Kuaishou Technology B

And this meeting is being recorded. And now I'll turn the call over to Mr. Matthew Zhao, VP of Capital Market and IR at Quashel. Thank you, Arbiter. Good evening and good morning, everyone. Welcome to Budget Technology Q1 2096 Financial Results Conference Call. Joining us today are Mr. Cheng Yixiao, Co-Founder, Chairman and CEO, and Mr. Xinbin Xiaobo. Before we start, please note that today's discussion may contain forward-looking statements which involve a number of risks and uncertainties. Actual results and outcomes may differ from those discussed. The company does not undertake any obligations to update any forward-looking information, except as required by law. All important information about this call and forward-looking statement is referred to the company's public information. All first quarter 2026 results announced and ended March 31, 2020, as issued earlier today. So in today's core management, we'll also discuss certain non-FRS financial results. These are provided for additional information and should not replace the FRS-based financial results for a definition of non-FRS financial measures or reconciliation of the FRS to non-FRS financial results and related risk factors. Please refer to our Q1 20.6 results announcement. For today's core management, we'll use Chinese as the main language. A third-party interpreter will provide simultaneous interpretation in prepared remarks session and a consecutive interpretation during the Q&A session. Please note that English interpretation is for convenience purpose only. In case for any discrepancy, management statements in their original statement will prevail. Lastly, unless otherwise stated, all transit units mentioned are in RMB. Now, I hand the call over to Yixiao. Welcome to Qashul's Q1 2026 earnings conference call. In Q1 2026, we made a complex and evolving market economy environment to continue to advance our AI strategy execution, clean AI, and its global leadership in multi-module video generation. And AI technologies continue to enhance the vitality of our economy ecosystem. It expands monetization and improves organizational sufficiency. In Q1 2026, the average DA use on Qashul averaged over $413 million total. Revenue increased by 3.4 YOY to 33.7 billion. Revenue from our core commercial business, including online marketing services and other services, primarily e-commerce, increased by 10.7 YOY. Adjusting net profits reached 3.4 billion RMB with an adjusting net margin of 10%. CleanAI demonstrated strong momentum, driving our second growth curve, sustained its global leadership in model capabilities and product experience, while achieving rapid monetization growth. In Q1, CleanAI generated revenue of over 650 million RMB, representing worldwide growth of more than 300%. Next, I will walk through the details of our major business segments in Q1 2026. First, our AI strategy and the progress of our large video generation model, CleanAI. In Q1, CleanAI continues to advance its vision of empowering everyone to craft capitating stories with AI. Through ongoing model iteration, intelligent product upgrades, and deep penetration across professional scenarios, CleanAI further reinforces global leadership in AI video generation. At a modern technology level, in February 2026, we launched the CleanAI 3.0 model series. Due to an all-in-one product framework, the CleanAI 3.0 model series was full of model inputs and outputs, spanning text, images, audio, and video, integrating video understanding generation and adding in one streamlined AI workflow. While supporting video generation of up to 15 seconds, the Cling AI 3.0 model series delivers highly flexible storyboard control and a more precise semantic alignment, incorporating simultaneous audio-video generation capabilities with strong subject consistency to further extend the boundaries of AI storytelling. At a product level, Cling Lab agents more upgrade the creative experience from standalone to an intelligent agent-driven system. with features including multi-turn conversational editing and a multi-perspective intelligent shot expansion. In addition, CleanAI launched the Team Plan, supporting real-time collaborative creation for up to 15 members and enabling creators to efficiently manage content creation workflows among teams. Recently, we launched Baseball Live Effect, which once again ignited a global AI creation frenzy and swept across social media platforms worldwide. This viral hit propelled Queen AI to the top of the overall App Store charts in 42 countries and regions, including Germany and Brazil. We have continued to focus on the core needs of professional creators across the field in television, advertising, e-commerce, and gaming sectors, supporting common production with end-to-end industrial-grade capabilities, and driving meaningful cost reduction and efficiency improvement. Clingy Eye was used in the creation of selective virtual scenes and visual effects, resulting in a Chinese historical drama that swerves into file shares in the Hollywood series House of David. Clingy Eye has supported the generation of hundreds of high-quality shots, including grand scenes and complex battle sequences, demonstrating its exceptional strength in commercial film and television production. With broader adoption across professional creative scenarios, Clingy Eye's commercialization has accelerated. In Q1 of 2026, Cling AI generated a revenue of over $650 million, representing growth of more than 300% year-over-year. In March, Cling AI's analyzed return run rate reached approximately $500 million. We made continued solid progress in advancing the research and development of our general-purpose models and in deploying AI-driven inhabitants across our commercial and organization ecosystem. At the general-purpose model level, we released the CAD Coder Pro V2, an eugenic coding model with strong capabilities from MUI aesthetics, command-line reasoning, and agent execution. It is compatible with the mainstream AI coding tools such as Cloud Code. It has been trained and optimized for open call and is capable of navigating complex real-world application workflows. In terms of AI empowerment for our commercial ecosystem, we continue to deepen the application of regenerative recommendation launch models and intelligent bidding models in online marketing services scenarios, driving roughly 4% growth in domestic online marketing services revenue in Q1. To optimize our generative recommendation large models, we incorporate multi-model aligned advertising in settings. By combining value-aware, supervised learning with ranking-guided reinforced learning, we improve the quality of the material recommendation candidate set. We also enhanced inference efficiency through model optimization at inference stage for improving model performance. For e-commerce business scenarios, the new generative search framework, OneSearch v2, was fully rolled out across e-commerce search scenarios in Q1. Through tech innovation, we have introduced a self-destination generative search framework based on latent space reasoning and enhanced the model's inference capabilities and search experience without incurring additional