Q2 2026 Netdragon Websoft Holdings Ltd Earnings Call
Speaker #1: Dear investors, good morning. Thank you for joining NetDragon Websoft Holdings Limited 2026 Interim Results Presentation. 各位投资界的朋友早安，欢迎各位参加网龙网络有限公司2026年中期业绩发布会。 Before we start, please allow me to introduce the management who are joining us today.
Speaker #1: 首先，请允许我为大家介绍今天公司的管理层。我们有 Dr. Simon Leung，Group Vice Chairman and Executive Director 梁念坚博士，集团副董事长兼执行董事；Mr. Wood Liao, Group Interim CFO, 刘克健先生，集团代理首席财务官；Mr. Ling Chen, Group Senior Vice President, 灵琛先生，集团高级副总裁。Now, let's pass our time to Dr. Simon Leung to host our presentation.
Speaker #1: 现在我们就把时间交给梁博士为我们主持今天的会议。有请。
Speaker #2: Okay, thank you. Good morning, good afternoon, good night—whatever. So, thank you for joining us today. Let me kind of—well, I'm not going to go through the agenda, because actually, you guys are probably very familiar with the agenda.
Speaker #2: So what I'll do is, actually, I'll hit the highlights of our half-year result, and then I'll hand it over to Wood. Then we'll also go into the businesses.
Speaker #2: In more detail, the good news is that on the gaming side, we delivered growth half over half, which is actually very good news for us.
Speaker #2: That shows the resilience of our flagship IP, which is actually very good news. The second big one is actually something we started much earlier, but I think we are kind of wrapping it up as a product.
Speaker #2: So, which is the AI Workforce OS, which I will talk about a little bit later. That continued to support the cost reduction, and in turn, supported the growth of our gaming business.
Speaker #2: And also, the education business going forward—again, Ling Chen and I will talk about how it will support our business going forward. Last but not least, in gaming, AI continues to help not only on the cost side but also on the growth side.
Speaker #2: For mine, the education business in the US and in Europe is still facing a lot of domestic issues in the US, and the two wars in both Ukraine and Iran.
Speaker #2: So it's actually facing the headwind. We successfully narrowed our loss outgoing to our global detail later on. Another milestone that we hit is actually we delivered our latest product, which is AP10 for the premium market, and also the APLE to address the value market.
Speaker #2: The AP10 premium is actually quite critical, because that supports the third point I want to mention. The good news on SaaS is that we have actually grown our business year over year.
Speaker #2: This is going to be our future. When hardware got commoditized, SaaS revenue became very important. The AP10 is actually designed to support all the SaaS businesses going forward, including deploying AI on our panel, which I will talk a little bit more about.
Speaker #2: So with that, I'm going to turn it over to Wood to talk about the results.
Speaker #3: Well, thank you, Simon. Good morning. Everyone. And I will walk you through the financial highlight for the first half of the 2026. And for the revenue, the revenue, total revenue of the group is $2.1 billion for the first half of the 2026, which is done 20% year over year.
Speaker #3: And for segment gaming and application services, revenue was $1.6 billion, which is down about 9% year over year. But despite the year-over-year decline, we have seen a significant signal of stabilizing and recovery, with half-over-half growth by 3% for this half over the second half of last year.
Speaker #3: And for my.ai, revenue was about $505 million, which, due to the delayed pace of the industry-wide demand recovery and at constant exchange rates, decreased a little bit, narrowing to about 17%.
Speaker #3: Gross profit is about $1.5 billion, which is 49% with the revenue trend. On the cost side, we continue to execute an AI strategy, and the overall OPEX decreased about 80% to about $1.1 billion.
Speaker #3: And for three major OPEC and selling and marketing expenses, they decreased by about $218.3 million. And for the administrative expense, it decreased by about 12% to about $418 million.
Speaker #3: Reflecting our continuous effort to deploy AI applications to enhance our internal strategy and internal efficiency, R&D expenses decreased by about 34% to approximately $409 million. This reduction is driven by the operational deployment of our AI employee metrics, which support structural transformation and efficiency gains.
Speaker #3: As a result, our profit attributable to the owner of the company was about $56 million, an increase of 20% this year. And I would like to draw your attention to the two major items here.
Speaker #3: First, we made an impairment provision for the first half of about $115 million, which reflects the price hit—the multi-year lows—at around US $1,600 in the assets of our interim ad.
Speaker #3: Which the $11 price already covers a bit as of now. And secondly, the service payment was decreased by about 42%, to about $8 million.
Speaker #3: This indicates our operational structure is gradually reaching a new normal. Excluding these two factors, adjusted profit attributable to the owners of the company was about $254 million, which is down about 8% year on year. This is relatively stable given the market situation.
Speaker #3: Finally, our work has declared an interim dividend of about HKD 0.5 per share, which is unchanged from last year, reflecting our commitment to our shareholder return plan, as outlined in March this year.
