So good day, ladies and gentlemen. Thank you for standing by and welcome to the DSC Holdings second quarter, 2026 English earnings webcast. I'm Gary Dvorak, Managing Director of the Blue Shirt Group, DSC's investor relations firm. I will be the host for today's presentation. I'll go over a few logistical items, then turn the presentation over to DSC. Currently, all participants are in a listen-only mode. After the speaker presentation, there will be a question and answer session. At that time, we will explain how the Q&A will work. Please edit your name in the webinar to show your name and firm name. We'll need this to call on you for any questions. As a reminder, today's program will be recorded. The recording and associated transcript will be available on the IR section of DSC's website shortly after we conclude. The earnings release is available on the company's IR website now. Please note that today's discussion will contain forward-looking statements made pursuant to 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 expectations expressed today. Further information regarding these and other risks and uncertainties is included in today's press release and the company's public filings with the Securities and Exchange Commission. The company does not assume any obligation to update any forward-looking statements, except as required under applicable law. Also, please note that all numbers presented are in RMB unless stated otherwise. Additionally, during the call, we may refer to non-GAAP financial measures. You can find a reconciliation of the most directly comparable GAAP measures in the materials posted on our Investor Relations website. On the webcast today with me, we have Ms. Quinn Zhou, Director and Chief Financial Officer, and Nancy Wang, Director of Investment and Financing. Let me now turn the presentation over to our CFO, Ms. Zhou. Please go ahead.
Thank you, Gary. And good morning, everyone, or good afternoon, wherever you are. And thank you for joining DSC's first earning webcast as a public company. So a couple of hours ago, we held the Chinese language webcast for our Chinese Chinese speaking audience, where our founder, Mr. Junhong Yao, covered operating highlights and I covered the financials. I will now deliver both in English and prepared contents of the two webcasts will be identical. The Q&A will naturally differ, but the full bilingual transcripts of both sessions will be posted on our IR website. So let me start with the operating highlights from Mr. Yao. So in June, DSC completed its IPO on NASDAQ, an important milestone for the company and our formal debut onto the international capital markets. We're keenly aware that we must repay shareholders' trust with sound disclosure, solid execution, and long-term operating results. In the second quarter, China's auto market remained under pressure, yet our key operating metrics stayed resilient with substantially narrowed losses. Our revenue grew at 3.7% year-over-year and our adjusted net loss narrowed by 61.5%. So two business developments in this quarter deserve particular attention. One, our AI products are now delivering real results in real economy, and two, China's used car exports are surging, where we have begun to prudently explore international opportunities. But before we address our business performance, let's take a look at the industry first. Next, intense price competition in the new car market in China triggered rapid depreciation of used car inventory value. DaFungche, which is the operating system we provide to the entire used car industry here in China, DaFungche's data showed prices of mainstream models held by mid-sized and large-sized dealers fell by 10% to 15% within just two months. Industry experts estimate that more than 70% of China's used car dealers operated at a loss in the first half of this year. This upheaval is breaking the old way of running a used car business that was based on experience and gut feel. Leading dealers are now relying on real-time market data for purchasing, pricing, and inventory risk management. Data-driven is no longer just a fancy concept. To the dealers in China today, it is a matter of profit and, in some cases, survival. And that pressure is rapidly deepening the dealer's reliance on our operating system and our AI services. Even against this backdrop, the used car transaction volume in the entire country still rose about 1.5% in the first half, and now it's at a record 9.72 million units, according to the China Automobile Dealers Association, a semi-government association body. And in June, June is a very important month. In June, used car transactions finally surpassed new car retail sales for the first time in history. It is a historic milestone. We expect turbulence in the industry to persist before a new equilibrium forms. When it comes to used car dealers, a classic Chinese saying sums it well. Wildfire cannot burn them out. When a spring breeze blows, they rise again. Wave after wave of resourceful dealers will adapt and grow, and we will stand alongside them through this transformation. Next, I want to reiterate DSC's identity. We are the enabler of the used car industry. We are not a dealer. We do not have any consumer traffic and nor do we do any consumer sales. We provide dealers with our intelligent operating system, Dafengche, and on top of it, the transaction services and AI services the dealers need. According to China Insights Consultancy, Dafengche's market share has remained above 90% for quite several years. This gives us two invaluable assets, industry data that is massive, granular, and real-time, and exclusively ours, plus this long-term, very durable and deep connection with the dealers nationwide. On that foundation, we offer the transaction services that are essential to their operations we offer b2b matching which includes china's number two online used car auction platform we offer used car inspection with over 4 000 inspectors across 250 cities the widest coverage in china we offer vehicle delivery with over 100 self-operated warehouses