Ticker: LI

Quarter: Q4 2025

Date: 2026-03-12 00:00:00

Operator: Hello, ladies and gentlemen. Thank you for standing by for Li Auto's Fourth Quarter and Full Year 2025 Earnings Conference Call. [Operator Instructions] Today's conference call is being recorded. I will now turn the call over to your host, Ms. Janet Chang, Investor Relations Director of Li Auto. Please go ahead, Janet.

Janet Chang: Thank you, operator. Good evening, and good morning, everyone. Welcome to Li Auto's Fourth Quarter and Full Year 2025 Earnings Conference Call. The company's financial and operating results were published in our press release earlier today and are posted on the company's IR website. On today's call, we will have our Chairman and CEO, Mr. Xiang Li; and our CFO, Mr. Johnny Tie Li, to begin with prepared remarks. Our President, Mr. Donghui Ma; and our CTO, Mr. Yan Xie, will join for the Q&A discussion. Before we continue, please be reminded 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. Forward-looking statements involve inherent risks and uncertainties. As such, the company's actual results may be materially different from the views expressed today. Further information regarding risks and uncertainties is included in certain company filings with the U.S. Securities and Exchange Commission and the Stock Exchange of Hong Kong Limited. The company does not assume any obligation to update any forward-looking statements, except as required under applicable law. Please also note that Li Auto's earnings press release and this conference call include discussions of unaudited GAAP financial information as well as unaudited non-GAAP financial measures. Please refer to Li Auto's disclosure documents on the IR section of our website, which contain a reconciliation of the unaudited non-GAAP measures to comparable GAAP measures. Our CEO will start his remarks in Chinese. There will be English translation after he finishes all his remarks. With that, I will now turn the call over to our CEO, Mr. Xiang Li. Please go ahead.

