Kuaishou Technology B

Good day, ladies and gentlemen. Thank you for standing by. Welcome to the Quasar Technology second quarter and interim 2026 Financial Resource Conference call. Please note that English simultaneous interpretation will be provided for management's prepared remarks. This English line will be in listen-only mode. I will now turn the call over to Mr. Matthew Zhao, VP of Capital Market and IR at Kuaishou Technology. Thank you, operator. Good evening and good morning to everyone. Welcome to Kuaishou Technology second quarter and interim 2026 financial conference call. Joining us today are Mr. Cheng Yixiao, co-founder, chairman, and CEO, and Mr. Jin Bing, our CFO. 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 obligation to update any forward-looking information except as required by law. For all important information about this call, including forward-looking statements, please refer to the company's public information where the second quarter and interim 2026 results announcement ended at June 30th, 2026, issued earlier today. During today's call, management will also discuss certain non-IFRS financial measures. These are provided for additional information and should not replace IFRS-based financial results. For a definition of non-IFRS financial measures and a reconciliation of IFRS to non-IFRS, Financial Results and Related Risk Factors, please refer to our second quarter and interim 2026 results announcement. For today's call, management will use Chinese as the main language. A third-party interpreter will provide a simultaneous English interpretation in the preparatory remarks session and a Q&A and a consecutive interpretation during the Q&A session. Please note that English interpretation is for convenience purposes only. In case of any discrepancy, management's original statements will prevail. Lastly, and last otherwise stated, all currency units mentioned are in RMB. I will now hand the call over to Yixiao. Hello, everyone. Welcome to Huaxia's second quarter 2026 earnings conference call. In Q2, amid a complex macroeconomic environment and industry competition, we remain committed to our long-term vision and strategic AI investments, and we achieve high-quality growth. In Q2, the average DA use on the Huaxiao app reached 412 million. Total revenues increased by 1.4% year-over-year to 35.5 billion RMB. Revenues from our core commercial business, including online marketing services and other services, primarily e-commerce and clang AI, increased by 7.4% year-over-year. Adjusted net profit reached 3.9 billion RMB with an adjusted net margin of 11%. Overall, profitability remains stable, further demonstrating the resilience of our business operations. Now, I'll elaborate the progress of each segment in Q2. are a strategy in progress of our larger video generation model, Kling AI. In Q2, Kling AI continued advancing its vision of empowering everyone to craft capability stories with AI. Through model breakthroughs, product feature upgrades, and a global creative ecosystem expansion, Kling AI reinforced its global leadership in multimodal video generation. Kling AI launched its native 4K video output in Kling AI 3.0 series As the industry's first video generation model to support native 4K output, it enables one-click generation of cinema-grade 4K video. Designed for film, TV, and advertising professionals, it delivers high-resolution visuals without complex post-production, achieving industrial-grade cinematic visual effects. Kuling AI also released the 3.0 Turbo model, which can maintain stable, high-quality dynamic output and precise audio-visual sync while improving creative efficiency and reducing production costs. Kuling MCP and Kuling CLI were officially launched as well, enabling AI agents to dispatch Kuling AI for batch content creation and expanding its use case in workflow automation and intelligent orchestration. Culling AI continues to empower professional content creation with its technical innovation and creative achievements, earning broad industry recognition. At the 2026 Cannes Lions, two advertising videos generated by Culling AI won one silver line and two bronze lines, demonstrating recognition of its creative capabilities by one of the world's premier creative awards programs. At the 2026 Beijing International Film Festival, multiple Kling AI-created works, including Paper Smartphone, were selected for the AIGC section. Stroke of Genius, with Liu Cixin serving as Literal Supervisor, won the annual Featured Work in Shortplay Micro Shortplay Award, highlighting Kling AI's strengths in empowering professional filmmaking and content creation. Driven by model breakthroughs, continuous product enhancements, and deeper penetration, Ke Ling AI's commercialization maintained a strong growth momentum. In Q2, Ke Ling AI generated a revenue of over 850 million RMB, up over 200% year-over-year, continuing to lead a global AI video commercialization. In Q2, we made a solid progress advancing the research and application of our general purpose large models. We released Ke-E VL 2.0, and I've created a multi-model model that enables deep perception across 256,000 ultra-long contexts and delivers nearly lossless reasoning for long video temporal understanding. It also became the first Koei-based model with a built-in agent collaboration mechanism demonstrating the potential for code parsing and a tool invocation. We introduced Agent X, a self-evolving AI agent for industrial recommendation systems that autonomously handles recommendation model and strategy design, performance evaluation, and knowledge accumulation, significantly improving iteration efficiency of our recommendation algorithms. It enables recommendation systems to autonomously drive recommendation model and strategies design, evaluate performance, and accumulate insights, significantly boosting the iteration efficiency in our recommendation algorithms. In Q2, we developed scenario-based agent capabilities for generating marketing materials across our online