in every province in China with collaborations with 40,000 car carriers and a reach into more than 2,600 counties and cities. This makes us China's largest single car delivery provider. And we also offer other dealer channel services, connection dealers with traffic platforms or financial institutions or whoever wants to reach Chinese used car dealer community effectively and efficiently. So our mission is to advance the digitalization of auto commerce. We firmly believe that a company's value ultimately lies in the value it creates for society and DSC will be adequately rewarded out of the incremental value we create for this industry so now the two developments that's worth noting the first one is of course ai we spent more than a decade digitalizing this industry ai now gives us the opportunity to lead it into the intelligence age As stated in our IPO perspectives, DSC is the AI application infrastructure for China's used car industry. Vertical AI depends on three elements, large language models, high quality industry data, and digitalized application scenarios. And we exclusively own two of the three, the data and the application scenarios. This is why our AI products suit dealers need very well and they can be adopted rapidly through the dealers existing workflows. Our AI products fall into three categories, tools, digital employees and a platform agent that basically has the collective brain of the industry and our nationwide execution capability. Let's go down the list to have an introduction. Among the tools that have already been launched, we have the Market Intelligence Assistant, which delivers hourly updated retail and wholesale intelligence on every car a dealer seller is considering acquiring, whether to buy the car and at what price. Our prospectus included the story of Mr. Yang Hong, a dealer from Kunming, Yunnan province. His story was that he was facing a seller who demanded over 50,000 RMB for a Volkswagen on polo and all of his four staff buyers said no you know it's not gonna sell you're we're gonna lose money on this car if we pay this price but mr yang decided to trust our marketing intelligence assistant and he went ahead and did the purchase and the car sold within 48 hours with a seven percent gross margin and it was the fastest selling gas car of the month he even told us that he could have sold for more but he wanted to make an example out of this to train his employees to no longer make their decisions based on their experience or their gut feeling and they should adapt a new mode of thinking and rely on data and rely on our market intelligence assistant so that's one of the many stories we have and alongside our marketing intelligence assistant we also provide several others like the imaging assistant which can turn a very casual photo of a car taken on the parking lot into a polished version like the car the photo was taken in a very professional studio but without losing the details of the car and we also have a sales script assistant that can summarize any vehicle selling points and pinpoint the consumer's needs from a conversation record we also have video assistants that can generate marketing short videos of different styles just using some raw photo and video clips from the dealer's employees. As you know, different social media platforms kind of have different personas and then to market on different platforms, you need the video to fit that persona. So we can do that just with one click. So since launching these AI-enabled assistants in March, we have seen very rapid adaptation and now more than 4,100 dealerships are using them by the end of June numbers. And the average daily token consumption, of course, went from zero to over 150 million by the end of June. And the second AI application we have is we call digital employees. So it's a step up from the tools, right? So we have several digital employees already deployed, working, and we have a few more coming out. First, we have the social media operations specialist. It engages consumers in live streams and short video comments, and it pushes the leads into the DaFungche CRM. And by now, we have over 3,215 paying deployments by the end of June. And we also have the management assistant. Once authorized to access a dealership's data, it can function as a real management assistant. can pull data and gives the owner of the dealership real-time analysis and management advice. So far, we have 295 of those paying deployments. The industry employs roughly 800,000 to a million people. That's based on our assessment. And we're assessing that role by role where AI can help. So, the biggest opportunities we see is in sales and procurement, basically buying and selling, right? So these positions are the most important to the dealer and also most costly and on average, we estimate that they cost about 35 to 50% of the dealer's gross profit just for their compensation so we are training digital sales employees and we're training digital procurement employees and they're progressing and we should be able to deploy them for our dealers in the near future and we will update everybody of course when that actually happens and the third one is in my opinion is most important and most exciting one is what we call a platform agent you know or you call it a platform level ai agent but it's more than that it basically you can understand it as an intelligent intelligent being with the collective brain of the industry because we can analyze everything and also it has legs and arms because it has our nationwide execution capability. So that one is very exciting to us and we are working intensely on developing it and then we hope to share with you the progress soon. So the second of the two business developments I would like to share is the international opportunities. So China has the world's largest car park and it's rapidly aging and then naturally while we were exporting new cars we're also exporting used cars and then according to China Automobile Dealers Association the export went up 61 percent year over year for the