Xiang Li: [Interpreted] Now translating from Mr. Li Xiang. Thank you for joining our earnings call today. Over the past year, Li Auto has been going through an important period of strategic adjustments. As we scaled, we've reassessed a number of core capabilities, especially how to sustain sales efficiency and organizational vitality in a direct sales model. The problem that we've identified is that in the past, we have used a dealership mindset to manage our store front. However, for our direct sales model, the key is really managing each storefront. And therefore, without dealers, we need to figure out how to manage the storefronts effectively on our own. Therefore, since the third quarter of last year, we've been focused on improving store rollout quality, strengthening day-to-day store operations and upgrading incentives, training and enablement for our teams. Ultimately, it comes down to one question. How do we sell well? With sales volume and productivity per salesperson as our key metrics, we implemented a series of targeted changes. First, we optimized division of labor and consolidated our sales force. By closing and replacing underperforming locations, we addressed site selection issues and moved sales teams from lower traffic, second-tier malls to higher potential locations, such as prime Tier 1 shopping districts and flagship stores in major auto hubs. This has directly improved store productivity and sales per head. To further enhance our frontline sales, we also upgraded our operating mechanisms. In March, we launched our store partner program, making each store the basic operating unit. Top store managers now have real operating decision power and profit sharing shifting from a pure management role to true store operators. At a time when profitability is challenging across the auto retail industry, we want to develop store managers who can earn over RMB 1 million per year and enable our top performers to make 3x the industry average. Just as importantly, this strengthens our frontline capability and helps keep our orders and deliveries firmly in the top tier of the premium segment. Turning to products. We will officially launch the all-new L9 lineup in the second quarter. With comprehensive upgrades from powertrain, autonomous driving to chassis technology, we aim to create a clear step change in user experience versus competing models and regain leadership in the flagship SUV segment. The new Li L9 will come standard with an 800-volt architecture and 5C ultra-fast charging. It would also feature our next-generation full stack in-house developed range extender 3.0 system, delivering higher generation efficiency and greater output. With further MBH improvements and our proprietary EGR low-temperature start technology, the new Li L9 offers a cabin quietness and driving experience comparable to best along with improved winter energy consumption. We will also debut the world's first AI-powered engine oil maintenance system, enabling long service intervals of up to 3 years or 30,000 kilometers. Our top-of-the-line Li L9 Livis, priced at RMB 559,800, reflects our vision for flagship SUV in a mirror of embodied AI. It will feature the world's first mass-produced fully drive-by-wire chassis along with an 800-volt fully active suspension system, delivering best-in-class comfort and handling. Response speed and safety performance across steering, braking and suspension are also significantly enhanced, providing the execution foundation for autonomous driving and embodied AI. In addition, the Li L9 Livis will be powered by 2 in-house developed 5-nanometer M100 chips, delivering 6x the effective computing power of 4U. Together with our data flow architecture and in-house intelligent driving stack, it enables end-to-end integration of our algorithms and computing platform. The success of the new Li L9 will directly determine the market potential of the entire L Series. If the previous L9's, competitiveness was driven primarily by smart product definition, the new generation L9 will build its core advantage through technology. Our BEV models are also continuing to ramp up. With sustained efforts alongside our suppliers, the supply constraints on the Li i6 have been gradually easing. We will keep increasing capacity to further shorten delivery lead times. Meanwhile, as owners put more miles on their vehicles, positive real-world experiences have driven the Li i8 NPS up by more than 20%. In the NEV Brand Health Study, recently released by Land Roads, the Li i8 also ranked #1 in NPS among all large SUVs, improving experience and satisfaction are now translating into a recovery in order in sales. Since March, Li i8 orders have increased 33% versus the same period in February and 179% versus January. Together, the Li i6, i8 are strengthening the market foundation of our BEV portfolio. Our priority is to fully resolve the issues we encountered earlier to ensure our BEV offerings establish a solid foothold in the market. Looking back at [ DevNet ] last year, the Li i6 faced multiple timing-related headwinds after launch, including initial sales policies, the production ramp and the phaseout of purchase tax subsidies which pressured gross margin. At the same time, these factors also set the stage for our margin improvements this year. In the second half of 2026, we'll be launching the Li i9, a new flagship BEV SUV further expanding our BEV portfolio to meet a wider range of customer needs. 2026 will be a pivotal year in Li Auto's evolution into an embodied AI company. As competition intensifies in the NEV market, we will continue to strengthen our technology moat and complete our transformation from a smart EV company to an embodied AI company, positioning us for the next phase of the competition. In 2025, our R&D spending totaled RMB 11.3 billion, which approximately was 50% was allocated to AI-related initiatives. We will maintain this investment strategy in 2026 as we continue to build the core capabilities required of an embodied AI company. For Li Auto, AI has 2 main dimensions: creating AI and applying AI, bringing our products to people to life while improving efficiency across the organization. On the Create AI side, we have rebuilt our R&D organization from the ground up to operate the way an embodied AI company should. We're developing capabilities and attracting top talents across the interface chips, foundation models, software and hardware. At the product level, we see the vehicle as an intelligent agent with real vitality and AI is what brings that vitality to life. Built on our next-generation technology platform, our products will all evolve in ways you will see over time. There will not be near extensions of traditional cars or EVs. Instead, they will be proactive and increasingly life-like in how they learn and improve, and that will be reflected in our high-frequency experiences in daily lives. From an efficiency standpoint, AI is helping reverse the slowdown in information flow and decision-making that can come with scale. By integrating AI and work alongside agents, where we're gaining the speed and agility of a start-up and our iteration and evolution, and we are already seeing early results of day-to-day operations this year. In other words, AI is not only reshaping our tools, it also is enabling a more dynamic high-velocity organization. In closing, I want to emphasize that we will convert the capabilities and systems we've built from the Li L9 launch in 2022 to today's broader automotive and embodied AI technology stack into a real user experience and measurable business value. This will serve as the cornerstone of our long-term competitive positioning for the next decade. We look forward to your continued attention of welcoming you to experience our next generation of products. With that, we will turn it over the call to our CFO, Johnny, to walk you through our financial performance.

