empowers our development for the future. Let me just give you an example. You know, the reason why so many companies, OEMs or smartphone manufacturers try to develop their own chips is because you will realize not only can you gain cost benefits from developing your own chip, but also it allows you to customize a lot of your products and also have a lot of potential of development as well. And the fact that we are able to self-develop our tooling chip allow us to maintain our leadership effectively. And because we do everything within our own system in the company, you will realize that it also allows you to achieve energy benefits and efficiency and improve your performance of your product significantly. And you also can work with other players and other suppliers more effectively as well. It allows us to have the unparalleled self-development capability to also leverage data and have very effective and fast iteration. That is the reason why we really invest heavily in our self-developed truancy.

Thank you. Thank you, Mr. Peng. The second question is about robots. I remember before, Mr. Peng had already published a robot dog. At that time, Mr. Peng said that he wanted to make a robot dog with very good performance. This is actually more difficult than making a car with very good performance. I would like to know about the progress of humanoid robots. It is expected that in 2026, in addition to mass production, it will also commercialize. So far, in terms of the progress of our research, which aspects do you think are more satisfied? Which ones are still bottlenecks? Maybe the cost is not low enough, or motion control, or some other places. So maybe we still have more than a year. So my second question is related to the humanoid robot product development. So since you just mentioned that you plan to mass production your humanoid robot in 2026, so will you also start to sell the product? next year, and currently, based on your development progress, which area you feel you are satisfied, and what is the area that you think you still need to continue to make more progress before you officially launch it?

Thank you.

You asked me a challenging question, because it concerns the next generation of products. First of all, the mass production of the machine in 2022 will definitely be commercialized. The machine is actually very different from the car. And I think it's more difficult than the AI car. So we see that most of today's machines are very capable of moving with the lower body. In fact, we think it is still based on the ability of L2. We think that real robots should be universal. Real robots need a lot of real data to build the simulation data. We think that the difficulty of robots is higher than that of cars, It's a full-level capability. For example, an AI car doesn't have AI capabilities, but you can drive it. But if a robot doesn't have AI capabilities, you can't use it. We think that for the bot, the bot needs a combination of a series of capabilities such as safety, reliability, stability, and natural language interaction in a large number of scenarios. This requirement is very complex for robots. This is also the reason why Xiaopeng is now doing all-person, even with skin, and close to the number of human joints. This is also the reason why Xiaopeng, from last year's autonomous driving to the physical world, at the same time considered why the structural system can be done until now, and can be done better and better. We are considering the robot and So I think today we are doing a good job, but we also hope to solve all the AI capabilities of L3, including feet, hands, mouth, eyes, ears, and part of the simple big brain agent capability. It is a foundation that can lead to business. Thank you for your question.

This is definitely not an easy question to answer because it involves our next generation of technology and products. Now, when we mentioned the mass production of humanoid robots by 2026, we're definitely talking about the commercial market and large production. But when it comes to the manufacturing of humanoid robots, it's actually more challenging than even producing AI vehicles. The reason for that is because it really requires fully integrated capabilities across the board because a true humanoid robot requires not only the hardware but also real data from our human activities and also simulated data, and it's very hard to achieve this comprehensive level of full integration. Right now, when you look at the available products in the market, a lot of our peers are actually focusing on the lower body motion capability of humanoid robot. And I believe that is still very limiting because it's restricting themselves to the level two or L2 level kind of capability, which is not enough. For example, if you have an AI vehicle, but you don't fully have the AI capability integrated in it, you can still drive it because it's a car, but a humanoid robot without the AI capability, without that L3 level capability is, it's basically useless. So even for the most basic model of the humanoid robot, you still require a lot, including all of the joints that you have for your arm and also the motion capabilities that you need and the agility that you need for a robot, which is not easy to achieve when you set the target of achieving the L3 AI capability. And right now, what I can share with you is that our humanoid robot R&D is progressing very well, and I believe that what we are able to do right now not only is just the basic lower body capability, but also legs, arms, mouth, and also some very simple but necessary brain capability. We call it the agent capability and the combination of basically humanoid or semi-human kind of brain intelligent capability that we incorporated in our humanoid robots. But that is very, very challenging, and I hope that in maybe the near future, I can share more update with you when we have more information. Thank you.

Thank you, everyone. That's all my questions.

So our next question comes from Din Wang with Deutsche Bank.

Thank you. I have two questions. First, I would like to ask the company to give us a little bit of advice. My two questions are all about financial. The first one, can you provide a first quarter gross margin guidance? And the second question is about your financial. We've got actually improvement in the long-term investment. The other one is the fair value games. which may be linked to the DD acquisitions. Thank you.