marketing services. . . . . . . integrated skills across key internal system began serving all employees. By June, over 92% of employees were using our AI agents, and the AI code contribution rate among R&D engineers reached 60%. Meanwhile, Wenqing, our enterprise-grade large model platform, integrates high-performance model inference, cost-efficient model customization, and fully management service. Wenqing supports Huaxiao's internal AI use cases and also external Enterprise clients with large model infrastructure. On Open Router, a global AI model aggregation platform, several vending hosted open source models ranked among the top by API call consumption. Second, user growth and a content ecosystem. In Q2, average use on the Quasio app reached 412 million. Average AMA use reached 797 million. We leverage AI-powered smart placement to improve user acquisition efficiency and boost retention among new and reactivity users. We consistently refine our traffic management to better serve our highly active core users. We also continue to emphasize our social features. Users with mutual followers engaging in private messaging grow over 15% year-over-year. We also optimize the Quasio app's core features, comprehensively elevating the user experience through systematic improvement We believe in the power of community and remain committed to strengthening our differentiated, high-quality content ecosystem. In June and July, we leveraged the World Cup buzz to launch our native IP, Huaishou World Cup Fans Trophy. Beyond covering trending topics, we launched original events, including the Huaishou XY Style FC and Dream Chasers Youth Football Tournament. Yixiao Cheng, Hua Su, Bing Jin, Hoi Lam Yip using a joint operation model to deliver more high-value content to users. We used an e-commerce live streaming model to secure live broadcasting rights for the 2026 CBA season. We also introduced a paid live streaming model for online music performance. In April, we hosted a TOP concert, generating over 10 million RMB in sales, achieving a synergy between content and commercialization. This model also driven grassroots sports events, achieving a notable regional scale, especially in northwest China. Third, online marketing services. In Q2, revenue from online marketing services reached a 20.6 billion RMB, up 4.4% year-over-year. Our non-e-commerce marketing services continued expanding across content consumption, lifestyle services, and AI applications. Supported by our Omni domain traffic synergy strategy and dedicated programs for brand merchants, e-commerce marketing services remained resilient. We also continue deepening AI applications across the full marketing services lifecycle. In Q2, the content consumption, lifestyle services, and AI applications continued to drive year-on-year growth in our non-e-commerce marketing services revenue. In content consumption, AI lowered production costs and reduced creation barriers, driving rapid short-play supply growth that catered to a broader user preferences. This enriched our content ecosystem and boosted related marketing demand. By June, short-play supply on Kuaishou, both live action and AI-generated, had grown over five-fold from January. In Q2, short-play-driven marketing spend grew over 100% year-over-year. In lifestyle services, we deepen our presence in sub-verticals like comprehensive and local services while exploring incremental growth opportunities. We also optimized the deep conversion capabilities, enhanced the full stack leader-driven marketing solutions, and launched a user cohort exploration AI agent. These efforts helped clients to more effectively identify high-intent users, improving lead quality and conversion. In AI applications, we worked closely with the clients to align ad placement with in-app conversion, helping them to improve user retention and conversion. This further strengthen our competitiveness in capturing ad spend from AI application clients. For e-commerce marketing services in Q2, we strengthen our Omni domain traffic synergy across e-commerce and commercialization business to improve merchant traffic matching efficiency. We conducted a more granular merchant segmentation with tiered operations, tailoring product strategies to address merchants' core needs. We also took a content supply side approach by managing marketing materials, including incentivizing first launch content and increasing recommendation diversity. These initiatives optimized the e-commerce marketing material content mix, enabling high-quality content to reach relevant traffic more efficiently and improving our long-term commercial ecosystem. Despite macro and merchant challenges, we remain committed to traffic support for high-quality merchants. The T2000 brand initiative launched in Q4 last year has delivered promising early results. Brand merchant marketing spend outperformed our broader e-commerce marketing in Q2. Its revenue contribution continues to grow. At the product level, our net Transaction ROI product continued to evolve by enhancing omni-scenario transaction bidding capabilities and refining bidding mechanisms and model strategies. Client penetration rose from 45% in Q1 to 55% in Q2, effectively helping merchants reduce return rates. In Q2, we continued optimizing AI applications across industry-specific scenarios, improving client marketing placement efficiency and strengthening our capacity to capture incremental marketing budgets across sectors. In content consumption through content understanding, user matching, and a smart placement, AI helped quality content reach users more efficiently. In lifestyle services, AI is applied to marketing material generation, digital human live streaming, conversational business operations, user intent identification, and deep conversation prediction. These help merchants lower costs across content creation, placement, and customer services. Number four, our e-commerce business. In Q2, we advanced our strategy across three areas, growing our paying user base, expanding