Tie Li: Thank you, Li Xiang. Hello, everyone, given time constraint my remarks today will be limited to fourth quarter financial highlights. All figures will be quoted in RMB unless otherwise stated. For further details, including the corresponding U.S. dollar amounts and full year financial results, we encourage you to refer to our earnings press release. Total revenues in the fourth quarter were RMB 28.8 billion, down 35% year-over-year and up 5.2% quarter-over-quarter. This included RMB 27.3 billion from vehicle sales, down 36.1% year-over-year and up 5.4% quarter-over-quarter. The year-over-year decrease was mainly due to lower vehicle deliveries. The sequential increase was mainly due to the increase in vehicle deliveries, partially offset by lower average selling price due to the different mix following the commencement of the i6 deliveries. Cost of sales in the fourth quarter was RMB 23.6 billion, down 33% year-over-year and up 3.3% quarter-over-quarter. Gross profit in the fourth quarter was RMB 5.1 billion, down 42.8% year-over-year and up 14.8% quarter-over-quarter. Vehicle margin in the fourth quarter was 16.8% versus 19.7% in the same period last year and 15.5% in the prior quarter. The year-over-year decrease was mainly due to different product mix. The sequential increase was mainly due to the estimated Li MEGA recall cost booked in the prior quarter, partially offset by lower average selling price due to different product mix following the commitment of the i6 deliveries. Gross margin in the fourth quarter was 17.8% versus 20.3% in the same period last year and 16.3% in the prior quarter. Operating expenses in the fourth quarter were RMB 5.6 billion or 5.8% year-over-year and down 1.3% quarter-over-quarter. R&D expenses in the fourth quarter were RMB 3 billion or 25.3% year-over-year and 1.4% quarter-over-quarter. The year-over-year increase was mainly due to the cost related to AI and other programs to support product portfolio expansion and technology advancements. SG&A expenses in the fourth quarter were RMB 2.6 billion, down 14% year-over-year and 4.4% quarter-over-quarter. The year-over-year decrease was mainly due to decreased employee compensation. Loss from operations in the fourth quarter was RMB 442.6 million, versus RMB 3.7 billion income from operations in the same period last year and RMB 1.2 billion loss from operations in the prior quarter. Operating margin in the fourth quarter was negative 1.5% versus 8.4% in the same period last year and negative 4.3% in the third quarter. Net income in the fourth quarter was RMB 20.2 million versus RMB 3.5 billion net income in the same period last year and RMB 624.4 million net loss in the prior quarter. Diluted net earnings per ADS attributable to ordinary shareholders was RMB 0.01 in the fourth quarter versus RMB 3.31 diluted net earnings in the same period last year and RMB 0.62 diluted net loss in the prior quarter. Turning to our balance sheet and cash flow. Our cash position remains solid with a year-end balance of RMB 101.2 billion. Net cash provided by operating activities in the fourth quarter was RMB 3.5 billion versus RMB 8.7 billion, provided in the same period last year and RMB 7.4 billion yields in the third quarter. Free cash flow was RMB 2.5 billion in the fourth quarter versus RMB 6.1 billion in the same period last year, a negative RMB 8.9 billion in the prior quarter. At the end of 2025, we had a total of 30,728 employees. And now for our business outlook. For the first quarter of 2026, the company expects the delivery to be between 85,000 and 90,000 vehicles and quarterly total revenue to be between RMB 20.4 billion and RMB 21.6 billion. This business outlook reflects the company's current and preliminary view on the business situation and market condition, which is subject to change. That concludes our prepared remarks. I will now turn the call over to the operator to start our Q&A session. Thank you.

Operator: [Operator Instructions] Your first question comes from Tim Hsiao with Morgan Stanley.

Tim Hsiao: [Interpreted] So my first question is about the channel. I think the management just mentioned that Li Auto now plans to optimize the sales networks and reportedly close up to 100 stores. What are company's plans and progress regarding channel optimization? Separately, could you elaborate a bit more about the store partner mechanism? CEO just mentioned in addition to the incentive program to the store manager, any further implementation details you can share with us? And when should we expect it to show positive results? That's my first question.