inference costs and or service latency, withdrawing incremental GMV growth after approximately 50% of our e-commerce search business. At an organizational ecosystem empowerment level, Bioshock's proprietary coding tool, CodeSeeker, has driven AI-generated code penetration to over 50%. it's also evolved into a company-wide general presentation to my flickr explaining its use scenarios from engineering coding to broader employee facing scenarios it now empowers functions across r d product operations data etc enhance overall organizational efficiency second user growth and content ecosystem Q1, an average data use on partial app reach of 413 million, an annual reach of 772 million. Average daily timestamp per user remained relatively stable. By providing differentiated premium content, generating our traffic mechanisms, and expanding social interaction scenarios around Chinese New Year, we offer users a higher quality and a more heartwarming online community with distinctive partial characteristics. In refining our distribution mechanism, we balance user experience with monetization efficiency, resulting in increased exposure for premium content. In terms of user growth quality, we further refine operations across all channels. By integrating user acquisition spanning with monetization scenarios and continuing to innovate our user retention strategies, we successfully improve user growth ROI. During the 2026 Chinese New Year holiday, we created an immersive online spring festival experience to drive faculty user growth through innovative interactive features and extensive premium content matrix, achieving a new historic peak in DAUs. In terms of our content ecosystem, we launched a series of special characteristic content IPs, including special spring festival galas for the Year of the Horse, the Ulaanbaatar Grand Stage, the Northeast Comedy Show. Leveraging Queen AI, we released a generated Chinese New Year animation front-page service. The show got a song. These initiatives foster vibrant, festive atmosphere for users while amplifying the popularity and influence of cultural native content. User social interactions increased significantly during the Chinese New Year campaign. The number of user pairs using our social interaction products was a group of 25% compared with the pre-Spring Festival period. And the number of users that sent private messages increased by 15%. These initiatives fostered a vibrant, festive atmosphere for users while amplifying the popularity and influence of quite short-native content. Data showed that our 20.6% tenancy-year programs generate over 15 billion live streaming views and more than 250 billion for video views and over 6.5 billion cumulative likes upon user growth and reinforcing using MyShare for broadcasting press. Third, online marketing services. Q1, revenue from online marketing services reached 19.6 billion, up to 9.3% of our YOY, with revenue from domestic online marketing services growing by more than 10% year-over-year. In Q1, the content, consumption, lifestyle services, and AI application sectors were the primary drivers of our non-economic marketing services revenue. In the content, consumption sector, AI reduced production costs and lowered creative thresholds for coming-style short plays, driving rapid growth in content supply and related marketing demand. As of the end of March 2026, the peak of daily marketing spend on special comic-style short plays exceeded 20 million RMB. Within the lifestyle service sector, where clients primarily operate on a lead-based model, we advanced more refined industry operations across 20 verticals, including healthcare industry and equity-compromised service allocation and automation. By tailoring our solutions to conversion characteristics of different industries, we help merchants improve customer acquisition efficiency and improve incremental advertising spend across these sectors. At the same time, we helped the merchants reach potential customers more efficiently and improve user conversion rates through product upgrades. In addition, during the Finance New Year period, demand for marketing placement in the application vertical was strong. We effectively captured revenue budgets and optimized the deep conversion outcomes contributing to AI application clients' increased marketing spend and commitment to our platform. In Q1, we accelerated penetration of AI across diverse online marketing services scenarios, covering the full pre-placement, in-placement, and post-placement lifecycle. This improved clients' placement experience and drove growth in total spending from online marketing services. In pre-placement, the generation of AI-based marketing materials enabled merchants to produce materials at a lower cost and more efficiently. As of March, AIGC short video marketing material spending contributed 10% of total for video online marketing spending on our platform. At the in-placement stage, our universal also X placement solutions became the dominant placement tool adopted by most online marketing clients. During the quarter, we added a new AI agent feature to UAX placement solutions. Learning from the best practice in placement optimization, creative generation, and editing across verticals, our UX placement solutions automatically assisted our advertising teams with tasks such as ad unit creation and feed management, improving overall placement efficiency, and post-placement stage AI. Driven analytics automatically review the performance data and provide atomic feedback to clients, reducing manual operational workload. Meanwhile, additional employee solutions offer 24-7 automated responses, enabling clients to handle consuming parts in real-time, especially during period sessions over night hours when human support is limited. For e-commerce marketing services, we further advance and deepen our omni-domain traffic synergy strategy through improved coordination between our organic and commercial traffic pools, Enhancing e-commerce traffic exposure and business growth for brand merchants. In Q1, we onboarded an increasing number of brand merchants and small and medium-sized merchants. The number of active merchants using marketing placements increased by 33% YOY and brand advertising spending increased by 42% YOY, supporting GMB growth for brand merchants' self-operated business across the Omni domain and areas. Meanwhile, AI capabilities have been fully integrated across our end-to-end e-commerce marketing placement workflows. With coordinated efforts of user interest in, for instance, AI agents, creative and production AI agents, and appealing production AI agents, both marketing placement precision and efficiency improved. On the product side, with continuous upgrades, our Omni software marketing solutions accounted for a greater share of our total spending from e-commerce marketing services and became the primary placement offering for our e-commerce marketing services. Our Net Transaction ROI product helps merchants optimize their Net Transaction GME, enabling more stable settlement outcomes. In Q1, client penetration rate reached 45% across industries and meaningfully reduced product return rates. Meanwhile, our full-store