supply, and deepening e-commerce and commercialization traffic integration. We optimized our merchant ecosystem and mix, strengthened brand and new merchant acquisition and their growth, growth synergies between e-commerce and commercialization. During the quarter, we focused on growing high-quality buyers, While active paying users remained largely stable quarter over quarter, as the users' omni-domain consumption habits continued developing, we strengthened our private domain advantages by aligning traffic across diverse scenarios. This enabled a content-driven product recommendation, shelf-based conversion, store repurchases to reinforce one another in a positive growth cycle. We also enhanced cross-scenario synergies and optimized the subsidy efficiency, driving balanced growth across content-based and pen-shelf-based scenarios. On the supply side, in Q2, we continued onboarding new merchants and advancing brand expansion through cost reduction, efficiency improvement, growth incentives, product empowerment, and operational support. We helped the new and small and medium-sized merchants grow while further improving our merchant mix. We launched our upgraded Starlight initiative, offering tiered support programs for brand merchants, large merchants, industrial zone merchants, and SME merchants, helping more merchants scale faster. Supported by these initiatives, newly onboarded merchants grew year-over-year and rose nearly 10% quarter-over-quarter. New merchants achieving scaled growth in their second month rose nearly 30% year-over-year, reflecting continued improvements in new merchant quality. On the brand merchant side, self-operated GMV from T2000 brands maintained strong year-over-year growth, while their contribution to Omni domain GMV steadily increased. Marketing spend on the brand and commercialization also grew rapidly year-over-year, further boosting brand merchants' contribution to the both overall e-commerce GMV and online marketing revenue. By industry leveraging content-based e-commerce trends, merchants counting tea, alcohol, and health products, beauty, and cosmetics, and fresh food continued growing, unlocking structural growth opportunities. We continue to improve our catalog ecosystem structure, enhancing content supply quality, We deepened collaboration with the top tier KOLs, increasing support for mid-tier KOLs in our strong verticals like 3Rural and NME, and improved the consistency of existing KOLs performance, reinforcing our e-commerce content foundation. By integrating KOL resources with the distinctive product offerings nationwide, we deepened our penetration in industry zones and launched content marketing initiatives like product origin, tracking, live streams. These efforts strengthened the synergy content and supply, empowered KOLs and improved the merchants conversion. We also expanded our KOL base through in-platform incubation, talent agency partnerships, and External Acquisition. To boost the streaming frequency, we refined our incentive policies. In Q2, the number of streamers with over 10,000 followers grew year-over-year, while KLF streaming frequency continued to increase steadily. On the distribution pool development, we leveraged the AI to enhance product capabilities, creating a more targeted system, further boosting the vibrancy of our distribution system. In Q2, active KOL distribution penetration continued rising year-over-year, and merchant KOL merchants' matches grew over 20% year-over-year. In Q2, throughout the full lifecycle of merchants, we continued optimizing AI capabilities across e-commerce scenarios, helping merchants reduce costs, improve efficiency, and driving intelligent operations. These initiatives validated AI's evolution from a productivity tool into a comprehensive business execution solution. In Q2, over 850,000 merchants used our free AI business tools across product selection and listing, marketing material creation, business analysis, smart placement, and AI-powered customer service. These AI tools provided merchants with end-to-end operational support and capability enhancements. Next, our live streaming business. Q2, live streaming revenue reached 8.7 billion RMB. We focused on supply side health and leveraged AI to empower live streaming products, driving ecosystem quality and product integration. On the supply side, we launched the Confluence Initiative, providing streamer acquisition incentives, early stage growth support, and ecosystem governance to steadily expand the supply of new streamers from talent agencies and improve their early traction efficiency. We also strengthened independent streamer operations, focusing on identifying high-value independent streamers to solidify our live streaming supply foundation. We also encourage the top streamers to expand into Group 5 live formats, leveraging their traffic and influence to enrich high-quality live streaming content supply. On the product and technology front, AI further empowered live streaming rooms powered by clean AI. AI gives with customizable special effects continue to evolve, offering more formats and capabilities and boosting users' willingness to pay. In Q2, AI gives a send to buy users past 6 million. AI-driven content understanding continued optimizing our live streaming recommendation strategies, enabling more precise matching between streamers and users, supporting paying users' growth. Intelligent live streaming gift recommendation and ranking features based on real-time multimodal signals improve the user's payment experience and efficiency. AI tools like AI interaction of systems and digital avatar solution will further refine improving streamer service efficiency. Finally, our overseas business progress. In Q2, we remain committed to high-value growth strategies, strengthening our overseas foundation in profitability, long-term operation, and localization. On traffic and content, we maintain refined user acquisition in host local content and expanded community creator networks, fostering an