Xiang Li: [Interpreted] I need to first start with the clarification on channel optimization. The rumor about closing 100 stores is false. In reality, we've always conducted routine optimization of our stores facing out a small number of underperforming stores that cannot reach their sales targets. And this is simply part of the normal operations to address past issues like core store locations or declining foot traffic in certain shopping districts. Our core channel strategy this year is very clear, quality over quantity. We will add new stores this year. New stores will prioritize top-tier shopping malls and premium auto parks to strengthen brand presence and attract higher quality traffic. As for city coverage, our footprint in lower city tier cities is already fairly complete. And going forward, we will focus on increasing store density in higher-tier cities aligned with our ramp-up of BEV sales. Meanwhile, we continue to improve sales and service experience, covering in-store reception, test drives, delivery, highway supercharging stations, staff with attendance during holidays. As a result, we've been seeing user satisfaction and positive feedback keep climbing. While we are on the topic, let me also share a bit about our new store partner program officially launched on March 1. We're treating each store as a core business unit and building a great sales model that's truly unique to Li Auto. First, we remain fully committed to direct sales. This ensures consistent service quality and a unified national pricing strategy. Meanwhile, we're now delegating decision-making power and sharing profits with store managers to really motivate our frontline teams and enable them to think like real business operators. In terms of store operations, store managers now have autonomy in 3 areas: customer acquisition, day-to-day operations and managing their own teams. And we've changed how we evaluate them as well, no longer just on sales volume. Now the performance measured on operating results of the stores. The goal is to make every store manager feel like they're running their own businesses and be fully accountable for the results. This new model also helps us solve past problems at the root, like opening stores without thinking through and other issues on store expansion. Going forward, store managers will be involved from day-to-day site selection with clear ownership and accountability. This way, we raise store quality right from the start, and the company will back it up with financial support and digital tools to empower our frontline team. We aim to see significant sales and operational improvements from Q3. Ultimately, we believe a healthy, efficient sales and service system is the foundation of strong sales and market leadership. Since last August, we've spent 7 months to systematically recalibrate our direct sales management framework. This includes high-quality store expansion, refined operations and store manager incentives as well as frontline training. That all is to build a truly sustainable competitive sales and service network for the long run.

Tim Hsiao: [Interpreted]. My question is about the product -- the new product for the upcoming all-new L9 and the L9 Livis, could you please shed light on the launch time line, pricing strategy, product competitivity, competitiveness in a crowded race and vehicle profitability?

Xiang Li: [Interpreted] So first, to your question about products. This year, we will be launching -- in Q2, we'll be launching our all-new Li L9 Livis equipped with our in-house developed MAC 100 chips. And the reason we call it an embodied AI robot is really because we've completely revamped the technical stack from a sensing brain and body from all these 3 dimensions. So first of all, let me start with the autonomous driving front. In the past, the technical paradigm is really to drive -- to have the machines to learn to drive through watching videos. They're not really understanding the physical world, but rather they're just watching videos and trying to imitate human behavior from these videos. However, we -- as we conducted our R&D, we identified that the most efficient way is really to understand the physical world. So when we say VLA model, we're really trying to use the language to understand and understand how the world works as opposed to just simply interpreting the video. So as we move from 2-dimensional cameras to 3-dimensional vision transformers, we can really understand the 3D world much better. And that requires a revamp of the full stack from video encoder to the chips to the algorithms, we can -- how we can enable compute directly from a large model to actually enable physical behavior. And I believe this will be a major technological shift across the world, across autonomous driving as well as physical robots happening this year. So as I mentioned earlier, this is going to be a real VLA. The language models really understand and thinks about the physical world before it makes the decision to move. So that's on the sensing side, perception side. And then next on the hardware itself on the execution side, Li L9, as we've released earlier, will be equipped by -- with a fully drive-by-wire system, which includes drive-by-wire steering, 4-wheel steering, electrical mechanical break, 800-volt fully independent active suspension where we have pumps powering suspension each wheel independently. What that all combined will do is that it will provide us with the level of agility that's never achieved before in a vehicle. And beyond that, all the signals and the decision-making doesn't happen through an MCU, which was the paradigm before. But rather, we have large models to directly process these inputs and the output goes directly to the actuators as opposed to going through a much less sophisticated MCU. So that's what we believe is the real smart car is going to be like. And we believe that L9 is going to be the beginning of all this. Thank you.

Operator: Your next question comes from Paul Gong with UBS.