hosting model allows merchants to overcome single product placement constraints through one quick store-wide placement, spraying up manpower. Who's e-commerce? We will advance our e-commerce strategy through three key upgrades, prioritizing clean user growth, supply acquisition, and deeper integration of e-commerce and commercialization traffic. This will enable merchants to better capture the end-to-end synergies across omnidomain scenarios, amplifying growth momentum. In Q1, by centering our advancing strategy, our e-commerce business achieves a sustainable healthy growth. We remain committed to strengthening omni-domain traffic synergies and refining operations across the full buyer lifecycle to drive long-term growth in our e-commerce buyer base. In Q1, on the supply side, we prioritized onboarding brand merchants and new merchants while further improving the quality of our supply in the first quarter. For brand merchants, we continue to advance our Voyage initiative, which launched in the fourth quarter last year. targeting top-tier brands across diverse verticals to support them on multiple fronts, including traffic operations and brand building. And at first, the quarter driven by incremental growth from new brand merchants and brand merchants' contribution to overall e-commerce, CMV, and commercialization continued to increase. maintaining strong year-to-year growth. Meanwhile, existing merchants continue to scale and stabilize their operations, while strengthening the health of residents of the e-commerce supply ecosystem. In new merchant acquisition, we work closely with our service providers across one of our targeted priority industrial zones nationwide. In Q1, the number of new merchants onboarding in these industrial zones increased about 41.8% worldwide. In addition, in collaboration with industry teams and SME merchant teams, we actively empower the new merchants to thrive across our full lifecycle from onboarding and early growth to scaling through a mix of merchant-incentive programs, operational tools, and business safeguards and marketing initiatives. In Q1, the number of SME merchant groups is leading to a healthier merchant structure and a more diversified supply. In Q1, we thought we'd improve our KOL ecosystem structure. This enhanced the supply of high-quality e-commerce content, and we'll continue to strengthen support for mid-tier KOLs. In January, we launched a treasured streamer spotlight initiative leveraging platform resources to identify and support outstanding KOLs across verticals and help them scale. In addition, we further refined our incentive policies, which significantly improved KOL streaming frequency. In Q1, the number of average daily active streamers hosting live sessions with over 10,000 followers grew 10.1% year-on-year. To empower KOLs, we began working with distribution and industry initiatives and providing top-tier merchants and KOL as we diversify monetization growth and market opportunities by expanding the reach of high-quality products with distinctive cultural e-commerce characteristics for a broader user base. Through the subsidy initiative, especially the KOL Blockbuster initiative, along with continued optimization of our distribution product allocation capabilities, we improved the efficiency of KOL product matching. In addition, we hosted offline matchmaking events between merchants and KOLs and introduced tiered services for KOLs at different levels, enabling high-precision distribution matching. In Q1, the number of merchant KOL matches in the distribution pool increased by 47 year-over-year, while the number of active KOLs participating in distribution grew by 23.5% worldwide. Our omni-domain e-commerce operations ecosystem continues to show strong growth momentum as we strengthen coordination between content-based and shelf-based scenarios through intelligent subsidies and refined operations across the full user lifecycle. We achieve end-to-end efficiency improvement from product recommendation to repeat purchase, building on stable and sustainable growth of foundation for merchants. Content-based scenarios continue to serve as the important driver of user demand. As user consumption habits increasingly expanding through browsing, search, and shopping mall exploration, our dominant domain consumption mindset among users have gradually taken shape. In Q1, e-commerce intent driven search TV grew 11% here over the year, while new and returning buyers in the shopping mall increased by approximately 36% in March. indicating users strong activity in shopping in time. In addition, we ensure supply will increase in purchase frequency among pen-shell-based e-commerce users. Furnishing end-to-end AI capabilities, we deliver tangible operating efficiency gains for merchants and upgrade user experience. Leveraging large model capabilities, we comprehensively upgrade the shopping decision-making process to end-to-end empowerment, spanning user-latent demand activation, search engagement, guided shopping and live streaming rooms, and finally, subsidy distribution. Due to large-model technologies, we launched an upgraded search experience by introducing an AI agent-based one-stop intelligent shopping assistant. With this upgraded search, we evolved from user-initiative product lookup to an AI-led product recommendation, significantly improving search conversion efficiency. In addition, leveraging AI to empower sales operations, we optimized the process of guided shopping in live streaming rooms. Illustrating scenarios of real-time product and highlight summarizations on an AI-powered outdoor real-time hosting feature generated over 10 million in incremental GMB per day per merchant. Next, regarding our live streaming business, in Q1, live streaming revenue reached 8.5 billion RMB. We remain committed to the health of live streaming ecosystem as our core priority, focusing on supply-quality improvements, content enrichment, and AI innovation to build a sustainable live streaming ecosystem for long-term growth. We continue to support high-quality content categories such as premium group live streaming and strengthen the professional operations of our current agencies. Solidify and the foundation of live streaming supply. On a product and technology front, AI capabilities fully empower live streaming rooms, AI tools including AI interaction assistance, additional avatar solutions, and AI private messaging improve the streamer's service efficiency and enhance the viewer's engagement experience. Quina's video generation experience significantly empowered live streaming data creation, accelerating a gift rollout, and enriching creative expression while boosting user willingness to pay. In Q1, the AI Universe series goes with the customizable special effects sent by users with 1.1 minutes. In content, we launched a diverse range of live-streaming interactive features during the Chinese New Year to build engaging live-stream scenarios. The Kuaishou Mastermind Quiz series featured over a large AI digital human streamers generated by CleanAI, interacting with users throughout the experience and attracting nearly 50 million users to actively participate. At the same time, we