Speaker #3: Turn to the next pages to our segmental financial highlight. And starting with the gaming and application services, the revenue was about $1.6 billion, down 8.8% year on year.
Speaker #3: Gross profit is about $1.3 billion, which is down about 12% this year, with a gross profit margin of over 80%. Our core segmental profit was about $461 million, which is still a little bit lower compared with the same period last year. However, year-over-year, the core segmental profit has already improved. Segmental operating expenses were reduced across key fee categories.
Speaker #3: For our R&D, the decrease about 23% and for the selling expenses decrease about 9% and the administrative expenses about 5%. Overall, the operating expenses for the gaming segment is about done about 40% as of the cost saving impact our AI strategy continue to in.
Speaker #3: Importantly, our core segmental profit improved half over half, demonstrating our business stabilization and operational efficiency. Moving to the My.AI revenue, it is about $505 million, which is down about 21%, mainly driven by lower customer spending due to delayed demand recovery.
Speaker #3: Our gross profit was about $140 million, which is down only 40%. We have a gross margin improvement because of our cost savings strategy. And our core segmental losses leveled to about $126 million, which is down from $195 million in the last year.
Speaker #3: The improvement was mainly supported by our continued cost reductions. R&D expenses are down about 27%, selling and marketing expenses are down about 25%, and administrative expenses are down about 56%.
Speaker #3: And overall, my operating expenses are down about 29% year-over-year as we continue to optimize our cost structure while waiting for a broader industry demand recovery.
Speaker #3: So to conclude, although the top-line growth was under pressure in the first half, we made solid progress in our cost control, a bit under the AI strategy.
Speaker #3: These actions help us to improve our operational efficiency and maintain a stable shareholder return. This concludes the first half highlights. Now, I will turn it over to Min Chen to talk about the gaming business.
Speaker #2: 大家好，很高兴为大家介绍网龙游戏2026年上半年的业绩。那首先从收入端跟研发端我们在上半年都取得了比较好的成绩。首先从收入部分，我们做到了环比的增长。那从趋势来看，我们也保证了游戏业务的收入上的稳定跟未来有机会的增长。那从拉动收入增长的要素上来讲，首先我们第一个是持续过往的战略，在聚焦内容创新上持续做深耕，在长线运营跟优化社交生态上不断去努力。那这个带来的这个增长的基数主要是由我们的用户的增长拉动的，这对于一个MMO游戏来讲，我们觉得是非常健康的一个增长。那同时也为未来的收入的企稳跟恢复增长起到了很好的助力。那接下去大家也会看到在成本端，尤其是研发的成本的部分，通过AI的应用的深化，我们的研发成本跟我们的这个研发效率实际上是不断在提升的。那从这个收入的，从研发成本上也很好的能看到这个的趋势。OK，下一页。那接下去为大家报告一下我们几个核心IP的在上半年的一些成绩。那首先是我们的旗舰IP，也就是我们的魔域IP。可以看到上半年我们的用户活跃度是有一个显著的上升。那这也带动了我们在收入的环比的一个修复。那这个时候的话呢，对我们的游戏的可持续发展起到了非常大的作用。