engaging atmosphere around real-life scenarios and deepening content consumption among core users. For overseas online marketing services, we capitalize on major events, such as the Fista Honita and the World Cup, driven by AI, in-depth ROI analysis, user group insights, innovative product features, and industry-specific strategies. We helped marketing clients capture key marketing periods and achieve rapid growth. We also unlocked the monetization potential in short plays and other content formats. Together with our marketing services capabilities, this formed a dual engine growth model while accelerating expansion into growth sectors such as e-commerce, Our e-commerce, e-commerce, GMV, and order volume continue solid year-over-year growth in Q2. At the same time, we drove growth in average order value through product mix optimization and quality supply while maintaining solid operational efficiency and profitability. Looking ahead to second half, the middle growing external challenges will remain steadfast in advancing our core AI strategy, leveraging our technological and ecosystem strength to navigate headwinds while we'll face short-term revenue pressure and heavy investments, AI investments, We will maintain our long-term focus and continue expanding the commercial exploration of AI to empower crisis content and commercial ecosystems. As we pursue near-term breakthroughs and high-quality growth over the long term, we remain committed to creating long-term value for our users and platform partners. That concludes my prepared remarks. Next, I will hand it over to Junbing, who will review the company's financial update for the second quarter. Thank you, Yixiao, and hello, everyone. In Q2, we continue to deeply integrate AI technologies across our business scenarios. As a result, we achieved high-quality growth throughout our overall business amid a complex environment. The comprehensive application of AI has become an engine driving the company's long-term development. It'll continue to empower our content ecosystem, enhance user experience, our providing merchants and advertisers with end-to-end intelligent business operation tools. For our platform partners, these capabilities helped to reduce costs and improve efficiency. It also injected a new growth of momentum into our business. In Q2, the group's total revenues reached 35.5 billion RMB, adjusted net profit reached at 3.9 billion RMB in the second quarter, with an adjusted net margin of 11%. The group's overall profitability remained at a healthy level. Now let's take a closer look. Our total revenue grew 1.4% year-over-year to 35.5 billion in Q2. The increase was mainly driven by growth of our online marketing services and acclaimed AI business. Online marketing services revenue increased to 4.4% to 20.6 billion in Q2 from 19.8 billion in the same period last year. This growth was primarily attributable to deepening application of AI in online marketing services, which effectively enhanced our clients' marketing placement efficiency, driving more marketing spend. Revenue from other services, including e-commerce and clean AI business, reached 6.2 billion in Q2, up 18.5% from 5.2 billion in September last year. The increase was mainly driven by continuing expansion of our clean AI business. As clean AI achieved breakthroughs in model capabilities, continuous product enhancements, and deeper penetration across professional creative scenarios, its commercialization continued to maintain strong growth momentum. In Q2, our live streaming revenue was $8.7 billion. We consistently cultivated high-quality content, leveraged AI-powered product innovations, and enhanced the quality of our ecosystem to develop a rich and healthy live streaming ecosystem and diverse high-quality content. Cost of revenue increased to 10.7% year-over-year to $17.2 billion in Q2, accounting for 48.4% of our total revenues. The increase was mainly due to high revenue sharing costs and related taxes in line with our revenue growth. Based on the Evolve, our gross profit was 18.3 billion in Q2 compared to 19.5 billion in the same period last year. Gross profit margin was 51.6% compared to 55.7% in the same period last year. Turning to expenses in Q2, selling and marketing expenses were 9.9 billion compared to 10.5 billion in the same period last year. Selling and marketing expenses decreased to 27.9% of total revenues from 30% in Q2 last year, primarily attributed to the lower spending for promotional activities. R&D expenses increased to 34.7% a year to 4.6 billion, accounting for 12.9% of total revenues. The increase was mainly due to increased investments in AI, including related training costs. Administrative expenses were 895 million, Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Huaxia Zhao, Hua Su, Bing Jin 30th, 2026. Net cash generated from operating activities in Q2 was 5.9 billion. Additionally, we actively leverage and deliver on our commitment to shareholder returns based on market conditions. As of today, we had repurchased approximately 1,970 million Hong Kong dollars, or around 43.3 million shares, representing about 1% of our total shares of sitting for 2026. Looking ahead to the second half, As Yixiao mentioned, amid external challenges, we expect to face near-term challenges. As pressure on revenue and our continued investment in AI both weigh on profitability, we will continue to uphold our technology-driven, user-centric self-philosophy while maintaining deeply focused on our users' needs. We remain staunchly committed to advancing AI and leveraging our leading AI technologies to further empower our content ecosystem and commercial value chain. At the same time, we will apply prudent financial displaying to reduce costs and improve efficiency. This will further strengthen the company's core competitiveness mode and create long-term value for our users, partners, and shareholders. That concludes our preparatory box. Now we can open for the call for Q&A. And the phone of UPS. Please ask your question.