Paul Gong: [Interpreted]. So my first question is regarding the 2026 sales volume target based on the current environment. And more importantly is how should we balance the volume and the market share targets versus our own margins? How important the volume target is in our overall balance of the development?

Xiang Li: [Interpreted] So 2026, as you all know, is going to be an important year from a product perspective as we release the third generation of product. We are very confident about our products. We also noticed that this is going to be the most competitive year to date. This year, you will be seeing more cars released -- more than cars released in the RMB 200,000 and above market than all of the years previously combined. But at the same time, the overall growth of the market is very limited. So considering all this, our total goal is still to reach a 20% year-on-year growth for all 2026. And to support all of this, we have a 3 plus 2 strategy, first, starting on 3. The 3 pillars to support our sales. The first one is the sales system. We will continue to be committed to our direct sales model. And as we've implemented new mechanisms, we will start to see benefits this year of the direct sales model. And the second pillar is the L Series launch, starting with the L9, this new generation of L Series will really be a key pillar to our sales this year. So we will make sure that we get every detail right from the product release to product -- to supply, delivery to sustain -- to ensure the success of L9 and upcoming L Series models. And the third pillar is BEV ramp-up, which includes i6, i8, i9 and MEGA. In the past few months, we've addressed the supply constraints. We have also fixed issues around the release, sales and marketing. So this year is, as the year goes on, we believe that the volume on our ad products will be steadily ramping up which is going to account for a very significant market in the premium market. Now moving on to the 2. The 2 -- the first one is the AI-related investments. In the past year, we spent billions on chips, on models. This is going to be the year I believe that all these investments start to bear fruit, meaning that they will provide a very differentiated product experience to our users. And to summarize, this experience will be proactive, will be high frequencies that every consumer can feel and benefit from in their everyday life. The second of the 2 is our overseas strategy. This is going to be the first full year where we officially run our overseas markets. But through the years of accumulation before, and we believe that this year is going to start to see results and support a long-term growth and overseas remains what we see as a long-term growth opportunity.

Paul Gong: [Interpreted]. So my second question is regarding the impact of raw material cost inflation, including the metals, memories and batteries. What would be the strategy for the company to face this challenge? Should we absorb that within the supply chain? Or should we pass this part of the cost inflation to the downstream?

Xiang Li: [Interpreted] We believe that the current cost pressure is still largely concentrated on key components like batteries and memory chips, which has indeed had some impact on unit vehicle cost. In response, we've already put in place the following measures. First, we are strengthening supply chain collaboration to stabilize pricing while securing supply. On the cost side, we've signed long-term agreements with our core suppliers to lock in both pricing and volumes for key raw materials upfront. This helps us hedge against short-term market volatility and on the supply side especially for AI-related components like memory chips, we have been tight recently. We are continuing to work with key suppliers and to secure dedicated allocation, ensure priority support for production and new model launches. Where contracts include clear pricing terms and adjustment mechanism, we strictly adhere to these terms where there is no such agreement, we work hand in hand with suppliers to share the cost pressure and navigate the cycle together, aiming for mutual benefits in the long term. We'd like us to sincerely thank all of our supply chain partners for their ongoing support for Li Auto. And secondly, we're driving end-to-end cost optimizations in the meantime. We are identifying cost-saving opportunities across the entire value chain from product to R&D to manufacturing, logistics and cost quality. At the same time, we're maximizing economies of scale through platform-based development and higher part commonality across models. This allows us to absorb as much as of the external cost pressure internally as possible. Our in-house developed and manufactured range extender, electric drive units, power modules, self-developed and contract manufacturing domain controllers, silicon carbide power chips and 100 autonomous driving chips and battery packs, all of this is helping us to better manage costs. Third, we're taking a more rational and steady approach to new vehicle pricing. For our 2026 models, pricing will reflect a balanced consideration of raw material volatility, R&D investments and user value to ensure the healthy and sustainable profitability. Our goal is to bring gross margins of new products back to normal to a healthy range. Overall, we're confident that by combining supply chain collaboration, long-term agreements that lock in key costs, platformization, proprietary technologies and rational pricing, we can effectively contain the impact of raw material price increases with a manageable range and maintain stable gross margins and operational quality.