first half but much of last year's volume in our based on our estimate were those zero mileage pseudo used cars but the government quickly caught up on that and a new regulation came out in january requiring automakers to consent to export vehicles registered under 180 days effectively closing the loophole so this implies you know genuine used car exports grew several fold when you look at 2026. so i believe there are two fundamental drivers underneath this trend. One is rapid price declines, making Chinese used cars globally competitive. And two is the new rule that's pushing the exporters from the pseudo used cars to real used cars. And as an industry enabler, our path is to fit our systems, our services to export oriented dealers. And through them, we can reach overseas these used car dealing ecosystems. Today there seems to exist an exceptional window for us to go abroad. To these exporters we have two distinctive advantages. One is that we have China's largest real-time verified inventory pool. Over 1.7 million vehicles live on our system versus a fraction of that number on some other platforms. And together with a one-stop inspection service, logistics and warehousing, we can service the exporter's needs pretty much seamlessly. And then we also believe that given in a lot of these markets nationwide worldwide, used car dealing, we're still in a very under-digitalized state. The replication of our operating system, dafungshu, in other markets is also highly possible if this export trend continues. So we are prudently exploring these opportunities, But how fast we proceed will be determined by a number of factors that we will consider. Anyhow to close, DSC, in Chinese we say , which is behind me, is in its 14th year. We're not a very young company. We have survived several black swan events. We have had large-scale spinoffs and reorganizations and restructions, and we have started building most of today's business in the times of great adversity. Resilience is in our DNA. We have just set out on this new journey, and we're very pleased to have your interest. So that concludes Mr. Yao's prepared remarks. And now onto the financial highlights I shared on the Chinese webcast. Despite a pressured auto market, our key operating metrics remained resilient. Revenue increased 3.7% year over year and 14% sequentially compared to Q1, which is largely in line with the industry conditions and our seasonality. For our industry, especially in China, the first quarter is typically the slowest you know winter and Chinese New Year and the activities pick up in the second quarter and then the third and the fourth quarter are the peaks. So our adjusted net loss narrowed by 61.5% year over a year from 19.2 million to 7.2 million RMB, reflecting our focus on the quality of revenue growth and on profitability. Before the income statement, I would like to say a few words on the operating metrics we just started disclosing this quarter. We previously disclosed similar operating metrics but that one covered both new car brokers and used car dealers and beginning this quarter we will focus exclusively on used car dealers. This change also incorporates investor feedback from our IPO roadshows and it is designed to help investors easily track the underlying drivers of our most important business segment over time. So the framework has two components And the first layer is user engagement on大风车. We have the average monthly active dealership counts or dealership MAU. You just think of it as this is about the dealership, the stores, the businesses. And then we have the average monthly active user accounts. We call it user MAU. And those are the individual accounts under the dealerships. but they're not consumers they're basically the employee accounts you know a large dealership could employ five people or 50 people or 500 people and then those are the employee accounts right so these metrics are intended for gauging the health of our digital infrastructure you know since the operating system is so important to us and to our overall story we want you to have the data to be able to easily track whether the foundation is healthy. And also indirectly, you can also have a peek into China's used car industry by looking at those data. And then the second part of our operating metrics is all monetization metrics, right? It has two levels too. The first level is the dealerships. because dealerships or dealers, they are our monetization units, right? So you basically count heads, right? How many dealerships have you monetized? So it's an important metric to assess the breadth of our monetization in terms of, in a sense, of the coverage of our monetization. but then our revenue is actually generated from transaction services right so each service each time we provide a service we generate a fee and then dealerships are not all created equal right so one dealership could generate five inspection transactions for us inspection orders for for us a month and another dealership could generate 50 inspection orders and plus 10 other orders for us right so dealerships are not created all equal in our monetization in order to really get to the bottom of it we need to go down to the transaction level so we wanted to give you the number of monetized the transaction services for any given period right so that that's the the logic behind designing those operating metrics alongside these two metrics we of course will disclose the average the average number right the average revenue per used car dealership and the average revenue per transaction are both monetized right so these are more like on the unit economics level. Those are smaller numbers and much easier to digest and track and analyze, right? So together, you know, we have the operating system engagement metrics and we have the monetization metrics. We want to give you those two metrics combined so that you can easily assess how stable our foundation or our moat is and how fast our revenue growth is, right? So together that's the complete picture. Now, I'm going to finally talk about