put our strength in our gaming content ecosystem. During the Chinese New Year, we launched a spring festival player carnival campaign 50 game developers and collaborating with major game ITs to roll out exclusive content and user benefits. Moreover, our esports business, a key strategic pillar of our gaming system, achieved a breakthrough. In April, Qualcomm's PS3 team won the King Pro League Spring 2016 Championship, driving the further evolution of the gaming system. Finally, about our overseas business programs. In Q1, we continuously explored high-value growth strategies for our overseas business while responding to market changes with business resilience. In terms of traffic, we continue optimizing user acquisition efficiency, and user growth makes a cultivated economy ecosystem Meanwhile, we further expanded the content verticals favored by our core users to deepen their engagement. Brazil's key market purpose is developing a maintained steady average DAUs and average daily timestamp per DAU quarter-over-quarter. For online marketing services, we consistently strengthen our advertising product availability by leveraging AI to fully empower our end-to-end marketing. workflows, improving placement efficiency and performance stability for clients. At the same time, we capitalized on our strength at content platforms to look like a commercial creative to unlock incremental advertising budgets, particularly from work that was led by cross-border e-commerce suppliers. Our e-commerce system pursued the solid year-over-year growth in GMB and order volume in the first quarter. but strength in key categories and high quality products supply improved conversion and repeat purchase rates. Meanwhile, through AI-assisted content reproduction and content reformation, we continue to optimize operational efficiency and profit generating availability. In summary, confronting many challenges and rigorous tests We remain steadfast in our core AI strategy, breaking through barriers with our deep tech and ecosystem strengths. Looking ahead, we will navigate certain needs with long-term results, continue steepening AI integration to a long-term value for all stakeholders, and allow new blocks of data. And that concludes my remarks. Thank you. I'll hand it over to Jinbin. Thank you, Yiqiao. In Q1, we continue to advance our AI strategy and main meaningful programs, leveraging our leading AI capabilities. We deepen our empowerment of customers' time and business ecosystem, delivering healthy, steady growth across both our operational metrics and financial performance. We continue to broaden the application of AI-locked models across business scenarios, further optimizing user experience and enhancing operational efficiency for our business partners. At the same time, AI is rapidly realizing its commercial value as the company's second growth curve. In Q1, the group's total revenue reached 33.7 billion RMB, with revenues from our core commercial business, which includes online marketing services and other services, primarily e-commerce, growing 10.7% year-over-year. Notably, QINGA generated revenue of over 650 million RMB, representing a year-over-year growth of more than 300%. Adjusting net profit was 3.4 billion RMB, with an adjusting net margin of 10%. Well, continuing to scale our investment in AI, we maintain the group's overall profitability and operational cash flow at a healthy level. Now let's take a closer look at our Q1 financial performance. Our total revenue grew 3.4% year-over-year to $33.7 billion in Q1. The increase was mainly driven by growth across our online marketing service, e-commerce, and clean AI businesses. Online marketing services revenue increased 9.3% to 19.6 billion RMB in Q1 from 18 billion RMB in the same period last year. This growth was primarily driven by the accelerated generation of AI across diverse online marketing services scenarios, which views marketing material regeneration costs for clients and optimized their placement experience while also improving conversion efficiency and driving higher spending by our marketing clients. Revenue from other services, including our e-commerce and Queen AI businesses, reached 5.6 billion RMB in Q1, up 15.9% from 4.8 billion in the same period last year. The increase was mainly driven by the continued expansion of our Queen AI business by continuously refining Queen AI's foundation models and developing more innovative features. We have broadened this doctrine of professional creative scenarios of salary and commercialization. In Q1, our live streaming revenue was 8.5 billion RMB. We consistently cultivated high-quality content offerings, expanded live streaming scenarios, and leveraged AI-empowered innovations to develop a rich and healthy live streaming ecosystem and diverse high-quality content. Cost of revenues increased 11.1% year-over-year to 16.5 billion RMB Q1, accounting for 48.8% of revenue. The increase was mainly due to higher revenue sharing costs and revenue growth, as well as increased bandwidth expenses, server custody costs, and depreciation of the property and equipment and the right of use assets, and amortization of intangible assets. Based on the above, our gross profit was 17.2 billion RMB in Q1 compared to 17.8 billion in the same period last year. Gross profit margin was 51.2% compared to 54.6% in the same period last year. Starting to expenses in Q1, selling and marketing expenses were 10.3 billion RMB compared to 9.9 billion in the same period last year. Selling and marketing expenses slightly increased to 30.6% of total revenue from 30.4% in one last year, primarily attributable to increased selling in promotion activities. RMB expenses increased 9.8% year-over-year to 3.6 billion RMB, accounting for 10.7% of total revenue. The increase was mainly due to higher employee benefit expenses, including share-based compensation expenses and increasing investments in AI. Administrative expenses were 770 million RMB compared to 830 million in the same period last year, primarily due to a decrease in employee benefit expenses, including related share-based compensation expenses. The group level net profit for Q1 was 2.9 billion RMB. Group level adjusted net profit was 3.4 billion RMB, with an adjusted net margin of 10%. Our balance sheet remains robust. Cash and cash equivalent time deposited to financial assets and restricted cash totaled $117.7 billion as of March 31st, 2026. Net cash generated from operating activities in Q1 was $3.1 billion RMB. Additionally, we actively delivered on our commitment to shareholder returns based on market conditions. As of today, we had room purchase approximately 854 million Hong Kong dollars or around 17.96 million shares, representing about 0.42% of our total shares outstanding for 2026. We can add, we'll continue to prioritize user needs and remain committed to investing in the app, leveraging our leading activities that will fully boost the vitality of our content ecosystem and expand monetization, reinforcing our competitive edge in every changing market, and creating long-term value for users, partners, and shareholders. This concludes our prepared remarks. Now we can open the poll for Q&A.