Operator: Your next question comes from Jing Chang with CICC.

Jing Chang: [Interpreted] So my first question is about in response to recently, we have heard about some media reports about the company's consideration of share buybacks. So please confirm if there are any related plans?

Tie Li: Yes. Hi, this is Johnny. I think this is not a media request about a media reporter at Weibo. As a dual primary listing company, both in U.S. and Hong Kong Stock Exchange, we recognize that share buyback is one of the ways or tools we should consider for enhancing shareholder value. And with respect to share repurchase, currently, we don't have any -- we don't have additional information need to be disclosed. Thank you.

Jing Chang: [Interpreted] So my second question is about R&D expense. So what will be our guidance for the R&D expense in 2026? And also last year, we spent almost half of our R&D related to intelligence or AI-related areas. So what is the portion guidance for this year?

Tie Li: Okay. Thanks for your question. This is also Johnny. We expect the R&D expenses this year to remain around RMB 12 billion with AI-related initiatives accounting for half of the cost. This includes the investment in AI infrastructure such as in-house chip development and computing power as well as R&D for AI products like autonomous driving system and Li Xiang engine invested in the last several years. So to clarify, we don't -- the automotive and AI as different -- as independent business. We invest on the R&D side to build AI capability and put it in our company as a whole business model. We monetize all the R&D investments through our current business model. It's not a separate business model.

Operator: Your next question comes from Feixiang Gao with Citic.

Feixiang Gao: [Interpreted] So my first question is about iSeries. So could you give us some details about i8 and i6 orders, sales and especially about the ramping up production of i6? And how do you evaluate Sunwoda battery safety and its cost reduction contribution?

Xiang Li: [Interpreted] Now let me start with Li i8, since its launch in July last year, user satisfaction has continued to rise as owners accumulate miles on their vehicles.and they are satisfied, particularly with driving, charging experience and autonomous driving. The NPS of i8 has risen by over 20% compared to the early post-launch period. During the Chinese New Year holiday, our 5C ultra-fast charging and OTA 8.3 autonomous driving upgrade received strong user acclaim, pushing NPS to an all-time. In Land Roads second half of the 2025 survey, Li i8 ranked #1 in NPS among all large SUVs. Fueled by the strong word of mouth, Li i8 orders have steadily rebounded. Daily orders in early March were up threefold over threefold -- sorry, over nearly 180% versus January. This clearly upward trends reflect strengthening market demand. And now turning to Li i6. We've successfully moved past the most challenging phase of production ramp-up. We are now in the stable delivery phase. Li i6 product strength has been thoroughly validated with this distinctive exterior design, spacious interior, efficient energy consumption and agile handling. It precisely meets the needs of young families. It experienced strong order momentum since launch. And at the same time, all of the supply chain bottlenecks have now been fully resolved. We used to face short-term volatility in battery supply, but we worked closely with our core suppliers to scale production capacity. We also introduced a purchase tax subsidy and extended care policy. I'd like to thank all of our Li L6 users for their understanding and patience. We expect Li L6 to sustain a steady monthly sales of around 20,000 units, and we are on track to efficiently deliver the current order backlog within the next 1 or 2 months. Most importantly, Li L6 success clearly shows that Li Auto's brand appeal has successfully extended from the EREV segment into the BEV segment. Moving on to our battery strategy. We will commit -- we are committed to an open partnership approach while working closely with industry leaders, we retain control in all these partnerships. When it comes to the vehicle performance, we led battery architecture design and rigorously controlled quality at every step. Regardless of which partner supplies the cell, all batteries must meet Li Auto's unified Li's standard for performance, quality and safety. To the user, the experience is identical. There are no differences whatsoever. Additionally, starting in 2026, all Li Auto vehicles will be equipped with batteries from only 2 brands, Li Auto brand and CATL. This marks a deeper level of integration with our core partners. We ask for your continued trust in the Li Auto brand because Li Auto's quality has never dependent on any single supplier. It's defined by our full stack in-house R&D, our rigorous quality control systems and the core values we've upheld from day 1. When people choose Li Auto, it means they're choosing the most reliable assurance. Thank you.