our second quarter numbers. Dealership MAU was 65,000, over 65,000, and the user MAU was close to 200,000. The exact numbers are on the website. Everybody can read them. So the former, the dealer MAUs actually went down a little bit compared to q1 but the user mau went up a little bit compared to q1 and uh we think that it's actually just a reflection of the industry reality right because the the operating pressure was so great uh certain uh lesser dealerships kind of just went out of business but the people in the industry didn't disappear they just moved on they moved to another dealership So we see more user accounts, but slightly less dealer accounts. And then we generate a revenue from over 9,000 used car dealerships. The average per dealership revenue is over 6,600. And we recorded 214,000 monetized transaction services. And average per service revenue is RMB 259. again exact numbers are on the website so our revenue generating services are still mainly transaction services like b2b matching used car inspection vehicle delivery and other b2b collaboration and our operating system and other digital solutions we have they have never been and they will never be main driver for our revenue and actually as our transaction services revenue grow we expect the contribution share of any kind of digitalization solutions to continue to decline okay and now let me turn to the income statement our total revenue increased modestly at 3.7 percent the used car business was generally stable the incremental growth primarily came from our software services to OEMs and it was partially offset by our decision to discontinue certain OEM marketing services so here I would like you to have two takeaways one is that our existing business remained stable and resilient against industry backdrop and then also we're actively improving the quality of our revenue mix. I would also want to note that the quarter's incremental revenue carried a lower gross margin than our company overall level, so the cost of revenue for Q2 grew faster than revenue. But our disciplined expense finance management nevertheless delivered a sharp narrowing of adjusted net loss. Speaking of loss, our gap loss was 240.5 million RMB, but that included approximately 227.8 million of share-based compensation that was recognized upon the IPO and some other IPO-related expenses. the gap number last year was 25 million when we didn't have any of those expenses related to the IPO and then our adjusted net loss was 7.4 million RMB for Q2 26 compared to 19.2 million in Q2 25 and that's down 61.5 percent and it was primarily driven by just across the board reduction in our operating expenses so our general and administered gna expense went down 14 percent sales and the marketing went down 14 percent and notably our r d expenses went down 29 percent respectively excluding share-based compensations. And then the two forces made this pretty significant OPEX reduction possible. One is our disciplined resource allocation. We continuously seek to optimize our business mix. Sorry, my computer was running out of power. So give me one second. Okay, all good. Sorry. So yeah, two things made it possible. One is that we tried to improve our business mix. And we have become very disciplined with how we allocate resources. So we shrank some of the businesses and also we are striving for a more streamlined organization. And two, very importantly, for this OPEX reduction, we had a company-wide AI adoption campaign. And this is for real. So starting earlier this year, we launched this company-wide campaign called AI Era, Race Against Time, Fast and Furious. You know, the Chinese name of this campaign actually took after that famous movie. And this campaign required every single one of our employees, except for those in the front lines offline, like, you know, those who work in warehouses or on dogs or in the markets, you know, like they're in the field, right? So excluding these employees, everybody, everyone else in the company, whether you work for a business unit or you work in a finance department or human resources or PR, everybody had to participate in this campaign. And what does it do? They have to present their AI application achievements. And it's a competition and there are rankings. And then we reward those ranking high and then we give punishment to those ranking at the bottom. And the rewards include extra stock auctions and the punishment sometimes can be as severe as a disconnect determination of employment right so after several months uh this ai application or ai adaptation is no longer just uh what you know the senior management uh dreams of or wants everybody else to do it has already become how our employees operate and how they work and naturally you know we we reaped some of the benefits and we have better processes and we we actually saw like very good ai application tools being created by our own employees and then that gave the management enough room to further reduce our opex right So finally, let me turn to the balance sheet. The IPO proceeds obviously enhanced our financial flexibility and strengthened the foundation of our business. But this flexibility will not change our discipline. Whether it's AI or overseas markets or other growth areas, we will invest cautiously and in stages guided by tangible customer value and monetization results and looking forward management is focused on three things i can share today first is still grow our transaction service revenue we are always looking at potentially expanding our service types and deepening our per dealer monetization and raise the average revenue per dealer And second, we would like to very quickly test out different ways to monetize all the AI applications we have created so far and the ones we will create in the future. And then third, keep improving efficiency and maintaining our cost discipline so that we can achieve overall profitability as soon as possible. Well, that concludes my prepared remarks, and also Mr. Yao's remarks, and Gary, back to you, and we're ready for questions.
