Good day, ladies and gentlemen. Thank you for standing by. Welcome to Kuaishou Technology's second quarter and interim 2025 financial results 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 Markets and IR at Kuaishou Technology.
Thank you, operator. Good evening and good morning to everyone. Welcome to Kuaishou Technology's second quarter and interim 2025 financial results conference call. Joining us today are Mr. Cheng Yixiao , Co-Founder, Chairman, and CEO. Mr. Jin Bing is 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 or the second quarter and interim 2025 results announcement ended June 30th, 2025, 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 results and measures and a reconciliation of IFRS to non-IFRS financial results and related risk factors, please refer to our second quarter and interim 2025 results announcements. For today's call, management will use Chinese as the main language. A third-party interpreter will provide simultaneous English interpretation in the prepared remarks session and consecutive interpretation during the Q&A session. Please note that English interpretation is for convenience purposes only.
In case of any discrepancy, management's statements in their original language will prevail. Lastly, unless otherwise stated, all currency units mentioned are in RMB. I will now hand the call over to Yixiao .
Hello, everyone. Welcome to Kuaishou's second quarter 2025 earnings conference call. In Q2, we built on our relentless pursuit of delivering compelling user experience and our strong technological foundation, using large AI models to empower our content and business ecosystems. These efforts created value for both our users and business partners while setting record highs across multiple operational and financial metrics. Notably, average DAUs on the Kuaishou App reached an all-time high of 409 million in Q2, showing steady year-over-year growth. In Q2, our total revenue rose by 13.1% year-over-year to CNY 35 billion. Adjusted net profit totaled CNY 5.6 billion, with a 16% adjusted margin, marking a new high in quarterly profitability.
We achieved this remarkable growth while maintaining our commitment to strategic investment in AI, even amid macro uncertainties, demonstrating our strategic result. It also highlights the resilience of our content and business ecosystems, as well as our efficient organizational execution. As you have seen in our results announcement, given the company's business performance, the board has declared a special dividend of HKD 0.46 per share for the first time since the listing, amounting to approximately HKD 2 billion in total. This reflects our confidence in the company's long-term growth prospects and solid financial position, as well as our commitment to continuously enhancing shareholder returns. We are sincerely grateful for our investors' continued support. Kuaishou's steady growth would not have been possible without the trust and support of our shareholders. Looking ahead, we will consider measures such as share repurchases and dividend payments as appropriate to reward our shareholders.
Next, I'd like to elaborate on the progress of our key business segments in Q2. First, our AI strategy and the progress of our large video generation model, Kling AI. For Kling AI, we remained dedicated to delivering state-of-the-art large visual generation model technology and product performance. We launched the Kling AI 2.1 model series in May. The latest iteration boasts comprehensive quality improvements, including better motion performance, more realistic physical simulation, and more acute somatic responsiveness. It offers two modes: standard and high quality, providing creators with differentiated video generation solutions. At the end of July, Kling AI also released a Kling App, which integrates unlimited visualization space, intelligent creation assistance, and real-time multi-user collaboration. The all-new feature brings creators a seamless and efficient one-stop creation experience. Kling AI is committed to becoming a one-stop creative engine for creators, empowering everyone to craft captivating stories with AI.
Kling AI has assisted prosumers and corporate clients in exploring more application scenarios, including advertising and marketing, film, television and short plays, gaming interactions, and smart hardware, among other areas. In the short play industry, Kling AI and Kuaishou Astro Short Plays jointly produced the world's first AI-generated astrology series. A new world is looming. So far, it has racked up almost 200 million cumulative views globally, setting a new benchmark for AI-generated film and television content. These in-depth explorations of Kling AI's application scenarios also led to commercialization growth. In Q2, revenue from Kling AI surpassed CNY 250 million. Large AI models also empower our content and business ecosystem. We launched OneRec, an end-to-end generative recommendation large model that represents a new technological paradigm.