Feixiang Gao: [Interpreted] So my second question is about the in-house chips MAC 100. So could you give us more color about these chips such as mass production time line and where can we see the cost savings and efficiency gains. In addition, how should we understand the company's software and hardware integration and when we'll see the gap between different automakers because of this integration? That's all.

Yan Xie: Okay. This is Yan. Let me answer your question. And M100 we delivered with L9, a new series, and we already started mass production. For M100 itself, with the same silicon area, M100 delivered significantly higher effective compute, giving our VRA algorithms much more design space. For example, we can run our VRA model with about 6x the parameters and 10x the compute of the previous generation while still achieving high frame rate and faster influence. More importantly, as our in-house models, compiler stack and operating system evolved together through codesign, we are beginning to unlock the real potential of our full self-developed autonomous driving stack. The performance gain we see today are important, but the bigger impact is that this system now integration will significantly accelerate how fast our autonomous driving capabilities improve over time. Once the system is officially deployed, we expect the pace of capability improvement to increase substantially. M100 also works closer with Halo OS and vehicle by wire system. In every entire coordination between autonomous driving compute, computation pre and post processing and vehicle control. This shortens the end-to-end latency from sensor photon input to vehicle actuation to around 200 to 300 milliseconds, directly improving the driving experience. The higher local compute also allows us to deliver more intelligent capabilities beyond autonomous driving. Over time, the car will behave more and more like a robot. Some of these capabilities will first appear on the new L9, and we will continue to expand them in the future and more M100 also brings significant cost advantage. First, the BOM cost per chip is much lower than external solutions. And second, we removed the XCU controller used in the previous generation platform. By replacing it with M100 combined with the Halo OS virtualization receive over RMB 1,000 per vehicle. Third, thanks to our data flow architecture and codesign of model and chips, we achieved higher long-term efficiency today and maintain much greater headroom for future performance improvement. When we started developing our own chip in 2022, we believe that around 2025, the industry would move into a new phase where models, chips and operating systems have to be designed together. That kind of vertical integration creates real differentiation in performance efficiency and user experience. Over time, the gap between companies will start to look like the gap between Apple and Android in cell phone world. Once you achieve full stack hardware software integration, the advantage became structural and it keeps widening. Thank you.

Operator: Your next question comes from Tina Hou with Goldman Sachs.

Tina Hou: [Interpreted] So my first question is regarding embodied AI. So for the next 1 to 2 years, wondering what's the strategic plan from Li Auto? What kind of products are you -- are we expect to see? And then also the progress or time line of these products. Also, in terms of the strategic priority, how do you decide between EVs, robotaxi as well as humanoid robot?

Xiang Li: [Interpreted] So in terms of the embodied AI strategy, first of all, on the technological and product front, we believe that there's a lot of commonality regardless what kind of product, physical product we're talking about. So this is an area we will be investing heavily in because we believe that all these investments will be shared across different product shapes and forms. This includes the device side inference chips, the foundation models and operating system as well as the entire data training workflow. In the meantime, on the commercial side, we will be very careful as we invest and explore. We'll be adopting the start-up model in areas like AI glasses and robots instead of using a traditional large company approach where we invest heavily in unverified direction. We will work more like a startup company to start these new initiatives and wait our way through.

Tina Hou: [Interpreted] So my second question is regarding our R&D restructuring. Wondering if the restructuring has been completed and then also under the new structure, how is the progress regarding our autonomous driving?