At this time, if you would like to ask questions, please press star 1 and wait for an interview.

我们现在进入问题环节 第一个问题有来自UBS的Kenneth Fong The first question comes from Kenneth Fong of UBS Please go ahead 请提问 谢谢管理层接受我的提问 恭喜一个非常稳健的业绩 以及肯宁非常迅猛的发展 我有一个问题是关于肯宁AI的 我们看到26年一季度的生涯化的发展 也是非常超预期 我想问一下就是主要收入增长的驱动是什么呢? 以及可林目前主要的应用场景主要是什么? 应该如何展望水平生成大模型的未来的应用的前景呢? 谢谢,我激烦一下。 Thank you, Manishen, for taking my questions and congrats on the very robust growth on Kerlin. I have a question on Kerlin AI. This commercialization progress in first quarter has exceeded expectations. What are the main drivers behind this strong revenue growth? What are the recent use case and primary application scenario for Kerlin? And how should we assess the future application process for the video generation of large language models?

Thank you. As you may remember, last March, ARR was worth $1 billion. So ARR has increased by nearly 4 times in a year. The increase in business income from ARR is due to the API of B-end business clients, the lack of income, and the lack of income of P-end users. No matter the number of users or the amount of monthly income, they all achieved very high growth. In terms of flow, B-end business clients and P-end users all showed better flow. Thank you for your question.

In Q1 2026, CleanAI generated revenue of over 650 million RMB, up more than 300% year-over-year. In March 2026, the ARR of CoinAI was approximately 500 million US dollars. For those who recall, our ARR stood at 100 million USD in March of last year. So it's increased by nearly fourfold in just one year. The rapid revenue growth was driven by both corporate API services and paid subscriptions from prosumers. The number of paid prosumer subscribers and their monthly ARPU increased rapidly. From the retention perspective, our corporate clients and paid consumer subscribers maintained healthy retention trends, and are scoring CleanAI's robust technology and product capabilities in professional creative scenarios. This has translated into strong user stickiness, providing a solid foundation for CleanAI's long-term sustainable revenue growth.

关于可灵目前的用场景,主要包括广告营销, video, video, video, video, video, video, video, video, and provide visual support for the advertising team to determine the advertising tone. But the hidden scenes can not be deeply integrated into the full-fledged production. From script analysis, role setting, and scenario analysis and setting, the intelligence division and the central presentation, and the high-resolution automation creation, the production cycle and cost are greatly compressed. At the beginning of April this year, more than 200 million A.I. books were released on the whole network, all of which are non-professional. The creator used Killeen to create almost all the images from the beginning to the end. In the game production scene, Killeen can make the concept original in a dynamic way. Helping Killeen to save a lot of time in the production of the dynamic atmosphere. At the same time, it can also deal with the scale of the subsequent production. The plot and the process are also animated.

CleanAI is viewed mainly for professional creative scenarios, spanning advertising and marketing, film, television, and short plays and gaming. In advertising and marketing scenarios, CleanAI covers the entire visual production lifecycle. Early in the ideation phase, it rapidly generates near-final demos that turn creative concepts into tangible visuals. Those same visuals carry through to guide the final production shoot. CleanAI's model capabilities maintain strong ad character and product consistency across different scenes and camera angles, offering advertisers reliable visual support to establish a coherent creative tone. In film, television, and short play scenarios, CleanAI can be embedded into the entire production workflow, from script breakdown and concept design for characters and scenes to intelligent storyboarding and final video generation. enabling a highly automated creation process that significantly reduces production timelines and costs. The AI short film paper smartphone, which went viral in early April this year and accumulated over 100 million online views, was produced by two non-professional film creators in just three days using CleanAI. The creators leveraged CleanAI to generate almost all visual elements, covering the full pipeline from storyboarding to final cut. In gaming scenarios, CleanAI supports the project initiation phase by turning concept artwork into dynamic previews, saving visual effects artists considerable time for using animated mood boards. CleanAI also supports downstream production by generating assets such as gray box renders, in-game narrative sequences, and cutscenes.

Let's look at the future of video production models. In the first half of this year, we saw that AI technology is very popular. AI technology has made the supply of the software industry increase by several times. We also have reasons to believe that as the ability of video production models continues to break through, including the professional production content in DIMM, supply is also expected to explode. This will not only lead to more high-quality professional production content supply, but it may also be integrated into the content In terms of the outlook for the applications of large video generation models,

We have already seen a compelling signal so far this year. AI-generated short plays have shown strong momentum, with AI technologies driving multifold increases in content supply. We have also reasons to believe that as large video generation models continue to achieve breakthroughs, professionally generated content, PGC supply, including films, is likely to see the face of breakout growth. This would bring not only a dramatic expansion in high-quality professional content supply, but could also enable the integration of more personalized and interactive elements into content, such as bringing users' own identities and personalities into the content itself. This level of highly personalized content consumption could fundamentally reshape the relationship between users and content. Against this backdrop, CleanAI is well-positioned to unlock substantial commercialization potential. Overall, we are highly confident in CleanAI and the broader AI video generation sector, and we will continue to invest in CleanAI's computing power and talent development.

Thank you.

I'll read to our next question, please. The next question comes from Lincoln Kong of Goldman Sachs. Please go ahead. Thank you.

Thank you, Mr. Guan. That's my question. I want to ask, other than CleanAI, what other progress does our company make in other areas of AI this quarter, especially in the field of agents? So thank you, Benjamin, for taking my question. So my question is, other than CleanAI, what are the other areas we have seen progress on AI front this quarter, especially like AI agents, et cetera? Thank you.