By introducing a reward-based preference alignment method and leveraging reinforcement learning, our large model can generatively predict videos and match users' preferences from the video library. With this architectural advantage, the new large recommendation model significantly improves the efficiency of training and inference computing power. The technology is being used for recommending short videos and has improved user time span and retention rates. For online marketing services, we explored the correlation between users' omni-domain behaviors and marketing service conversion by leveraging raw knowledge and inference capabilities of large language models. The process included supervised fine-tuning of omni-domain data and optimization of models based on user preferences. These efforts enabled end-to-end prediction of tailored marketing materials, increasing click-through and conversion rates, driving low single-digit growth in online marketing services revenue.
In the e-commerce scenarios, we applied multi-modal large language models to enhance product information extraction by leveraging LLM knowledge reasoning, which generated a structured description of target users, use cases, and user interests, amplifying user product matching efficiency. Second, user growth and content ecosystem. In Q2, average DAUs on the Kuaishou App reached 409 million, and MAUs reached 715 million, increasing by 3.4% and 3.3% year-over-year, respectively. Average DAUs on the Kuaishou App set another new record. The average daily time spent per DAU on the Kuaishou App was 126.8 minutes, while total user time span rose by 7.5% year-over-year in Q2. The steady growth in our user traffic was attributable to our high-quality user growth strategy, an optimized traffic allocation mechanism, a more engaging community atmosphere, and our differentiated content operating strategy.
By staying focused on our high user value growth strategy, we improved ROI by reducing user acquisition costs. We also refined our traffic allocation to better align commercial content with user retention, which strengthened both user experience and stickiness. At the same time, we increased traffic exposure for top-tier original content, fostering a virtuous cycle between content creation and consumption. We upgraded the private messaging experience across various scenarios and expanded innovative social features that have made communications more engaging. As a result, the daily percent penetration rate of private messages among users with mutual followers increased by nearly five percentage points year-over-year. Our differentiated vertical operation strategy also further strengthened our connection with users as we continue to deliver standout content with distinctive Kuaishou characteristics.
In the three rural verticals, we launched the Happy Village Spring Farming Season Series activities, featuring short videos on agriculture technology and knowledge, e-commerce live streaming to promote agricultural materials, tools, and products, as well as the Spring Farming Grand Stage pet entertainment event. These activities boosted interactions between three rural creators and interested users with related short video content, accumulating 12.7 billion total views. As the National College entrance examination approached, we brought together teachers from schools, education experts, and high-quality creators in the pet knowledge vertical. Through live streaming and short videos, they provided students and their families with practical information before, during, and after the exam. Third, online marketing services. In Q2, revenue from online marketing services reached CNY 19.8 billion, up 12.8% year-over-year, with the year-over-year growth rates accelerating from the first quarter. We secured additional budgets from marketing clients across multiple industries.
With proactive outreach and by optimizing our intelligent marketing product solutions with large AI models, these improvements lowered the threshold for the clients to place marketing materials and also enhanced marketing conversion efficiencies, leading to notable year-over-year increases from both external and closed-loop marketing services. In Q2, we also upgraded our AIGC marketing material solutions. The newest iteration moves beyond general capabilities to better align consumer-specific industry attributes and application scenarios, which further improved conversion rates for AIGC materials. In Q2, revenue from external marketing services continued to grow, driven by strong demand from the content consumption, local services, and automotive industries. For the content consumption industry, advertising spending in the short place remained high, double-digit growth year-over-year. Smart pricing for the in-app purchases and dynamic adjustments of in-apps as promotional events drove this momentum.
They improved users' willingness to pay and increased their time spent, which in turn expanded revenue for our marketing clients. For clients in the local services and automotive industries, who mainly relied on lead-based operations, our native private messaging products proved especially effective in reaching users and driving conversion efficiency. In Q2, we leveraged large language models and model sequence modeling for native private messaging between clients and users, which raised the lead conversion rate in terms of intelligent product placement solutions. The penetration rate of our UAX plasma solutions continued to rise, making up around 65% of total external marketing spending in Q2. In Q2, for closed-loop marketing services, we helped e-commerce merchants elevate the efficiency of our omni-domain intelligent operations by refining our product and traffic strategies.