Okay, thanks, Quinn. We're now going to begin the Q&A session. To ask a question, use the raise hand function, and we'll call on you and open up your audio. I think a couple of people have dialed in, and if you dial in by phone, excuse me, press star nine to raise your hand, then star six to unmute. When you start, please state your name and your firm before your question so we know who you are. Alternatively, if any of you prefer, there is a question box, a Q&A box, and you can type your question into the question box. It's part of the webinar, and we can then pose that question to Quinn. So we'll pause for a couple seconds to let the queue form and then we'll start the Q&A. Our first question is going to come from Ella G with CR Global. So Ella, you can go ahead.
Yes, thank you for taking my questions. I have two. First, it is quite interesting regarding your latest development in your AI related business. I wonder, can you further elaborate on the business model of these AI products? For example, is it transaction based and you do a take rate? Or is it gonna be subscription based? And also, So who is bearing the cost of the AI token training and usage? And also, what are some business targets relating to your AI products that you would like to achieve, for example, by end of this year?
Okay. Thank you, Ella. Let me address them one by one. So the first one is AI monetization, right? Yeah. So as I introduced, we have three types of AI products, call them products, the tools, the digital employees, and, you know, the platform AI. You know, I don't actually have a good name for it yet. So for the tools, as of now, we provide them mostly for free to our used car dealers. And then the operating system itself, DaFuqchua, also comes free for most of its functions. And then we do have membership fees, like subscription fees for a higher level of functionality. And then certain of these AI enabled functions are embedded into the higher categories, but largely they're free for the tools. And then for the digital, oh, actually, sorry, let me rephrase. Of the tools for the marketing intelligence agent and certain other agent i think we do charge a small fee it's it's still very small you know it's uh it's it's not gonna form any meaningful revenue source for us yet but we we are testing out uh the dealer's willingness to pay for intelligence and then uh so far uh as long as the cost is uh at a reasonable level we we do see a lot of paying agents and then uh for the digital uh employee the AI employee part, the second of our AI products, most of them are chargeable. So, you know, the dealers kind of pay for the employee and then we deploy them, right? So the number I just shared, they're all paying deployments. And then for the last one, the platform level or industry scale AI, I will have more to share, but it's not going to be something that we charge anyone a fee for. We would believe it's going to be more exciting than just charging a fee. And that's the monetization part of the AI products. And then you also ask who's bearing the cost of the training, the tokens, etc. That's interesting because we have the data. So we do use all of the frontier models that are available in China's market. But because most of the training happened on our data, to my knowledge, as a chief financial officer, we're not incurring any significant cost in external token use. It's actually one of the advantages of owning your own industry data. Yeah. And then the last one, Ella, sorry, remind me, the targets, right? The time, like, you know, time targets. Yeah. So we, for the tools, we're just continuously rolling them out depending on, you know, what the dealers need. And then we have rolled out like four or five so far. You've just seen a few short months on the last four or five months. And for the digital employees, we have already deployed two or three, and then we're, you know, as I just introduced, we're looking at the most costly and the most important critical functions of dealers' work, you know, and then we're training those employees. I don't have a specific target date yet, but, you know, once that happens, we will share with the market, maybe on the next call. Yeah. And then for the third one, the very exciting, you know, the industry scale collective brain, super intelligent AI platform, we're also working very hard on that. I don't have much to share this time, but, you know, hopefully soon.