请提问。 谢谢管理员介绍我的提问。 也恭喜我们有一个非常稳健的主业 还有肯定的非常强劲的增长。 Thank you management for taking my questions and congrats on the very stable cooperation and robust growth of Kerlin. I have a question regarding Kerlin competitive landscape and the iteration direction. Recently, multiple video generation large language models have been updated successively. So how should we view the current competitive landscape of the video generation models and what is Kerlin competitive strategy? Thank you.

Thank you, Renjia. The global video production channel is a huge market with a value of $1.5 billion. The AI video production model has a great potential for penetration. We believe that the AI video production channel is currently a turning point in the development of the industry. Many players rely on their own platform ecology, vertical scenarios, and technical skills to compete. They continue to expand in advertising, e-commerce, Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

Thank you for your question. The global market for video content generation represents a massive 150 billion US dollars opportunity. AI video generation models have substantial potential for user adoption. We believe the AI video generation segment is currently in a vibrant phase with the diverse stakeholders maximizing their strength to drive industry upside. As industry players leverage their respective platform ecosystems, vertical scenarios, and technological capabilities to compete through differentiation, the commercialization boundaries of AI video generation continue to expand across diverse scenarios, including advertising, e-commerce, films, short plays, and gaming, compared with relatively Thank you very much.

In the past two years, CleanAI has been the first team in the production channel of stable AI video generation. The CleanAI team continues to have top-notch strategic judgment and mobility in the industry. In June 2024, we launched the world's first DIT structural video model product that can be used in warehouses. In April 2025, we released the world's first multi-modal visual language MVL communication system. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

Over the past two years, Kling AI has consistently remained among the Tier 1 players in the AI video generation segment. The Kling AI team continues to demonstrate strategic foresight into the industry and strong execution capabilities. In June 2024, we launched the world's first commercially available video model product based on the DIT architecture. In April 2025, we introduced the world's first multi-modal visual language interaction architecture, NVL. In December 2025, we released the world's first omni-model-based multi-modal video generation model. Our track record has consistently proven that the Kling AI team is among the industry's top teams worldwide, leading with strong research, engineering, and strategic execution capabilities. Originally, Cling AI closed an independent financing round, which will further enhance its competitiveness in the industry.

Cling is a company that focuses on service and professional content creators. Through video production model technology, it improves the professional community's work efficiency by using video creation as a profession. CLEAN AI has a strong prompt understanding and controllability ability to help users achieve accurate and consistent creative expression. In terms of image performance and production extension, CLEAN AI can also meet the needs of creative creators for high-quality video content.

CLEAN AI is the world's first video model to achieve native 4K.

In this launch, CleanAI has focused on serving professional content creators by leveraging video generation large model technology to improve the productivity of professional creators. who make video creation their career and have sustainable purchasing power. Kaling AI features strong, prompt understanding and controllable store-bought capabilities, enabling users to achieve precise and coherent creative expression. In terms of visual quality and production scalability, Kaling AI also satisfies professional creators' needs for high-quality video content. Kaling AI is the world's first video generation model supporting native 4K output. Users can generate 4K videos with a single click without the need for additional upscaling or post-processing, resulting in clearer visuals, richer details, and a more cinematic look and feel.

In short, we are full of confidence in the long-term competitiveness of CLEAN AI in video production channels. With the development of multi-capacity, product experience, and professional creators' ecological and commercialization scenarios, CLEAN AI will continue to play an important role in the rapid penetration of the AI video production industry.

Overall, we are highly confident in Culling AI's long-term competitiveness in the AI video generation segment as model capabilities continuously iterate, product experience continues to improve, and the professional creator ecosystem and commercialization scenarios further expanded. At the same time, the AI video generation sector is rapidly gaining broader adoption, and Culling AI will consistently unlock greater growth potential. Thank you.

Thank you, operator. Next question, please.

下一道問題是來自高盛的Lincoln Kong. The next question comes from Lincoln Kong of Goldman Sachs. Please ask your question.

請提問。 謝謝關於曾接受我的提問。 我是想問關於AI戰略。 除了Clean AI以外,這個季度在其他的AI方面的業務有哪些進展?

Thank you, Benjamin, for taking my question. My question is also about AI strategy. So other than clean AI, this quarter, what are the other areas in terms of the AI development Quaishou has seen?

Thank you. Thank you, Benjamin.

During the quarter, our AI advancements extended beyond calling AI iteration. We also made tangible achievements across organizational efficiency, AI applications in online marketing services scenarios, and our core recommendation systems iterations.

在线上营销服务落地方面, We have integrated AI deep into AI-based material production, smart pricing, and the key parts of recommendation when producing. At the material production stage, we use user-interested modeling and video production to realize the upgrade from marketing video material to creating video for users. At the smart pricing stage, the agent system can combine the investment account history and transformation goals of marketing customers, provide self-learning, and continue to modify the automatic investment strategy. Regarding AI applications in online marketings,

We have deeply integrated AI across key processes, including the generation of AIGC marketing materials, intelligent bidding, and generative recommendation. In the generation of marketing materials, we connected user interest modeling with video production capabilities, enabling an upgrade from searching for marketing material for videos to creating videos tailored for users. On the intelligent bidding front, leveraging marketing clients' historical account data on marketing placement and conversion goals. Our agent system develops self-learning and continuously optimized automated bidding strategies. This effectively improved long-term customer value and marketing placement performance, delivering strong ROIs. For generative recommendation, our models can truly understand advertising content and user needs. By converting products, live streams, search queries, and industry information into semantic representations, we improved user product matching efficiency.