We continued to upgrade our omni-platform marketing solution, iterating key tools such as intelligent bidding, supplementary material placement, and omni-domain multi-product marketing to simplify the placement process and enhance the ad placement stability for merchants. In Q2, our omni-platform marketing solutions accounted for an increasing share of total closed-loop marketing spending. Scenario-wise, revenue from online marketing services related to shelf-based e-commerce has achieved rapid growth. In addition to continuously introducing high-quality content and merchandise, in Q2, we zeroed in on optimizing our traffic distribution strategy for shelf-based e-commerce, strengthening the synergies between e-commerce business models, especially for shelf-based e-commerce. This refinement led to simultaneous improvement in the sales of a conversion rate of e-commerce merchants and effective cost per mile of our closed-loop marketing services. Fourth, our e-commerce business in Q2, e-commerce GMV, rose by 17.6% year-over-year to CNY 358.9 billion.
We continue to deepen the efficient integration and connection of exceptional content and superior products, further meeting more users' consumption needs and elevating their user experience. As a result, the number of e-commerce monthly average paying users continued to grow year-over-year to 134 million in Q2, while the repeated purchase frequency of active e-commerce merchant users also increased year-over-year. In Q2, our shelf-based e-commerce efforts also further helped merchants to capture new business opportunities for omni-domain solutions. Shelf-based e-commerce GMV continued to outpace overall GMV growth in Q2, accounting for over 32% of the total e-commerce GMV. During the 16/18 shopping festival, GMV from the shelf-based e-commerce products cards rose by over 50%, and search-induced e-commerce soared by over 140% year-over-year. For new merchants in the shelf-based e-commerce, we further deepened marketing host applications.
We utilized materials and merchandise information from content-based scenarios, helping new merchants achieve basic operations for one-stop marketing. For existing merchants operating on a larger scale, we offered a big brand, big strategy, a marketing event for top-tier high event interest products, and super links, the official channel of products, recommended products, leading to a 30% year-over-year increase in average daily active merchants in the shopping mall in Q2. As shelf-based e-commerce expanded rapidly, short video e-commerce GMV grew by over 30% year-over-year, largely due to increasing supply in short videos with embedded shopping links. Growth in short video e-commerce GMV also benefited from enhanced supply quality, driven by short videos, blockbuster products from live streaming sessions, which boosted the matching efficiency between targeted demographics and core merchandise. In Q2, new merchant expansion and optimized scenario development drove the number of active small and medium-sized merchants to rise further year-over-year.
We are deeply committed to increasing incentives and resources for new merchants through the Golden Bounty Initiative, service provider empowerment, and traffic support across scenarios. We have helped them quickly adapt to the platform and grow their business. These efforts drove a 50% year-over-year increase in the number of newly onboarded merchants in Q2. We established a product sourcing center to screen high-quality merchandise supplies for merchants and reduce operational thresholds. We also developed a task and incentive system aligned with the growth stages of merchants across tiers, covering four core scenarios: live streaming, short video, shelf-based e-commerce, KOL distribution, further optimizing the merchant ecosystem. In Q2, we placed a strong emphasis on achieving long-term synergies with KOLs. We helped them enhance their content capabilities by introducing new features that support full-on growth and re-engagement, boosting the operating efficiency in the private domain.