那同时呢，AI的应用，那一开始我们是从一些的非重点项目开始，但随着这些AI的应用的深入，它已经逐渐渗透到了我们相对非常重要的核心IP的核心的研发流程上。那它不光是在我们的这个策划的设计，到我们的内容量的开发，它都能够显著提升我们的效率。那这对我们未来的可持续发展也起到了很好的作用，因为从整个市场的趋势来看，为用户提供持续提供优质的内容是整个我们游戏行业必须要去做到的事情。OK，下一页。那这里在上半年的收入，用户量的增长，它来自于主要来自于两个方面。那今年呢，是我们魔域的20周年。那基于这20周年的这个活动，包括我们配套推出了我们新的片区剧情跟我们的新的职业跟我们的社交系统的优化，这一系列的在游戏内的功能跟内容的优化，显著带动了我们用户的回流。那这个用户回流呢，不光是回来的用户多了，同时呢，用户的稳定性，包括留存也有很大的改善。那对于长线游戏来讲，原本我们的这个长线留存已经在一个很高的一个水平。那目前呢，对于这次的优化，我们在长线留存上至少也有2.8%的一个长线留存的提升。那这个对我们的用户的基数也是一个非常稳定的一个现象，企稳的一个现象。那我们预期这样子的一个生态是能够持续改善与稳定的，这对于未来我们的这个用户生态跟收入起到了一个非常好的一个作用。下一页。那接下去就会不得不再次提到一下我们对AI对我们的帮助。首先呢，AI对我们的内容产出，虽然我们的研发成本是下降的，但是我们在我们核心旗舰产品上的内容产出的量实际上同比是上升的。相对于去年上半年，我们推出了更多的我们的版本内容。那至少说在核心魔域产品上，我们都至少提升了10%以上的内容产出。那这些内容产出不光是帮助我们去给予用户，无论是从这个内容向的玩法向的社交向的，包括我们的工艺的部分，我们有能力去通过AI推出更多的内容。那同时呢，我们在内容的丰富度上跟深度上也有很大的提升。这个对这个是我一种长期能力的建设。我们相信在未来随着我们AI的能力的部署，在核心产品上的内容产出跟品质也能够长期稳定的不做增长。OK。下一页。那接下去是我们的政府IP。那政府IP呢，它的海外收入今年上半年是一个亮点，它不光做到了这个用户量的提升，那同时呢，扣除一些汇率的因素，可以看到在我们的MMO持续的MMO的深耕跟休闲板块的营销的带动下，我们在这个收入上也做到了显著的增长。那其实呢，AI对于政府这个项目其实它的帮助的空间其实会更大，因为尤其是面对海外不同的国家、不同的语言版本，我们有机会通过AI去开发我们的外部版本，我们的手机版本跟不同我们各样的语言版本，让我们的用户跟内容，让我们的版本内容去触达更多的用户。同时呢，把我们的历史的一个的用户可以激活起来，能够持续的去推动我们用户活跃跟收入的增长。那在这点上政府IP在下半年预期也有一个非常好的一个增长的趋势。下一页。那在英魂之刃这个IP上来讲的话呢，它也在上半年的端游其实已经也取得一个很不错的成绩，因为去年是我们英魂之刃的十周年，也就是说这是我们已经运营到第11年的一个长线IP的一个产品。那它今年也做到了同比的增长。那这时候来讲的话呢，对于政府一个MOBA游戏来讲，它的IP年轻化也是一个特别重要的一个重点。所以说今年在我们的上半年，在我们的这个小黄鸭IP的联动下，用户其实给到了尤其是年轻用户给到了非常积极的反馈。这对我们也是一个非常大的一个帮助，它能能够持续去活跃整个游戏的生态跟活力。那在我们在长线的这个内容迭代跟我们的赛事的帮助下，我们的IP跟整个游戏热度也在不断的增长。OK，下一页。那接下去呢，是在今年上半年我们推出了更多的游戏加文旅的一个合作。那其实上在整个内容行业上来讲，游戏加文旅其实上已经进入到一个政府鼓励。那政府无论是个体的文旅局都非常的乐意跟我们游戏做更多的互动，赋予我们更多的内容。而游戏在AI助力底下，我们不光只是一个简单的一个联动，我们可以把游戏的文旅的内容深入植入到我们的游戏的玩法当中去。那这些的改变，那事实上在用户端无论是用户参与度跟商业化都得到了非常正向的一个反馈。所以说的话呢，从长期来看，对于一个长线运营的IP持续文旅提供持续优质内容，做不断的融合跟扩展已经形成了一个非常良性的循环。我们也期待在这方面的投入能够为我们的长线游戏焕发更多的活力。下一页。那接下去是在AI员工的的方面，那事实上其实在上次的业绩发布会上，我们已经做了一个今年的这个整体的展望。那从整个的规划来讲，我们在今年上半年其实也很好的完成了我们原来的规划。那首先呢，是从AI占到我们整体工作量占比上来讲，已经达到了35%到40%，已经是一个同比持续增长的一个趋势。那同时呢，它带来的不光是我们研发成本的下降，同时它也带来的我们研发内容跟品质的提升。而AI员工不仅只是一个工具式的一个使用，它已经成为了一个AI员工，它能够去自主去接单，自主去分析工作，自主的去交付。那同时我们看中了很大的一点是说，因为游戏的研发它都有所谓的周期性，那包括那我们的AI员工目前已经实现了基于工作内容的实时扩容，也就是在一个岗位上我们有能力同时启动100个、200个实例，那等同于原来的100到200个员工在同时的工作。这在我们在游戏的研发性质上也是一个颠覆。而在新游戏孵化上，由于是说它的历史债务会更小，所以说呢，AI对我们的帮助的，尤其是工作量占比其实可以达到一个更高的比例。而在今年下半年，事实上随着我们这些单点的能力的积累，我们有能力打造更强的一个内容的一个闭环，尤其是在我们的内容生产跟内容应AI运营上，是一个有很大的一个机会点的。那业界可能都在提出研运一体、研运一体，那这里的话呢，就是通过AI我们可以更好的把游戏的研发内容跟游戏的运营结合在一起。在下半年我认为在AI推动的一个真效，尤其是在收入的拉动上可能会起到更好的一个作用。那尤其是说整个集团对AI的持续投入，我们会看到AI在自主决策跟自学习的前沿上的探索也发现了一些很大的机会。我们是有能力去抓住AI这个领域的一些红利，帮助我们去做到更好的一个增长。OK，下一页。好，那这里其实就是一个小的案例。那事实上我们在部分领域已经实现了全AI自动化的闭环，也就是说我们把我们的海外版本，也就是我们的英魂口袋版，那可以通过我们的AI的员工，它针对100多个国家同时去制作针对这100多个国家的投放素材，而且把这些内容去推到所有这100多个国家里去。那在我们跟人传统人的投放团队的A/B test的时候就会发现，那AI已经在部分领域去超越了人的能力。首先呢，在过往我们是没有能力去把我们的游戏推到100多个国家的。