在推荐系统迭代方面,我们推出了Agen驱动的研发闭环产品Agen X. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

Regarding the iteration of our recommendation systems, we launched AgentX, an agent-driven R&D closed loop. In the past, taking a recommendation strategy from idea to launch often involved multiple steps, including data analysis, solution design, product code modification, experiment configuration, A-B testing and monitoring, metric attribution, and post-launch review. This process relied heavily on manual execution by algorithm engineers, which limited efficiency. Today, our agent serves as the execution engine for recommendation iteration, freeing engineers from a wide range of repetitive tasks and enabling them to focus on goal setting, critical reviews, and higher level judgment and decision making. As a result, we significantly enhance both the iteration efficiency and performance of our recommendation strategies.

In terms of the overall efficiency and capacity, the company has already exceeded 90% of the employees' usage rate, including H&M's products, including MyFlickr. It has greatly improved the internal performance of many departments such as technical research, data analysis, business operation, and so on. Taking the technical research performance as an example, in the second quarter of 2020, the average demand for technical staff was reduced by more than 4% in the first quarter. In terms of organizational efficiency improvement and organizational enablement,

Our in-house developed agent tools, including MyFlickr, have achieved over 90% employee adoption. They fully cover diverse functions including technology R&D, data analytics, business operations. These tools have significantly enhanced our internal productivity. Using technological R&D productivity as an example, In the second quarter of 2026, our technological R&D team's average delivery cycle was shortened by more than 10% compared with the first quarter. Meanwhile, the average daily lines of code submitted on AI-assisted R&D increased by over 70% sequentially. AI's contribution rate to newly added code exceeded 60%. Huaxia Wang Qing not only efficiently supports internal AI use cases, but also provides large-model infrastructure services to a wide range of external enterprise clients, achieving strong revenue growth.

同时,我们将持续以AI技术增助赋能用于组织。 In summary, we will consistently harness AI to deeply empower our business and organization, continue to expand the boundaries of AI applications, and work together with our partners.

to jointly explore and create more innovative business value and greater growth upside. Thank you.

Thank you, Aubrey.

The next question, please. The next question comes from Thomas Chong of Jefferies. Please ask your question, Jingti Wen.

Good evening, thanks management for taking my questions. In light of the uncertainties of macroenvironment and industry competition, What are our strategies to reduce merchants' operational pressure and empower merchants to grow? On the other hand, how should we think about the online shopping outlook in second half? Thank you.

跟您的问题,二季度宏观需求仍然强压,行业竞争仍然激烈。 在这样的环境下,商家追求的是经营确定性,这也是我们商家车工作的出发点。 基于商家的经营类型做分型经营,

Thank you for your question. In Q2, macro demand remained under pressure, while industry competition remained intense. Against this backdrop, merchants are seeking greater visibility and certainty in their business operations. This also shapes our merchant-side initiatives. We segmented merchants based on their business profiles and implemented tiered operations. By offering more tailored operational support, initiatives, and resources, we helped merchants of all types find the right growth opportunities on Quaishou.

First of all, we adjusted the work of the supply and demand from the second quarter to the current type of work. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

First, starting in Q2, we restructured our merchant-facing teams by business type, aligning differentiated strategies with each segment's core needs and bringing policies and resources back to more appropriate levels. Previously, support was misaligned. Brand merchants lacked dedicated professional services. Ad-driven merchants received e-commerce resources disproportionate to their revenue contribution. Influencer and content-driven merchants already rich in organic traffic needed sustainability above all. We have since centralized brand merchants under a specialized team. We allocated mismatched resources away from ad-driven merchants and used full platform fan reach to improve long-term viability for influencer and content-driven merchants. Building on this, we continue tiered At the same time, AI is a key mastermind in helping merchants reduce their profits.

In the context of vegetable production, smart marketing, smart sales, and smart customers, AI ability is deeply applied to the daily business and cost management of merchants.

Meanwhile, we deeply integrated AI capabilities into merchants' daily operations and cost management across scenarios such as marketing material creation, intelligent marketing, intelligent after-sales services, and AI-powered customer service. This helped merchants reinvest into more productive growth initiatives.

. . . . . Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip Yixiao Technology B, Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

This year, brands and industry zones remained our two strategic priorities on the supply side. Since Q4 last year, tightening relevant compliance regulations has narrowed the gap in compliance costs across different merchants, leading merchants to set higher ROI requirements for their marketing investments. Against this backdrop, brand merchants' advantages in operational capabilities and business stability became increasingly evident. Meanwhile, AI has significantly lowered the barriers to content creation. We have further focused to brand subsidies on key blockbuster products. As a result, brand merchants have entered a positive growth cycle on Yixiao. We were pleased to see brand merchants demonstrate a genuine commitment for brand exposure or to drive transaction off platform. In Q2, the contribution of self-operated GMV from brands to overall GMV increased steadily, while the share of marketing spend from brand merchants and total marketing spend also grew steadily. On the industry zone side, we leverage the service providers to deepen our presence across industry zones nationwide and empower local merchants. This year, we have gradually rolled out of this model in regions including Inner Mongolia, Yunnan, Shandong, and Jiangsu. Through these operational initiatives and growth incentives, we provide a tangible support to merchants with the distinctive regional offerings, helping them establish a foothold on Yixiao and scale their businesses.