Xiang Li: [Interpreted] We went through a major organization change in January, where we completely revamped our hardware and software functions. And all this has a shared goal, which is to build a silicon-based digital human being or a smart human being. That's our overall goal. And in order to achieve this reorganization, we've basically reconnected across all our businesses and regrouped the functions by the specific part of the human that they're responsible for as opposed to previously dividing by the so-called business units or the product. So this organization or the specialization happens in 3 areas. The first one is what we think of as brain, and this includes like data sets, which is comparable to human lungs and the chips, which is comparable to the heart, and operating system compared to the neuro system. And then all of these are organized together so that we can work through pre-training and post training and infrastructure together and think of these as the brain. And my rule for these teams is that they cannot touch applications because in many companies, the fundamental research team are attempted to work on applications. But in our case, the brains should not have their own hands, but they should focus on building a very robust brain that can be used across different hardware and software applications. And then on top of the brain team, we have what we call the applications team or in our language, we call it the core software team. This core software team includes software tools like MCP. They include teams that build agents, includes teams that focuses on skills and teams that focus on memories. And all of these different components are designed to work with a shared operating system. That's how we can organize all these different components of software to actually do things for us as opposed to just become a conversational road map. So they can actually -- it can actually strategize and execute a task given by human beings. And then -- so everything I mentioned above is software, which includes skilled memories. And again, here, we avoid the team to the temptation to want to build their own brains. We don't want the body to have its own little brain, but rather we want them to fully utilize the powerful brain that was built by the previous team that I mentioned. So this team focuses mostly on the body, the applications. And then finally, we have the hardware team. This is about building dedicated hardware for AI, for embodied AI, which include an energy team, an electric drive team and the controllers team and when I say controller teams, we don't necessarily use actual controllers like the one seen in automobiles in the past because we've now -- we're now working with a model where we built MCPs for the different components so that the large model can directly talk to these MCPs and control these different components, which makes them a lot more intelligent and more efficient. And when I put these 3 teams together, the major benefit that I see is that they're able to work together across different products because they enable each other's work as opposed to the file that you see with traditional models. And that's how we can deliver products that are also differentiated from the traditional products. So after we made this rework many people at the beginning didn't understand why we are making this change. But over time, very quickly, they're starting to see the efficiency gain in the workflow. For example, for the autonomous driving team they used to iterate their models every 2 weeks. But after the change, they're now able to turn out iteration every 1 day, which is a 14x improvement. And again, as I mentioned earlier, this is partly thanks to the collaboration across the different teams. They all stayed in one room and worked together as opposed to having different departments talk between silos. That's how we believe we can really build in embodied AI that you will see very soon. There is one important thing that I want to add in the -- around the beginning of this year, we've seen many important managers, especially in the R&D field. We've seen several major departures, many of them have been with the company for over 5 years which is something I believe you've also noticed in the market. These leaders in many cases, they're first in line of their business. They went through the entire 0-1 cycle with Li Auto in a startup mode. And they've been very well received by the investment community and many of them have had major funding and as we've seen on the news already. So first of all, I wanted to congratulate these former colleagues for getting good reception in the market, not just funding but also overall reception. And we -- I really wish all of you go back in to compete in your respective market and wish your business to develop well and for you to have a success in your new venture. At the same time, I also want to point out that this change has really given rise to a very exciting change within the company. It has enabled many of the young leaders across technology, in business units and foundation models, product -- vehicle product lines and embodied AI. It has embodied new -- enabled new leaders in these areas of many were born after 1990 or 1995 to really take the spot and take the burden and they're really -- we are starting to see a full bench of very talented and motivated colleagues, which is prepping the company really for the next decade. I believe this overall is really a win-win situation for us and our employees and really ready us for the new decade of development. So I want all of you, investors and those who care about the company, rest assured that this young generation is taking over very well, and they're building on the solid foundation that's built by their predecessors and have yielded some really, really good results, especially in some key technical areas for -- including some of the thesis and key research problems are post-'95 generation has made a major contribution and even some newly fresh college graduates. Many of them were born after the year 2000, they've made some major contributions to our products, to our technology and our are at the forefront of our research and R&D. So this is a really exciting time and I'm really excited to see all this and also makes me very confident about the next decade that's to come. Thank you.

Operator: As we are reaching the end of our conference call now, I'd like to turn the call back over to the company for closing remarks. Ms. Janet Chang, please go ahead.

Janet Chang: Thank you once again for joining us today. If you have further questions, please feel free to contact Li Auto's Investor Relations team through the contact information provided on our IR website. This concludes this conference call. You may now disconnect your lines. Thank you. [Portions of this transcript that are marked [Interpreted] were spoken by an interpreter present on the live call.]