那其次是说呢，在效率跟探测机会上，事实上AI找到了人的团队没有找到的国家地理机会。那这个是我们非常看重的AI不仅是在样本上的一个潜力，同时在商业机会跟真效上的潜力，这个是我们在未来去逐渐去发掘放大的一个很大的一个重点。下一页。那在总结来看，AI在我们的内容产出部分，那在合并到我们的下半年，那事实上AI已经在部分的领域完成了完整的闭环。那第一个部分就是AI的本地化，那现有我们已经有很长线运营的几个核心IP，那AI可以帮助我们去快速的去推出不同的语言版本，包括针对每个国家的文化跟特点做内容本地化，把继续放大我们的本地化内容，把我们的这个现有的产品推到更多的国家去触达更多的用户，去拉动我们的收入的增长。那同时呢，AI基于我们现有的这个核心版本，无论是IP还是我们的玩法，它可以通过AI可以帮助我们去演化更多的内容版本，去扩展我们的用户，去推到更多的这个我们的可能我们的衍生用户这个领域。那同时呢，通过AI运营的闭环，这些内容有机会呢去更好的洞察每个市场的不同的需求，我们目前看到就是说不同的大区、不同的国家，它的内容喜好，它的不同的这个文化还是有很大的差异性。而过往呢，通常我们只能做一个通用的版本，那现在在AI内容生产跟AI内运营的帮助下，我们有机会针对每个市场跟国家做更加针对化的内容生产跟运营的推广。那这个也是在下半年我们会认为对我们游戏业务很大的一个助力。那接下去在新游戏的孵化的过程中，随着我们过去一两年的一些积累，我们目前也有了一些很好的一些knowhow，从AI如何做出游戏跟做出好玩的游戏，跟做出如何更有商业化机会的游戏上，我们也得到了一些迭代。那目前看在下半年的pipeline的角度来讲，我们会在AI肉鸽、AI文旅游戏跟AI短剧这些过往我们有比较强的知识沉淀，同时呢，AI能力能够完整的把它做出来的领域去不断去做尝试。那通过这个因为我们本身MMO就是一个集大成的游戏，在MMO的里面有各种各样的玩法，我们也会把我们过往的一些成功玩法单独做了出来，做成一些轻量化的游戏，相信这些随着我们的AI的赋能也会有很多的机会，有可能会在海外跟国内出现。那总体上看呢，我们的核心IP我们坚定的去在IP上做投入，长期去把我们的用户跟我们的IP建设好，这保证了我们的IP的长线的收入的稳定跟有机跟未来的持续发展。那在AI的助力下，我们更有机会无论从放大现有的游戏收入，还是拓展到未来的机会上，都给我们很好的一个未来发展的空间，相信说下半年包括明年我们都会有期待更好的一个发展。好，谢谢大家。
Speaker #1: OK, thank you, Lin Zhen. Let me kind of go through mine very quickly, and then I'll get into one of our strategic moves, which is the AI workforce.
Speaker #1: OS going forward. So for mine, I think I would talk about exactly how we managed to increase our gross margin and also decrease our opex in a very hard time.
Speaker #1: By the way, in terms of cost optimization, we have continued, and we will continue. So, you will see the result in the second half of 2026.
Speaker #1: And if you look at the performance exactly, we really narrowed the loss in Q2. So we are on the path to our EBITDA break-even in the second half of the year, which is a combination of cost reduction and optimization, and also taking advantage of the buying season in Q3 and the latter part of Q2 and Q3.
Speaker #1: So, we are encouraged. So, yeah, thanks. This is actually very critical to our path forward. Like I said earlier, the AP10 launch is actually very critical.
Speaker #1: For the deployment of AI services, and for the LE, it is actually our product that we can address the lower-end market, which we call the value market, which is mostly the market outside of the US.
Speaker #1: But that said, with budget pressure in the US, we also expect to see the LE will pick up momentum. That said, we also continue to look at whether we can provide even a better price-performance product in the lineup, which you will hear from us very shortly.
Speaker #1: Next, again, very important—this is actually related to our AI Workforce OS. We have over a million classrooms around the world, which is a great user base to deploy AI services.