Okay, thank you. And then my second question is, we actually get this question quite a lot from investors. So how does the DSC compare to some, for example, US online auto services companies such as Carvana? And who do you think your true comps are?
Thank you for that question. Yeah, it goes back to Mr. Yao's prepared remarks where he talked about our identity, right? Our identity. You know, when people hear about us, you're Oh, you're in the used car industry. So are you, sorry, are you similar to Carvana or, you know, CarMax or Auto Traders or some Chinese company like Yuxin or Auto Home? Because those are the names they know from the used car industry. So again, you know, we are an enabler for used car dealers. We are not a used car dealer ourself, and we never do anything that's in competition with the used car dealers or have a conflict of interest with them in the past. We've never went down those avenues. And then compared to Carvana, you know, Carvana is a used car dealer. It's a very successful used car dealer, even though it had some, you know, pretty scary moments post-COVID when the chip prices caused used car price to skyrocket and then it dropped. And then, you know, we all saw like how Carvana's stock price dropped. So anyhow, it's Carvana, CarMax and many others like Yuxing, they are used car merchants. They buy and sell used cars. as they are direct to see. We're not. We're not. And in terms of our comp, when you look around the world, really, it's really hard to find one for a , which created much difficulty for me during the IPO process. Because when you think about it, it's pretty amazing that in the largest car park economy in the world, where about 90% of the used cars are sold by this large number of small scale dispersed used car dealers, an army of dealers in anywhere between 200,000 to 300 dealers nationwide in this massive land. All of them are using the same operating system, right? So you cannot find a direct comp anywhere. It's not in the US. To our knowledge, just for the operating system alone, we believe that in the US about 200 to 300 SaaS companies provide different systems to different regions, different dealers, right? So they're very siloed and very separated or segregated. But somehow we had this amazing opportunity in China to leapfrog and to be persistent and to eventually have the whole market on our system. So yeah, so if you ask me, who's my comp? It's a very difficult question. In the car commerce industry, we don't see anyone else in the world. This is one of a kind kind of situation. But then if you talk about business model, we do believe and on the advice of our investment bankers, that our business model is actually quite comparable to several very established companies such as Shopify or smaller ones like Toast or Service Titan. In a sense that we were in completely different industries, but the structure of the business model shares some similarity that we all started out being the operating system provider right and shopify provides operating system for small merchants worldwide pan industry right but it is an operating system and then on top of that you know it started providing uh layers of different types of transaction services and if you look at their financial history you will see that uh in the very beginning i think uh the operating system generate most of their revenues. That's very different from us because it's very difficult to generate revenue from SaaS in this particular economy. And then over time though, their merchant solutions contributed more revenue after a while.
So if you just look at the structure of the business model, we do share similarities with these companies. got it thank you very much okay thanks ella uh our next question is going to come from james kistner of water tower research hi uh matt thanks for uh for taking my questions um i really love the uh leaning into ai so congrats on that it sounds like you're you're really making a lot of uh early progress there but um i guess maybe just stepping back um i mean just kind of talk about like the next few years just you know what you think the biggest drivers of revenue
growth are maybe you kind of rank them if possible like kind of what's the most promising like what do you think is the highest impact yeah um it definitely will come from transaction services because that's where nearly all of our revenues should come from and and it's an amazing place because it's volume driven, right? So in a sense, it's kind of like a stock markets or an exchanges business model. You know, like this past six months, the industry was facing so much pressure and some dealers do go out of business, but the total volume was going up, right? So it's like the stock market, like some people will make money, some people will lose money and certain stocks will go up and certain stocks will go down just like the cars you know some of them will keep their value very well some of them will lose their value very quickly but you know where do we monetize we monetize from the transactions and then can these transactions be skipped you know that that's the ultimate question any answers generally no you know because if any of you if you're interested please go look at our perspectives we have a very nice arrow I personally drew that page that shows the 10 steps of a used car's transaction process just from the beginning like locating the car to you know uh pricing it to buying it to inspecting it by pricing it buying it moving it you know and then cleaning it and then photograph it and then price it again for sale and then go market it and eventually sell it right so this is a long process and and regardless of which market you're in or you know whether the car is powered by gas or electricity or solar you know that the whole process doesn't actually change, right? So back to your questions that, you know, so our revenue will most likely come from the transaction services along the way, you know, it is essentially in a way indispensable, you know, especially for B2B matching in a market like China, 80%, anywhere between 80 to 90% of a used car dealer's cars came from another used car dealer, you know, it doesn't directly come from a consumer. So there's a very vibrant transaction community there, a market there. And then at the very tail end, when a consumer buys a car, the used car dealer kind of functions as the one and only and the most important router for the financial institutions. Because only the used car and the consumer, those two people know that oh there's a need for a loan right so the the the used car dealer is the router and then he gets to route the the the this demand to whoever pays him the the right uh uh channel fee service fee right so there there are like along the way there are naturally these places where as then uh internet uh sorry like uh industry vertical uh digital and internet company with With all the data and the connections and the AI capabilities, we can go in and kind of just reform and benefit from upgrading the whole process. And two, another revenue generator. This might be a little too premature to share, but we do believe that the third AI application or the industry-wide platform-wide agent could present some pretty significant revenue source for us. But I say that with a caveat because it's still premature. We will share more when we have more.