We also established stable distribution channels for merchants by building a platform indoors product. The KOL Blockbuster Initiative efficiently connected KOLs with exceptional products, while the Treasure Brand Spotlight Initiative strengthened the diversity of their product distribution by introducing more branded merchandise. In Q2, monthly available merchandise in KOL distribution increased by over 30% year-over-year. We encouraged entertaining KOLs to transition to small and medium-sized KOL, bringing a large number of new and small KOLs into leading KOL teams and organizations, and provided them with a tailored incubation path and commercialization models. These steps facilitated rapid growth of new KOLs. Over the mid to long term, these small and medium-sized KOLs will become a key growth engine for our ecosystem. In Q2, incorporated AI capabilities in merchants' end-to-end operations, particularly in AIGC content production, smart live streaming, and intelligent customer service.
Integration reduced merchants' operating costs on the content-based e-commerce platform. Our shelf-based AIGC capabilities are fully embedded across our e-commerce content ecosystem. This directly improved product conversion efficiency in various scenarios, including product cards and short videos, through the automated generation and optimization of high-quality materials. Going forward, we aim to use larger models to reconstruct connection between customers and merchandise, making a substantial upgrade from meeting immediate demand to predicting potential demand. Next, our live streaming business. In Q2, live streaming revenue grew by 8% year-over-year to CNY 10 billion. As we continue to develop a rich ecosystem and top-tier live streaming content to meet users' diverse needs, we ramped up our efforts to boost revenue efficiency in our core categories, including group and multi-host live streaming. More specifically, we refine operations and leverage regional attributes with a greater emphasis on combining short videos with streaming.
We also explored AIGC innovation and application in live streaming, voice chat, and other real-time interactive scenarios. By the end of Q2, our partner talent agency size had risen by more than 20%. Talent agency management streamers increased by over 30% year-over-year. On top of that, we announced online and offline live streaming scenarios, providing a broader stage for exceptional live streaming content from Unwon, Operation Opera Group, and Tangshan Bubble Dragon. In Q2, Grand Stage expanded its reach nationwide, establishing a presence in 57 cities across Hebei, Guangdong, Huizhou, and Jiangsu provinces, among others. Our live streaming business can deepen its cooperation with game developers by hosting e-commerce events such as Crossfire Mobile Champs Cup events. By combining content ecosystems, our advanced sources, and user consumption, we provided game developers with end-to-end support from exposure to conversion based on their full lifecycle stages and specific needs.
Our live streaming plus strategy continues to empower traditional industry. In Q2, average daily number of users submitting resumes on [Quai Har] increased by over 40% year-over-year, and the number of matches grew by over 150% year-over-year. In ideal housing, daily lead generation surged by over 60% compared with the same period last year. Finally, in terms of our overseas business and local services process, in Q2, we continue to develop overseas business at a steady pace. Revenue grew 20% year-over-year to CNY 1.3 billion. We continue to expand our new users through innovative channels and improve the customer acquisition efficiency with refined operations, strengthening the core user base. As a result, the DAUs in Brazil, one of our core international markets, remain stable with consistent growth in average daily time span per DAU year-over-year.
On the monetization, we actively expanded the industry coverage of advertisers to continually expand product capabilities, maintaining healthy year-over-year growth in online marketing services revenue. For our e-commerce business in Brazil, our end-to-end digital empowerment of local merchant operations enabled content-driven immersive consumption, real-time interaction, and rapid conversion-driven consistent, solid year-over-year and quarter-over-quarter growth in transaction scale and order volume in Q2. In Q2, GMV for local services increased steadily year-over-year. Category-wise, we strengthened our cost controls in our in-store dining services and strategically avoided subsidy competition while supporting high ROI product growth in our general in-store business. On the supply side, we continue to improve our core categories and price comparison capabilities, promoting the healthy growth of competitive products and low-priced supplies. In Q2, the amount of daily average available merchandise leaped by over 55% year-over-year.