Speaker #1: So we'll continue to work on this. We call it a million classroom actual sockets. If you can deploy AI services into all the sockets, we will create a very significant SaaS service going forward, which we are very excited about.
Speaker #1: Next. OK, let me turn to the AI workforce platform, which I talked about earlier. Lin Zhen also talked about it. I'm not going to go into great detail, because these slides are really designed more like a product introduction.
Speaker #1: Since this is an earnings call, I will hit the high level. So, if you remember, we talked a lot about using AI, embracing AI.
Speaker #1: This is the result of embracing AI. What we often realize is that we are using AI to support our core gaming business and also our education business.
Speaker #1: This can be a really great platform for us to support other businesses and other sectors. So, we decided to productize all of the AI features that we developed, and also, with the introduction of agentic AI, we feel this is actually a great opportunity.
Speaker #1: So this is the result of that. Next. So if you look at this, actually, we basically create a foundation with all the features in there, and then we can create workflows using agentic AI to help businesses grow.
Speaker #1: So we're going to focus on different verticals, rather than approaching it from a horizontal standpoint. You can kind of think about it as, really, we are like an employment agency.
Speaker #1: Whatever employee you want, you want the employee, you can come in and you can employ the AI employees to support your businesses, which we are very, very excited about.
Speaker #1: Next. So from a go-to-market standpoint, we see many different ways to deal with it. Well, first of all, first and foremost, our businesses—both gaming and education—will embrace this 110%, if you will.
Speaker #1: And then, of course, we'll go after other businesses, which I have a slide on the next page to show you. But the key on that one is actually our differentiation, which is our moat.
Speaker #1: Our moat is actually not really about the AI technology itself, because we're not in the frontier model. We're going to be using whatever model fits our purposes, so we're going to be using all the models, if you will.
Speaker #1: But along the way, we're going to gain a lot of domain expertise and knowledge, which is our differentiation. That's why we call it the data flywheel that will drive our competitive edge.
Speaker #1: So again, I mean, going back to what I talked about earlier, a lot of people think about doing AI as actually working on a large language model.
Speaker #1: That's not our case. Now, with that said, we're going to be doing something—which I'll explain later on—that we call Solvent Large Language Model.
Speaker #1: But the key is, actually, along the way we're going to gain industry knowledge that will work with our AIs—that will be our big differentiation.
Speaker #1: Next one. So this is the use cases that we talk about. So leveraging all the AI capability on our platform. The number one we can go B2C that we can you know we can do our games, which Lin Zhen talked a lot about.
Speaker #1: We can do learning, whether we should through comic books or interactive courses and all that. So we can continue to do that. But equally important, if not more important, is actually we can do support B2B and also B2G.
Speaker #1: So we continue to use this platform to support our movement in Thailand, but not limited to Thailand. I can't talk about it right now, but actually we're talking to a lot of countries in Southeast Asia to use the same concept that we have in Thailand to support their growth.
Speaker #1: A lot of nationwide AI education products that we can support, one being Cameroon that we talked about. That's the one where we will do what we call a sovereign large language model.
Speaker #1: Again, we're not going to develop large language model capabilities ourselves. We're going to leverage open-source or open-weight models from around the world, mainly in the field from China and then in the US.
Speaker #1: Probably it will be JAMA in China, and it will be either Queen, DeepSeek, or support our education large language model. The idea actually is to localize and verticalize the education model in a school.
Speaker #1: That we can support it, because in some of the African countries, they do not have access to the internet and they do not have access to the large language model.
Speaker #1: That is around the world. So with us doing what we call 'solvent large language model,' which basically runs on a $5,000 US dollar GPU PC, you can support the entire school.
Speaker #1: So we see that as a great basis to help us in our education businesses. Now, I mean, that approach is not only limited to developing countries.
Speaker #1: We are also piloting that in Hong Kong as we speak, in, you know, about four schools right now across different sections. So that's—I mean, when I talk about that, I also cover the smart education platform, solvent LLM.
Speaker #1: Sorry. So, to put it in a nutshell, this is actually what we do. We started out internally. We embraced AI, but that whole effort has morphed into our future strategy, which is leveraging the AI workforce platform to support our growth in gaming and education, but it can also help us to expand into other businesses.
Speaker #1: Next. Okay, with that, let me comment on the outlook very quickly. I think we talked about it. Lin Zhen talked about it in gaming.
Speaker #1: We continue our growth. We continue to embrace AI. And I'm looking forward to the AI-native game taking off in the second half of the year.
Speaker #1: For mine, I also talk about it. We'll continue, of course, optimization. Actually, on that one, we have deployed a lot of AI services, especially in customer support and all the backend services.
Speaker #1: To deploy AI services in the US is a little bit more tricky because of data privacy issues. Actually, the same thing applies to Europe.