Hey, Wen, this is James. I'll answer your question. Yes, so what you've seen, first of all, in Q4 is our vehicle margin has improved versus the prior quarter, and this is primarily due to our cost reduction efforts as well as higher scale. As to the forecasting on our... ...to... improve our vehicle margin to stand solidly on the double-digit vehicle margin. We'll continue to from a cost front, and as we continue to improve scale, we would benefit from a manufacturing cost perspective as well. With regard to your second question on the DD-related transaction, yes, there was an adjustment that was made. If you recall, as we closed the transaction with DD, there was an earn-out mechanism that has baked in with DD that is linked to the delivery of vehicles registered in the DD platform for ride-hailing purposes. Now, as we introduced Mona products, we have made decisions internally and obviously aligned with DD that we are going to focus on the 2C customer rather than the 2B customer. And therefore, based on our assessment, the first milestone is unlikely to be met. And therefore, we have released that liability, which is what you've seen in our financials. Thank you.

In other words, long-term investment.

Long-term investment, we had some long-term investment re-evaluations based on our invested companies, and the impact in the fourth quarter is relatively not significant, I would say. That would come from, some would come from, for example, the Hui Tian investment we had as an equity holder, And then there are a variety of other investment plans as well. But overall, the change in the fourth quarter is not as meaningful as what you've seen on the DD transaction release.

Thank you.

Very good.

Thanks. Your next question comes from Tingyu Wu with SciTech Securities.

Thank you. Thank you for the opportunity to ask the question. My first question is about the overseas market. And as we can see, the overseas market is one of the pillars for our long-term development. And are we considering to build a factory or some kind of corporation in terms of capacity overseas? Thank you.

Hi, it's Brian. Let me address this question. Yes, it's correct that I think international expansion is a key pillar of our growth curve, as Xiaofeng articulated in the earlier opening speech. We see there's tremendous demand for our high-quality technology products for a number of overseas markets. In fact, last year, I think we saw a very encouraging robust reception to our market entries into a number of new markets. So by the end of last year, we actually are now selling over 30 countries globally. And also in a number of those countries, we saw our product has been positioned as the top three EV models in the respective categories. So I think so far we are very encouraged by the response to our product and technology. In terms of going forward contribution, I think this year we continue to see increased expansion of our global platform. Our number of stores internationally will double. I think our sales outlets will double from probably roughly 150 to probably over 300 this year. Our number of countries that were selling our product will close to double as well. And also we anticipate close to also about doubling of our international sales as a consequence of all these expansion. We also are indeed invest in teams, invest in international sort of infrastructure. We mentioned that we're going to be building overseas R&D centers this year, which we'll help us to really recruit local talents to focus on international technology development. At the same time, we're also going to look at a number of local potential production solutions. Earlier, I think we have jointly announced with our partner in Indonesia to explore local manufacturing production solutions. We'll be also spending time in other regions, so we will obviously share with the market once we actually have these determined. But we do have a very detailed and also, I would say, aggressive plan to make sure that we are well positioned to enter these international markets with the products, with the brand, and also with the local capabilities.

Thank you, Brian. The second question I want to ask is about the price. Because our goal this year is to achieve a 100-kilometer city take-one project by the end of the year. I would like to ask, for example, for the L2++ function, is MPI equal to 100 a medium-term or medium-term goal? Thank you.

Let me translate the question.

My second question is about autonomous driving. We are going to enlarge the MPI to 100 kilometers by the end of this year. Do we think it is an ultimate goal for us or we aim higher? And also, how do we think of the the possibilities for the OEMs or the insurance company to take some of the responsibilities of the driving systems.

Thank you.

关于自动驾驶多少公里接管一次, 这个是一个在 the ability to move forward in the middle of the debate. How do you understand this? In other words, when your automatic driving has a greater coverage, for example, it can go to the community, to the parking lot, to all the small roads, then your driving response rate may be higher. When your user group is larger, because it will not be like the media, the media is testing its extreme ability, which is not to safety, it generally does not receive control. But a normal user may have a sense of security, may have a need for efficiency to change time. It will work. So from the point of view of our car, it depends on the big data of the car. So today, all the high-speed automatic driving of Level 2 are not far away from 100 kilometers per hour, including some L4 models that we have seen. So L3 is a huge challenge. But I also I believe it's just a foundation in the combination of the next generation of models, the next generation of calculations, and the next generation of data. I think a person's distance to drive in the future, even a month, may be hundreds or even 2,000 to 3,000 kilometers. If you can do it, if you can only take care of it once or twice a week, it will push the value of the entire society higher. Right, thank you for your question.