In the second half of the year, our judgment is that the consumer market is still in the process of mild recovery. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip Yixiao Cheng, Hua Su, Bing Jin, Hoi Lam Yip

Looking ahead to the second half, we believe the overall consumer demand will continue its moderate recovery, while the shift in consumer spending from discretionary to essential categories will persist. This means that merchants will increasingly prioritize greater predictability in their business operations, and the days of relying purely on traffic dividends to drive Business Growth are behind us. While we continue to provide merchants with the traffic resources and commission rate support, we expect e-commerce marketing service revenue and commission income to face pressure in the second half. For the platform, this represents both challenges and opportunities. We'll continue to steer our traffic synergy strategy toward brands and balance to profile merchants with both the ability and commitment to sustain long-term operations. while further strengthening our intelligent placement capabilities to help merchants achieve greater predictability in their marketing placement. We believe that by strengthening our tiered merchant operations and optimizing our supply structure, we will capture structural opportunities to drive high-quality growth in our e-commerce business. This is also essential to rising above short-term cycles and achieving sustainable long-term growth. Thank you.

Thank you. I'll read our next question, please.

The next question is from Daniel Chen from J.P. Morgan. The next question comes from Daniel Chen of J.P. Morgan. Please ask your question.

谢谢管理层给我机会问问题。 我想问一下下半年的非电商类的线上营销服务, 就是广告的发展趋势。

My question is on the non-e-commerce advertising.

How does management assess the trend of this segment service in the second half of this year?

And also, what has been the enablement and the impact of AI on the

Online Marketing Service for this quarter. Thank you.

Online Marketing Service for this quarter. Thank you. Online Marketing Service for this quarter. Thank you.

Thank you for the question. The growth of the online marketing services is affected by the broader macro environment and our clients' marketing budgets. Looking into the second half of this year, we believe visibility on the external environment remains limited. Meanwhile, AI applications and instant retail, which grew rapidly in the second half of last year, will face a relatively high base in the second half of this year, Customer budgets may also be affected by industry competition dynamics and changes in marketing strategies. That said, we continue to see structural opportunities in some industries.

第一是以短距为代表的内容消费。 我们认为这个行业还没有到增长的天花板。 A.I. continues to repeat the short-term industry supply, significantly reducing the production cost and production cycle, bringing more materials, more content types, and a rich supply. In 2025, we expect to reach 500,000 steps. In 2026, we expect to reach 500,000 steps. In the future, industry development will gradually expand from quantity to increase content quality. At the same time, we see that the value of short-term content is more prominent. In July of 2026, the daily number of reports of short-term users for short-term users has reached 230 million. The short-term telecom model is also evolving, and the IAEA ratio is getting higher and higher, which means that short-term users are gradually developing into a sustainable business that has the ability to maintain content consumption value and advertising. In addition, our future content form may be further surpassed by simple viewing, such as IAP extension, interactive devices, First in the content consumption sector,

led by short plays, we believe that this market still has room to grow. AI continued to reshape the supply side of short plays segment, significantly reducing production costs and shortening production cycles while enabling a broader range of genres, content formats, and reach your supply. In 2025, the number of comic-style short plays episodes on Kuaishou was 60,000, and we expected this number to reach 500,000 in 2026. Going forward, we expect the industry to gradually shift from quantity expansion to content quality improvement. Meanwhile, the value of short plays content ecosystem has become increasingly evident, with the daily impressions on Kuaishou short plays reaching 230 million in July 2026. The money transition model for short plays has also evolved. In-app ads account for an increasingly larger share, indicating that short plays are gradually developing into an ecosystem-based business model that combines content consumption value with incremental marketing inventory. Looking ahead, content formats could potentially extend beyond traditional viewings into areas such as IP adaptions and spin-offs, interactive short plays and virtual companionship, and integrating content with cultural tourism and creative industries. These new formats could unlock new opportunities for content consumption and monetization. Against these trends, we expected to capture the growth opportunities in content consumption through revenue-sharing incentives, omni-domain traffic support, and partnership with high-quality content copyright holders. Our platform's high-quality content will achieve more stable traffic and monetization returns.