Speaker #1: But we have ways to get around it, so we continue to be optimistic. We are seeing some signs of market recovery in North America; budget is being released.
Speaker #1: So that's why we are cautiously optimistic about the second half. Again, I talked at nauseam about AI, and also several solvent LLMs. So that's going to be our, our, our future.
Speaker #1: So, in a nutshell, we are cautiously optimistic about the second half. We believe it will be at least EBITDA break-even in the second half of 2026.
Speaker #1: With that, I'm going to close this presentation. Thank you for your patience. Now, we're going to open it up for questions, and hopefully some answers.
Speaker #2: Thanks, Simon. Thanks, management. Now it's time for our Q&A section. 感谢梁博士以及各位管理层。现在我们进入问答环节。 If you are joining us on the webcast platform, please click the Q&A button on the right-hand side of your screen and key in your questions.
Speaker #2: If you are joining us via phone line, please press star one, star one, and our operator will bridge you through shortly. 如果您希望透过网络平台发问，请在画面右方点击问答键，并输入您的问题。如果您希望透过电话语音提问，请在键盘按星一星一键，并耐心等候我们操作员宣读您的名字。 We are now coming to the next part—welcoming questions in both English and Chinese.
Speaker #2: 我们欢迎各位以英文或者中文提问。好的。 So let's welcome our first question. 我们第一道问题是来自国海证券分析师，是关于游戏方面的问题。上半年的游戏业务环比回暖十分明显。管理层对于下半年整体游戏业务的展望是如何？然后我们游戏收入继续增长的信心如何？谢谢。
Speaker #1: Okay, Lin Zhen.
Speaker #3: Okay. 所以说上半年主要我们的收入是通过用户量用户活跃来拉动回稳的。所以说我们相对比较期待上半下半年这个趋势是持续的。那当然说我们是我们在会争取做到今年说下半年比如说在同比持平的上下。那当然说是我们不会说刻意的去拉动收入增长，因为我们在我们还是比较注重游戏的长线运营。那在收入的增量的部分的话呢，其实我们还有一些比如说基站的IP授权游戏的上线，我们AI游戏的上线，包括我们各种各样的这个IP版本的上线，我们也期待在这部分能够对我们是有正向拉动作用的。谢谢。
Speaker #2: 谢谢林森总。接下来呢也是一条关于游戏的问题，是个补充提问。希望能够问一下就是看到今年上半年整体AI在游戏赋能上面做了很不错的一个成效。对。那我们管理层对于接下来下半年还有未来整个AI赋能这个内容开发，然后呢还有成本的一个运营的下降方面有怎么样子的一个期许。
Speaker #3: 我先提一下其实我们用AI在游戏里面已经蛮多年了。所以那个不是上只有上半年那个林郑你来尝试回答那个问题吧。
Speaker #1: Okay.
Speaker #3: 所以说的话呢可以看到就是说AI帮我们在效率上的提升。那可能在起始阶段我们对这个成本的帮助是是是比较大的，因为确实它有一个很大的一个空间。那在这个从今年上半年开始我刚才前面已经提到了就是说我们一部分的增效其实是对应到的成本的下降。但是我们有一部分的这个增效其实反映在了我们的内容产出的提升上面。那我们刚才在前面也会看到随着我们AI运营AI研发的这个闭环的产生，我们希望把更多的效率提升去放到拉动收入这个这个环节上去。那也就意味着说我们可能未来一段时间我们的研发团队的人数可能会保持一定的稳定，因为我们把一些这个效率提升完以后我们会有更多的这个研发投入到了AI游戏跟新业务跟系统拓展上面去持续让这个形成一个更加正向的循环。
Speaker #1: 我就我我我从这边补充一下就是因为刚您先提到就是说保持那个稳定就是它它看的是一个6月份的重点。所以说下半年呢其实有种数据才去看的话我们在本身上还是会有的。因为你可以看到上半年我们团队还是要要画我们还是刚刚林森总在那个之前也提到是游戏从AI的application从以前是蛮低的比例现在是5050%。27年是达到6070%。所以看到下半年我是27年还是会有下半年我估计就说可能在本身上的效果也是差不多的。27年会小一点因为可能是那个技术已经降低了你才你才再再就不可能每每年都是百分之多少。