Okay, that's fair. So it seems like you've had some sort of extra special items, you know, some favorable OPEX savings. It's just, can you kind of give us a little more detail on like what's driving that? I assume it's not agentic, but is it, you know, is it just very broad, you know, sort of cost focused? Was there a headcount reduction? Like, how are you, you're doing, you know, kind of doing that? And maybe you talk too about incremental kind of margin on revenue going forward.
I didn't catch the first part of your question. What were you asking about the CAPEX? Oh, sorry, the OPEX?
Yes, I believe you said you've had some, you know, some good cost control. And I just wanted to be sure if you could provide more detail on kind of how you're achieving that and what kind of incremental margin you might see on revenue kind of growing forward, perhaps what kind of leverage there might be in the model as revenue grows.
Okay, two parts, right? The first part, the OPEX reduction. You know, nearly all of our expenses are people, you know just salaries and wages and commissions you know nearly all of them you know we we do have like cloud expenses and uh uh de minimis like rental expenses and all of that but you know the absolute bulk of our expenses are just uh people right so you know that raises the question of how efficiently we're running our business you know how how lean and how efficient the organization is Right. So that's where we put the AI application focus on because, you know, people collaborate with each other. You know, there could be 10 people completing something, you know, along a long process or, you know, with the right AI application help, you can reduce that to two and even faster and better results. So, yeah, just to put it more straightforwardly, that's where the, whether it's GNA or sales or Andy, that's where reduction came from. And then the second part of your question about revenue growth margin and expectations. I apologize, but we are not prepared to give business outlook yet on this call. And hopefully after we, we're a new public company, right? After we get the hang of this and get more familiar and comfortable, we will start adopting the business outlook practice going forward.
Okay, totally appreciate that. One last one I'd sneak it in, just on agentic, you know, in just AI adoption in general, like, what do you think is, you know, the biggest barrier to that and to kind of accelerate? Is it model maturity or is it cultural, you know, in the dealers, like, you know, what has to happen for that to kind of take off?
Artificial intelligence and agentic adoption amongst your customers.
Ah, interesting. Good question. So we have the data and we have all those wonderful models we can access, you know, how fast can we go? I would put it on user acceptance, right? Even with some of our already deployed agents, we have to give it in stages, right? Because everything happens so fast, so fast. So even for some of the products we gave to our dealers, we have to give them like a half-stage product first for them to know what's going on and what is behind this. Oh, so this is the incremental version. And then we can move to the next stage. Yeah, so to answer your question, I think it's still the acceptance pace.
Great. Thank you for taking my questions.
Okay, if there are any other questions, please use the raise hand function and we'll call on you. We'll give a couple seconds to see if we have any others. There's also, there is the Q&A box to type in a question if you have anything you want to put there. Okay. It looks like we're done. I don't see any other questions. So let me just conclude. I want to thank everyone for participating today's webcast. Replays of both the Chinese and English sessions with transcripts are going to be posted on the IR section of our website pretty soon. If you have any other further questions, please contact the IR team by email. And we also invite you to join any of the investor events that we have, which will always be listed on the IR section of our website. So this concludes the webcast, and everyone may now disconnect.
Thank you. Thanks, everyone. Okay, bye-bye.