In the first quarter of 2026, the company continued to deepen its AI strategy. In addition to allowing AI to continue to maintain technology, products, and commercialization, we also achieved good progress in the field of business environment and marketing with the use of AI technology. In the beginning of this year, the AI technology in the open cloud has attracted everyone's attention. I would like to focus on the progress of AI technology in various scenarios.

Thanks for the question. In Q1 2026, we'll continue to deepen our AI strategy. Beyond clean AI mechanics, global leadership in technology, product capabilities, and monetization, we'll also make solid progress in leveraging AI to empower our business ecosystem and improve organizational efficiency. At the beginning of the year, AI agents such as OpenCore attracted widespread attention. I'd like to take this opportunity to highlight the progress we've made with AI agents in different scenarios.

在电商营销服务场景,我们推出了覆盖信息推理、创意生产、产品三选和营销出价决策等环节的全六项agent。 在信息推理环节,我们通过AI生产者推荐, The user's behavior on the platform will be sorted and formed into a model of understandable language. The image broadens the user's interest in understanding. At the same time, it combines the behavior series of the user to discover the depth of user consumption. In the creative and product section, using Admin, you can select and select for the merchant. There are more expensive and expensive materials and products. On the flow allocation, give more exposure opportunities. In the price decision reasoning section, Agents can integrate historical data, current traffic trends, and different areas of traffic performance to predict how the price will make the advertising investment stable and efficient. On the basis of ensuring that the total ROI is achieved, the largest sales investment scale can be obtained.

In e-commerce marketing services, we launched end-to-end AI agents, covering user interest inference, creative production, product selection, and marketing bidding decision. In the user interest inference stage, we leveraged AI generative recommendation to consolidate user behaviors on the platform into representations that are interpretable by models. We expanded the breadth of user interest understanding and harnessed behavioral sequences to uncover deeper user consumption patterns. In the creative and product selection stage, our AI agents help merchants identify materials and products with higher potential to become blockbuster items, offering greater exposure through traffic allocation and ultimately improving the art pool of material placement. In a bidding decision inference stage, our AI agents leverage historical data, intraday traffic trends, and traffic pricing across different scenarios to predict optimal bidding strategies, ensuring stable and efficient ad placement. By empowering merchants to achieve their full-day ROI targets, we effectively maximize the overall marketing placement scale.

针对生活服务业务强境 We have launched a deep search engine and sales engine for our customers. With the ability of the large model, the search engine of the customer can realize a comprehensive understanding of the product and information of the search engine. At the same time, based on the needs of the user in the platform before the action of reasoning, it improves the transfer rate of customer sales materials and their stability. The deep search engine 可以通过大模型的长文本理解能力, 实现对用户意图的准确理解, 并让模型围绕深度转化目标进行优化, 提升线索转化效率。 而Social Agent能够结合商家知识库和行业知识库, 让其能够像专业的客户一样回答客户的问题, 满足多轮对话需求, 甚至实现全流程托管。

Tailored for the local services scenario, we have launched three specialized agents, the target user's exploration agent, the deeper conversion agent, and the sales agent. The target user's agent leverages the capabilities of AMS to gain a profound understanding of lead-based products and services. By inferring users' latent needs based on their behaviors on the platform, it significantly improves the conversion rate and cost stability. of client marketing placement. The deep conversion agent utilizes the long context and understanding capabilities of LLMs to accurately grasp user intent. It optimizes model performance specifically toward deep conversion goals, thereby boosting the efficiency of lead conversion. Finally, the sales agent integrates merchant-specific and industry knowledge bases, This enables it to respond to customer inquiries like a professional service representative, handle multi-turn conversations, and even achieve fully autonomous end-to-end workflow management.

在组织赋能方面,由CodeFlicker 3.2的MyFlicker 1成为快手员工的AI工作火炸。 While Flickr has been incorporated into several leading large-scale models in the industry, it has the usual response to the core capabilities of planning, memory, and use of Agents. It can also be used through skills trading tools or trace agents to complete various complex work tasks. Next, we will continue to promote AI capabilities, infiltrate employees' daily work scenarios, and upgrade employees' work experience. In short, we will continue to increase AI technology,

In terms of organizational empowerment, MyFlicker, the upgraded version of CodeFlicker, has become the AI work companion for casual employees. It integrates the multiple industry-leading large models and possesses core capabilities of a general-purpose agent, including proactive response, reasoning, planning, and memory. Furthermore, it can leverage skills to invoke tools or vertical agents to handle various complex tasks. We're committed to continuously embedding AI capabilities into employees' daily work scenarios, significantly elevating their overall work experience. In summary, we will continue to accelerate the deep integration of AI into our business operations and organizations, creating greater commercial value and growth opportunities for both the company and our partners. Thank you.

Next question, please, operator.

下一条问题是来自Jeffries的Thomas Chong. The next question comes from Thomas Chong of Jeffries.

Please go ahead,请提问。 晚上好,谢谢华立诚介绍我的提问。 我们看到去年四季度公司完成了电商与商业化流量的融合。 请问在融合后,在六年一季度,我们看到了哪些成效? Hi, good evening. Thanks, management, for taking my question. We have seen companies completed the integration of e-commerce and commercial traffic in 2.4 last year. May I know the results of the integration we see so far in 2.1? How should we think about the monetization in e-commerce for the full year 2026? Thank you.

The other question is that the core of the integration mechanism of e-commerce and commercialization traffic is to improve the connectivity of e-commerce and e-commerce ads. In particular, the first flow helps the business to expand and gain advantages in e-commerce traffic. The second is to manage well. High-end GMs and business owners can also get more exposure in commercialization traffic. After the flow of traffic is integrated, we see that the e-commerce traffic is rising, the buyers are rising, and the business owners are rising. In the first quarter of 2020, e-commerce commercialization traffic still maintains a growth of high unit number. Effective marketing investment continues to stimulate and break down the interest of e-commerce buyers, leading to the increase of e-commerce marketing sales by more and more active buyers. At the same time, the increase in the number of buyers of heavy sales has presented a high growth trend in the supply chain market. The flow of integrated machines has increased and expanded, and the use of jumpers has increased. Thank you for the question.