Speaker #3: 对那个。
Speaker #1: 这个不可不可行的。所以刚刚林先生也说了未来重点不单单是成本的下滑还是说从内容的要画那边带动我们收入的增长从这点去看的。所以从数据上来看呢下半年27年还是有我觉得是另外一方面是说我们内容的优化会更加明显是从我们在本身上从游戏那块去看到的。谢谢。
Speaker #2: 好。谢谢。谢谢吴总。谢谢林森总。接下来这条问题呢也是我们国海证券这边分析师希望再补充的，就是呢看到这个木鱼AI很有趣。现在已经正式上线测试了，然后呢我们现在很关注的就是我们的商业化路径会是怎么去走的。然后呢第二条问题呢就是我们怎么看木鱼AI与现在的一些通用大模型啊或者市面上的一些AI agent这些平台上面他们的一个竞争边界在哪里。那尤其是在教育内容创作这些场景我们有更多的垂直数据，那这个方面我们的优势能够怎么体现。谢谢。
Speaker #3: 我们有多久期间呢那个可能要。
Speaker #2: 我们有。
Speaker #3: 我们我们看完这我看完这我这个因为那个题目是因为在英文是一个loaded question。
Speaker #2: 对。
Speaker #3: 就是就问了非常多的东西但是我先简单的解答一下。我其实我在已经强调了很多次我们不是做那个那个大模型所以跟其他大模大模型是没有竞争的那个关系其实有非常好的合作的关系因为现在其实没有一个大模型可是可以把所有东西做好所以其实每一个大模型呢它都有它的长处还有短处其实我们那个平台呢最好的就是就是按照我们的客户的需要的需求我们就去把去跟哪一个大模型去用它的那个功能比如说视频我们就特别是跑到某一个大模型比如说其他的我们就做一个到另外一个大模型我们可以拿到最好然后加钱或者是性能价格比最好的给我们那个客户这个第一个第二个呢就是在行业性的那个那个知识啊就是knowledge因为那个不不光是data那么简单不是数据那么简单最重要是我们收集那个数据以后我们怎么去把它就是变成那个那个那个那个knowledge知识那个所以那个呢是为什么我刚才也在讲我们那个所谓的sovereign LLM中文也可以讲是叫什么主权那个大模型啊那些其实也不一定是个大模型因为有不同的手段可以把它实现的可以用大模型的技术也可以从一个叫做RAG的技术也可以做一个叫做LLM wiki的技术然后那个是我们的differentiation所以我们是讲来做刚才其实我有一有一页也讲到我们怎么去做那个商业模式我们其实有B2C也有B2B我我觉得可能B2B呢就帮助我们在那个scaling就是那个会比较快一点所以我们当然内部就是说游戏跟教育然后我们在可能在在金融方面呢在培训方面都会利用我们那个平台其实那个游戏已经讲了很多次了我们就不再重复了在那个教育里边呢就几个方面啊在于我们教育第一个当然也是跟游戏一样我们在成本的那个优化里边也用很多很多AI了然后我们在做那个产品方面我们怎么去帮助老师在教室里面教学或者是在家里备课那些我们都在考虑在进去然后我们也在想到了我们怎么利用那个我们的AI员工那个OS来帮助每个学校去管理比如说那个我们有一个概念就是AI那个那个副校长那个可以帮助做很多我们做admin的那个administration的那个功能所以其实那个就回到我刚才讲为什么讲在那个SaaS model里边呢我们就比较encourage的花了就是我们有很多教室有我们的产品在里边我们就可以做把AI workforce那个platform把它放在那个我们的大屏其实我们的那个AI的surface或者是那个大模型呢我们在教育里边考虑到了第一我们一定会用其他的大模型另外呢我们也考虑了我们自己的那些所谓主权大模型呢可能要两个方法去做一个是也是放在云方面第二个呢也也可以把它本地化放在那个学校里面其实我们现在在测试了我们可以把那个大模型呢放在终端一个可能是一个手机里边都可以为什么放在那边呢又是放在手机了我们可以帮助学生学习这个那个因为其实现在可能一个那个参数比较小的那个大模型也可以在那个比如说在那个iPhone啊那些可以跑所以我们我觉得在发展方面其实空间非常多不好意思那个讲的比较多一点但是那个因为这个题目是比较比较那个heavy所以不好意思。
Speaker #2: 好。谢谢Simon。下一个问题呢是来自我们路演中后台的一个问题就希望呢看领导可不可以介绍一下我们现在Rocket跟帅酷这方面的两个投资还有我们的一个业务合作方面的一个情况。
Speaker #3: 我来就简单讲一下吧因为是我们是投资方所以不不可以就讲太多当然是我们觉得那个是很好的公司所我们就投入进去Rocket我觉得你就你们去那个本来我想讲那个搜索不用了问问AI他们现在做的什么事情我可以简单的讲一下那个Rocket的事其实我们在讲我们那个SaaS services然后在那个教育里边呢我们怎么用用那个主权那个LLM其实我们想象到了在很快的未来了也把Rocket那个眼镜也把它加进去就是好像是一个use case是一个老师在教室里面可以带着那个那个他的AI的AR的眼镜然后就到处跑然后呢他有问题比如说有些同学有问题其实他可以经过Rocket也去问那个AI然后在那个信息也可以在那个Rocket里面显示出来所以这有很多很多的东西可以用Rocket也有那个线上头在里面也可以帮助我们做一些AI的那个怎么那个capture的事情所以就我们充分的expect那个Rocket是我们教育的发展的其中一部分。
Speaker #2: 好。谢谢Simon。接下来呢是一条关于游戏方面的一个问题大家对于AI还是很感兴趣哦希望林曾总给我们分解一下这个AI对于用户时长啊留存率啊这方面这样子的一些提升幅度有什么影响还有分析方面的一个能力然后呢另外呢也希望林曾总帮我们介绍一下从暑假以来游戏的整体表现趋势环比有没有上升呢谢谢。