Now, about the human intervention or takeover per 100 kilometers, I think this is quite an interesting paradox. We're trying to navigate ourselves within this paradox. The reason for that is because, first of all, it depends on the scenarios and the application of the ADAS capability. For example, if you are covering different scenarios, if you are driving within a neighborhood or within a parking lot, the possibility is that you're going to take over more than, let's say, if you were just driving on the highway. And also, it depends on who you are and what your purpose is with the testing and calculation of the human intervention. If you are a member of the media, you're going to test the limit of the ADAS capability, which means that you're not going to take over unless it's absolutely necessary. However, for just an ordinary driver on, for example, daily commute, they might tend to take over more when they feel like it. So it really depends. And as an OEM, we need to really collect a huge amount of data before we can claim that we can achieve, for example, level 2 or level 3 ADAS capability or some sort of milestone. And when you look at a lot of our peers, even when they claim that they have L2++ or even L3, L4 capabilities, their 100 kilometers human intervention rate might actually be quite high. And so it really depends. I think it is part of our next milestone. And even though achieving L3 capability can be challenging, I also believe, I still believe that it's achievable depending on the next generation of model, computing power, and capability and data. And with that combination, I believe that we can achieve actually utilize and leverage ADAS more and more. And in the future, perhaps maybe on a weekly basis or on a monthly basis, we'll be driving maybe hundreds of kilometers or even thousands of kilometers. And if we only have to take over maybe once a week, two weeks or once a month, we definitely will promote the whole penetration rate of high-level intelligent driving for the whole society. And I think this is our goal. And when we mentioned that we need to achieve such a milestone of one takeover per 100 kilometers of mileage or driving range, We're actually talking about some sort of measurement of our milestone, and essentially we're talking about the L3 capability that we expect to achieve, and that we believe and we are confident that we can achieve with our capability as well. And also we believe that we are way ahead of the game, and we have a higher and more advanced capability than industry average. Thank you. And even for that, let me just add that it's only an initial or an early stage of L3 capability. And in the future, maybe we only need to take over every thousand kilometers or tens of thousands of kilometers.

Thank you. 关于您刚刚的第二个部分,就关于安心感的, 关于例如保险,例如中国的L3或者全球的L3, 什么时候的政治法会落地,以及出现安全事故的责任的论定, 实际上这一点小鹏... Thank you.

And about the second part of your questions regarding the safety issues of driving and also regulation and related accountability of traffic accidents, we actually have been exploring that very early on, including our proposal during the two sessions of the MPC recently, we have proposed to actually do more so that we can really push the next generation of regulation that can allow us to have safer environment of driving. And we believe that in this process, China is going to lead the world in becoming a safer environment and having more regulation in this regard. Thank you.

How about 50-60? Your next question comes from Nick Lay with J.P. Morgan.

Okay, thank you. I have two simple questions. The first question is, as Mr. Peng mentioned, there are three important growth drivers in the company. One is AI vehicles, and the other is the overseas market. Let me translate my question very briefly. Chairman mentioned that in the long term, we have three major business drivers, including AI cars, overseas markets, and humanoid robots. Fast forward to, say, 2030, how should we think about the scope of each of these products?

Thanks. Thank you.

It's hard for me to answer your question. In the next decade, that is, by 2034, I hope that half of the sales will come from China and half will come from the world. This is the first. We have a plan for 2030, but we can't talk about it. The second, when will the energy generation reach a huge technical or commercial point of view? Today, I believe no one can give a clear answer. Maybe from the end of 2026 to the end of 2027, when we see that some energy generation can be commercialized and produced, Thank you.

It's very hard to answer your question. What I can only say for now is that in the coming 10 years, meaning by 2030, We hope that half of our cells will come from China and the other half will come from the overseas market. We do have our internal plan and targets for 2030, but unfortunately, I have to keep it confidential. And the second aspect that I want to mention to respond to your question is regarding humanoid robots. It's very hard to predict when the turning point will come technologically and commercially or when it comes to the mass production. We might have more visibility by the end of 2026 or end of 2027 when we saw the emergence of mass-produced humanoid robots. I think right now, for now, things are still very foggy. We're probably at about 2010 or 2011 for new energy vehicles when it comes to comparing humanoid robot development. Thank you.

Okay, very clear. Thank you. My second question is that the EU government announced a new industrial policy two weeks ago. This industrial policy is more related to us, which is to promote smart electric vehicles and encourage foreign investment. In this part, we have cooperated with the public. I don't want to say that under the new EU government policy, we cooperated with the public. The current product structure is in China for China. My second question is really about new policy announced by European government lately on encouraging overseas foreign investment in European car market, especially in AI autonomous driving car. And given we have partnership with Volkswagen, is there any new opportunity we can explore from there?

Thanks.

Hi, Nick.

I think as we laid out our autonomous driving strategy before, I think we're setting up our overseas R&D center, and also we're in the process of bringing our autonomous driving technology to the global customers. I think that right now we're setting up the infrastructure to start the preparation. I think in terms of the collaboration with Volkswagen, Currently, I think that our teams, engineering teams, are working really hard to push for the SOP of the G9 platform collaboration as well as the EEA collaboration. We do believe that there is a potential and also the industrial logic to collaborate in the potential overseas market. Thank you.

That does conclude our question and answer session. I'd like to turn the call back over to the company for closing remarks.

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

That does conclude our conference for today. Thank you for participating. You may disconnect your line.