The integration between e-commerce and commercial traffic was fundamentally designed to align our e-commerce with e-commerce advertising distribution. To be more specific, first, marketing placement helped merchants amplify their advantages in acquiring traffic within e-commerce. Second, high GPM merchants with strong operating performance were also able to secure more advertising traffic. Following this integration, we achieved increases in e-commerce traffic, our buyer base, and a number of merchants across Omni domain scenarios. In Q1, e-commerce commercialization traffic maintained high single-digit growth. Effective marketing initiatives continuously stimulated and expanded e-commerce buyer interest. driving a more than 20% year-over-year increase in the number of e-commerce monthly active paying users attributable to marketing. While the number of active merchants using marketing placements shows strong growth momentum, increasing by nearly 40% year-over-year, this mechanism has amplified the scaling leverage available to brand merchants, enabling them to achieve breakthroughs in operational scale while contributing to a healthier merchant structure. As mentioned earlier, we launched the Voyage initiative last year, benefiting from our own strategic focus this year on brand acquisition and traffic synergy. GMV generated for brand merchants across Omni domain scenarios increased by more than 25% year-over-year, while advertising standing in brands under the Voyage initiative increased by 42% year-over-year.

We will continue to deepen the full-time promotion, full-time control, and other smart investment technology capabilities, combine the flow of integrated infrastructure, focus on resources to introduce high-quality products and content, and promote the brand and performance of long-term business capabilities and wishes on the platform, We believe that this is a good basis for the development of the business industry.

As the micro-consumption environment has yet to show significant recovery and compliance policies have weighed on operational efficiency for certain merchants on our platform in the near term, we'll continue to strengthen the foundational capabilities of our intelligent placement solutions, including our omni-platform marketing solutions and our food store hosting. We're further rolling out our traffic integration mechanism and will focus our resources on introducing high-quality products and content supply, while consistently offering traffic support to brands and merchants with balanced profiles who are capable of and committed to sustainable long-term operations. Although in the near term, our overall e-commerce monetization rate growth may get impacted slightly, we see this as an opportunity to strengthen our merchant ecosystem. Over the long run, I would believe this structural adjustment to the merchant ecosystem will lay a solid foundation for the healthier and more sustainable development of our business ecosystem.

Thank you. Next question, please, operator.

下一道问题是来自美银证券的Miranda Zhuang. The next question comes from Miranda Zhuang of Bank of America Securities.

Please go ahead, 请提问。 喂,大家好,感谢管理层接受我的提问, 然后恭喜亮眼的业绩。 我想请教一下就是因为我们最近看到AI漫剧能做飞快,那请教管理层怎么看背后的核心驱动因素,那我们应该如何思考未来增长的持续性能? 谢谢。 My question is about the AI video content. So the AI comic style short plays have been growing very fast recently. How does management view the drivers for this trend? And how do you think about the growth of sustainability into the future? Thank you.

另一个问题是我们大家所看到的 近期快手AI漫剧业务 与行业一样迎来了好发性的增长 26年的第一季度 快手AI漫剧的营销消耗同比增长超过了100倍 完比增长超过了150 截至3月底 快手平台上的AI漫剧 Thank you for your question.

As you noted, in Q1, marketing spend on cultural AI comic style short plays grew more than 100 times year over year and more than 150% quarter over quarter. As of the end of March, the peak of daily marketing spend on special AI comic style short plays exceeded 20 million RMB. The core driver behind this growth is AI's ability to substantially lower production costs, enabling content supply to expand rapidly, and subsequently driving stronger user consumption and marketing placement demand.

第一,从供给观看,AI显著降低了漫剧的制作成分和创作门槛。 AI has been able to cover a variety of settings such as depth, depth, effect, dubbing, and post-production editing. Actors, scenes, and shooting equipment are not required. The production cycle is also significantly shortened. Three to four weeks are enough to complete it. Open data shows that the cost of AI comics is about 70% of the cost of 8-15 million yuan, which is called the support budget. This makes it possible for more creators and institutions to enter this track and drive the supply of content. According to three-way data, in March and March of 2026, the top limit of AI comics was about 4.7 million copies, exceeding the short limit of 3.3 million copies in the whole year of 2025. In the first quarter of 2026, the number of Japanese first-rate comics on the fast-tracking platform increased by 215%. The expansion of the supply size promoted the growth of AI comics consumption. Second, from the demand point of view, First, on the supply side, AIS has significantly reduced production costs and lowered barriers to entry for creators.

AI can now handle multiple stages of production, including character design, storyboarding, special effects, dubbing, and post-production editing. It eliminates the needs for actors, filming locations, and production equipment, which is significantly shortening the production cycle to just three to four weeks. Public data shows that the production cost of a single AI comic-style short play ranges from $80,000 to $150,000. approximately 70% lower than traditional short plays. This cost advantage has lowered the entry barrier for creators and studios, driving rapid expansion in content supply. According to third-party data, around 47,000 AI comic style short plays were released in March alone, surpassing the 33,000 short plays that were released over the entirety of 2025. In Q1 2026, the average daily number of comic-style short plays with active marketing placement on quite a platform grew by 215% over a quarter. This massive supply expansion has propelled growth in marketing spending on AI comic-style short plays. Second, on the demand side, AI comic-style short plays not only cater to the colder needs of existing short play viewers, but also expand into broader user acceptance, especially in During such as fantasy, sci-fi, and science fiction, AI commissarial plays can transcend the cost constraints and technical limitations of live action production, therefore making more diverse content consumption needs.