Secondly, in some segmented industries, we have plenty of access to information in the past, but we still have room for improvement in the future. For example, in the Chinese service industry, Huaxia Zhao, Bing Jin, Huaxia Zhao, Hua Su, Bing Jin, Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

Second, we'll still have room to grow in certain verticals where our penetration remains insufficient. For example, in the lifestyle services sector, The business is closely tied to the broader macro environment, but it also covers a wide range of verticals, each with substantially different merchant operating models, user decision-making journeys, and conversion goals. This will allow us to continue unlocking incremental growth opportunities in verticals such as beauty and wellness, home renovation and decoration, and real estate through more refined industry-specific operations. In the gaming vertical, there's a strong gaming content consumption on Kuaishou, but monetization is still catching up. We're currently exploring opportunities to better connect the gaming live streaming and game content with marketing budgets to unlock more monetization potential. In the AI era, the barriers to developing mini games are also becoming even lower, which could drive more supply and incremental marketing demands.

Regarding the negative effects of AI on online marketing services, we focus more on the trend of managing efficiency and marketing investment. AI is an important means for us to help customers promote sales, reduce costs, and drive customer marketing investment. Specifically, for the first time in marketing materials, AI has clearly reduced the threshold. For example, in the industry of life services and AI tools, AI helps customers and service providers to reduce costs faster and more quickly. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

Regarding AI's empowerment of online marketing services, as our clients place greater emphasis on operating efficiency and marketing placement ROI, AI has become an important lever for us to help clients improve efficiency, reduce costs, and unlock additional marketing budgets. Specifically for marketing materials, AI has significantly lowered the barriers to content creation. For example, in industries such as local services and AI tools, AI helps customers and service providers produce content and marketing materials that are more suitable for Huaxiao users in a faster and more cost-effective way, which has also driven growth in marketing demand from these industries. Secondly, in customer operations, AI is being applied to more workflows such as business opportunity insights, product selection site suggestions, placement diagnostics, and performance review as well as customer service. This helps customers and service providers Standardize and automate processes that previously relied heavily on manual work, thereby improving operating efficiency Meanwhile, on the platform side, we continue to see AI models to improve the matching and conversion efficiency of our advertising system For example, through better understanding of marketing materials and user interests as well as intelligent bidding and intelligent price adjustment capabilities We can help advertisers match their budgets more efficiently with the right users and scenarios.

总之,下半年线上营销服务仍存在结构性增长的机会, AI将继续作为提升营销效率和投放回报的重要能力, 推动客户经营体效和商业化增长。

In summary, structural growth opportunities for online marketing services will continue to emerge in the second half. We'll continue to harness AI as a key capability to optimize marketing efficiency and marketing placement ROI, driving operational efficiency improvements and commercialization growth for our clients. Thank you.

Thank you, operator.

Last question, please. Last question will come from Yuan Liao of Citics. Please ask your question.

Good evening, sir. Thank you for accepting my question. Congratulations on your independent financing. My question is actually about the company's cash flow. After the financing is completed, what kind of impact will it have on the company's cash flow? The company has a cash flow expenditure in the second half of this year. Do you have any plans?

What is the situation of our cash flow in the second half of this year? Good evening, management. Thanks for taking my questions. So after Kerlin finished his financing, what is the impact on Huaxiou's cash flow?

Looking forward to the second half year, what is your plan for cash flow expenditure in the total cash flow situation? Thank you.

Kerlin will take a more flexible approach to cash flow after financing. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip

After the completion of CleanAI's financing, we will have more flexibility in managing cash flow expenditures including addressing additional computing power needs through leasing and other approaches. This will further optimize capital allocation and to some extent improve the group's overall cash flow position. Regarding cash management and capital expenditures, we will continue to adhere to a prudent financial strategy. In terms of execution, the majority of our CapEx was heavily front-loaded in the first half of the year. As a result, the company achieved a positive cash flow in the second quarter, and our objective is to maintain positive group-level free cash flow in the second half of this year.

In terms of shareholder returns, we will continue to create long-term value as the core trend. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip On shareholder returns, we have always focused on creating long-term value and consistently executed a proactive shareholder return strategy.

Since the beginning of the year, the company has completed share repurchases and an aggregated consideration of profit and loss. Yixiao Cheng, Huaxia Zhao, Hua Su, Bing Jin, Hoi Lam Yip These initiatives not only represent a tangible return to our shareholders, but also demonstrate the company's ability to generate sustainable cash flows and its confidence in the long-term development of our business.

At the same time as we embrace the age of AI, our company will continue to hold on to the bottom line of financial security, to reduce the cost-effectiveness of the financial cycle, and to maintain a plentiful health cash reserve. We will use a stable financial structure as a support,

We fully embrace the AI era and at the same time we'll continue to safeguard and strengthen our company's financial foundation. We're committed to driving cost reductions and efficiency improvements through prudent financial displaying while maintaining a robust and healthy cash position. Harnessing our resilient financial structure as the cornerstone We will achieve a healthy and sustainable balance between business expansion and shareholder returns, laying a solid foundation for high-quality long-term growth. Thank you.

Thank you, Operator, at the end of the Q&A session.

Thank you, Operator. Thank you once again for joining us today. If you have any further questions, please contact our capital market and IR team at any time. Thank you.