Speaker #1: OK那第一个问题是说这个AI对我们的游戏的关键数据带来什么样的帮助我分成两大类第一大类叫做间接的帮助第二类叫直接的帮助间接的帮助呢其实我前面提到了说通过AI我们其实提升了内容的产量提供了提高了10%从对内容的质量又提升那这个对比如说对我们旗舰的IP比如说魔域PC来讲它可能在线时长大概会有15到20分钟的一个增长那在我们的气流方面刚才也有提到了可能有2.8个百分比的绝对值的增长那这个对我们来讲还是一个相对比较长线运营的游戏来讲是一个比较可观的值所以说第一大类呢AI是帮助我们提升内容的产出跟质量那去间接带动游戏的这个粘性跟留存这是第一大类第二大类呢其实是我们可能把它全部放到叫做智能PC因为其实在我们今年上半年我们智能NPC其实也是有进展的我们通过AI陪伴AI的一些这个宠物一些的各种各样的AI内容那这样的话呢用户可以直接通过语音甚至文字跟智能PC进行交流那当然这个部分的话呢因为8月初国家也有一些政策的上台所以说从合规角度我们可能先保证合规然后在这个合规的范围之内去尽量去测试用户各种各样跟这个AI的交互那我们也发现说呢确实一些情感陪伴类的一些需求会从用户跟AI的接触中体现出来那我相信说呢这部分也会在未来去满足玩家这种个性化的需求去拉动我们的在线和活跃那这个是第一个第一个方面第二个方面是关于是说这个暑假这个趋势那可能稍微这个补充一下因为我们的这个尤其是魔域IP跟这个我们的这个征服IP它基本上都是这个成年人所以说它的这整个的活跃周期可能跟是不是暑假关系不是那么大就是说因为如果说跟暑假关系大可能是这个未成年可能会会多一些对但我们是当然说在暑假肯定是活跃是有增长的我们这个趋势也是延续的但并不会说是因为有新游戏或者说是这个未成年比较多会有一个显著的增长对谢谢。
Speaker #3: 其实太素体我们也也也不不不适合在这里讲在这里聊了吧。
Speaker #2: 好。谢谢林曾总谢谢Simon so our next question is regarding our education business how does the company view the recovery outlook for our overseas education technology market in the second half of the year and judging from our order on hand do we see any customer demand rebounds。
Speaker #3: I thought I talk about it so actually we're cautiously optimistic so we are seeing some movement in the US is actually so that's why we are cautious ous cautiously optimistic I have to admit is actually quite slow in the Middle East for obvious reasons I do not expect to see the Middle East recovering in the second half of the year Europe is actually highly depending on a couple of countries one being Germany so so again I mean you know we are cautiously optimistic I think 2027 will be the year of recovery for us。
Speaker #2: Thank you, Simon. Due to the time constraint, we are welcoming our last question right now. 我们最后一道问题呢，是希望管理层能够给我们分享一下，我们是否会维持一个较高的派息比率，然后未来的这一个派息，如果不够，还有我们整体的一个现金流的运营情况，我们是如何去平衡的。谢谢。
Speaker #3: 我我来回答那个问题吧这个其实我们在上次那个公布已经公布了我们我大概是我们company no less than。
Speaker #1: 港币6亿。
Speaker #3: 港币6个亿你都你可以做一些do some math你就可以知道那个那个我们这个是我们的一个计划了所以那个应该目前来讲是没有改变了。
Speaker #1: 我补充一下是说我从我们现今的情况来看我们在6月底我们还有包括我们一些可变现的一个投资里面我们还有18亿人民币的现金在手所以肯定是刚刚也提到蛮多是下半年我们的情况就无论是教育或者游戏方面我们还是觉得是会continue improve了就是就继续增长所以说未来我们比较有信心是说去commit我们刚Simon提到是3月份我们提到的一些shareholder return的program所以说我们是蛮有信心会完成这个program的。
Speaker #3: 6个亿港币还没用完呢。
Speaker #1: 还没用完。
Speaker #2: 好的。
Speaker #3: 不好意思我开开玩笑这个OK好谢谢。 OK谢谢大家。
Speaker #2: 谢谢孙总谢谢吴总。
Speaker #3: 拜拜。