In addition to the industry trend, the growth of AI comics is also related to the level of investment in the platform. First of all, as early as 2025, we began to pay attention and invest in the AI comic market, providing a level of flow support and new opportunities. In terms of content supply construction and commercialization, we have done a thorough layout. In addition, around the supply side, we continue to operate the toy ecosystem to attract more consumers to participate in the content production and expand more small and medium-sized toy workshops. In addition, we are also promoting the combination of small and medium-sized copyrights and slow-changing resources to help UZIP quickly transform into UZI toy content. At the same time, our mature content consumption and commercialization chain

The AI broader industry needs, the fast growth of AI comics.shoreplay is also closely tied to our continued effort in this category. First, as early as 2025, we began to closely track and invest in this emerging segment. We provided 100-minute scale traffic support and cash incentives while laying the groundwork for both content supply deployment, development, and commercialization. Second, on the supply side, we have been continuously developing the UGC ecosystem to bring more creators and more small and medium-sized activities into AI comic-style short plays. We're also helping connect novel copyrights with comic adaptation resources so that high-quality IP can be turned into high-quality AI comic-style short play content more efficiently. Meanwhile, Kuaishou already has a mature content consumption and commercialization ecosystem. This allows us to better support the growing content supply, distribute quality content more effectively, and further amplify their commercialization potential.

In the future, regarding the sustainability of AI management growth, we will focus on the optimistic attitude of growth in the medium term. The third-party data predicts that in 2026, the domestic comic market size is expected to break through 300 billion yuan, and in 2030, it will exceed 850 billion yuan. In terms of speed, we have confidence in the growth of the content and commercialization capabilities, so we are confident that through the deep integration of content and technology, we will promote the continued growth of the platform AI comics.

Looking ahead, we remain optimistic about a continued growth of AI comic-style short plays, and we expect to see robust growth in the medium term. Third-party data projects that the market size for AI comic-style short plays in China will exceed 30 billion in 2026 and surpass 85 billion by 2030. As a question, we are well-positioned with a vibrant user ecosystem, industry-leading video generation models, and mature capabilities in content distribution and commercialization. As such, we are confident that through the deep synergies between content and technology, we will drive the same growth in AI comic style trophies on our platform.

Thank you. Maybe the last question, please, operator.

Thank you. The last question comes from Xueqin Zhang of CICC. Please go ahead.

Thank you for accepting my question. My question is about finance. I would like to know how the company's cash flow management policy is. How about the future plans on capital investment and shareholder return? Will this affect the cash flow situation of the company? Thank you. Thanks management for taking my question. My question about financials. Can management share your cash flow management policy? And what are the plans for CapEx and the shareholder returns going forward? And will this test have any impact on the company's cash flow position? Thank you.

好的,谢谢你的问题。 公司一直保持着谨慎的资金管理策略, 在考虑安全性和流动性的前提下, 最大限度提高资金的管理效率。 公司预计今年的资本开支260亿元人民币, 目前来看,指引暂不会有调整。 Thanks for the question.

The company has consistently adhered to a prudent capital management strategy, maximizing capital efficiency while ensuring safety and liquidity. We still expect a pay tax of approximately 26 billion this year, and currently there is no update to this guidance. We expect the majority of the total capital expenditures will be incurred in the first half of this year. In addition, we have taken proactive measures to build advanced procurement and inventory buffers amidst rising competitive power prices. This is a deliberate step to manage market volatility allowing us to better control procurement costs for computing resources when practice moves upward.

那么我们始终重视为股东创造可持续的长期价值, 并坚定不移地执行积极的股东回报策略, 通过持续且稳定的分红与股份回购等多元化的方式, 切实回馈广大股东的信任与支持。 This reflects our firm faith in the long-term development of our company's future business and the ability to create strong cash flow. Based on the annual share price of 3 billion Hong Kong dollars this year, we will continue to actively carry out share return. We expect that the share return amount including share price and share return in the entire year of 2026 will be further increased compared to last year, and the share return rate will be further increased to about 4%.

We consistently prioritize the creation of a sustainable long-term value for shareholders. We remain firmly committed to our proactive shareholder returns policy. Through a combination of sustained and stable dividends, share repurchases, and other diverse mechanisms, we deliver tangible returns to our shareholders for their trust and support. This demonstrates our firm confidence in the positive development of our business over the long term and our strong cash flow generation. Building on an annual dividend of HK$3 billion this year, we will continue to actively repurchase shares. We expect total shareholder returns in 2026, including dividends and share repurchases, to increase compared to last year, with an overall shareholder return yield of around 4%.

从现金流的情况来看 尽管我们有260亿人民币的资本支出 我们的目标依然是继续保持全年集团层面顺向的自由现金流。 我们将以当下的精准投入撬动未来的影帝增长。 在坚定进行AI战略投入的同时,我们将严格把控财务风险,保持健康的现金储备,以稳健的财务结构为AI时代的长期高质量成长保驾护好。

In terms of cash flow, even with $26 billion in CAPEX, we still aim to maintain positive free cash flow at a group level for a full year. By executing high-precision targeted investments, we aim to unlock greater profitability growth in the future. As we firmly invest in our AI strategy, we will strictly manage financial risks and maintain a healthy cash reserve, ensuring a solid financial structure to safeguard our long-term, high-quality growth in the AI era.

Thank you, operator. That's the end of the Q&A session. Thank you once again for joining us today.

If you have any further questions, please contact our capital market and investor relations team anytime. Thank you.

Thank you.