On the demand side, our city-specific refined operations and elevated user experience, coupled with improved smart subsidy model, drove the overall healthy growth of the business. In terms of content, we remain committed to supporting the distribution of high-quality traffic on our platform. Furthermore, recommendation funnels aided by large models have also procured increases in video views, efficiency enhancements, local services, and short video scenarios. For monetization, we meet users' diverse placement needs by expanding and optimizing advertising products, resulting in almost 120% year-over-year revenue increases. In addition, stricter ROI displaying improved both subsidy and operating efficiency, further narrowing the operating loss in our local services year-over-year and quarter-over-quarter. Looking back on the first half of 2025, our advanced AI strategy at a steady pace, accelerating the commercialization process while achieving continuous technological breakthroughs.
Looking ahead to the second half, we will make steady, fast investments in AI with a continued commitment to exploring AI technology's value in creating efficiency for creators, marketing clients, and e-commerce merchants, as well as discovering new commercialization opportunities with greater potential. As we maintain high-quality growth in our existing business, we will continue to promote the synergies between our content and business ecosystem, allowing both to flourish. As always, we will base our operations on our user needs and strive to amplify operational efficiencies for merchants and marketing clients, working with the users, creators, and partners to jointly create long-term value through long-term sustainable growth. That concludes my prepared remarks. Now, our CFO will review the company's financial update for the second quarter of 2025.
Thank you, Yixiao, and hello everyone.
In Q2, despite the challenging external environment, we'll continue to make breakthroughs in both our financial performance and operational metrics by leveraging our industry-leading AI technology, vibrant user ecosystem, increasingly prosperous content, and solid diversified business ecosystems. AI technology is now deeply integrated across the majority of our business scenarios, in turn bringing new commercial opportunities and showcasing our robust growth momentum. This success in return has reinforced our commitment to long-term steady, fast investment in technology. In Q2, our total revenue increased by 13.1% year-over-year to CNY 35 billion, and adjusted net profit grew by 20.1% year-over-year to CNY 5.6 billion, with a margin of 16%, ensuring our ability to increase profitability. These achievements highlight both the resilience of our business ecosystem and also demonstrate our healthy and sustainable growth potential. Now, let's take a closer look. Our total revenue grew 13.1% year-over-year to CNY 35 billion in Q2.
The increase was mainly driven by growth in each of our core businesses, including online marketing services, live streaming, e-commerce, and Kling AI. Our marketing services revenue increased by 12.8% to CNY 19.8 billion in Q2, from CNY 17.5 billion in the same period last year. As we continue to leverage our AI capabilities to optimize our intelligent marketing product solutions, we improved the conversion efficiency of marketing materials, which in turn drove higher spending from our online marketing clients. Revenue from other services, including e-commerce and Kling AI business, reached CNY 5.2 billion in Q2, up 25.9% from CNY 4.2 billion in the same period last year. This increase was mainly attributable to growth in e-commerce GMV, which boosted e-commerce commission income, as well as Kling AI's continually growing revenue scale. Rapid iterations of Kling AI and expansion into more applications now continue to accelerate its monetization.
In Q2, our live streaming revenue was CNY 10 billion, an increase of 8% from CNY 9.3 billion in the same period last year, which further developed specialized verticals to better diversify the live streaming ecosystem. At the same time, we consistently refined the operations of our core categories and deepened the integration of short videos and live streaming, which led to greater user engagement with high-quality live streaming content. Cost of revenues increased by 12.3% year-over-year in Q2 to CNY 15.5 billion and accounted for 44.3% of total revenue. The increase was mainly due to increased revenue sharing costs and related taxes in line with our revenue growth, partially offset by decreases in the depreciation of property and equipment and ready-to-use assets and amortization of intangible assets. In Q2, gross profit grew by 13.8% year-over-year to CNY 19.5 billion.
Gross profit margin was 55.7%, an increase of 0.4 percentage points year-over-year, an increase of 1.1 percentage points sequentially. Moving to expenses, selling and marketing expenses increased by 4.6% year-over-year to CNY 10.5 billion, accounting for 30% of total revenue, a drop from 32.4% in Q2 last year. The increase in selling and marketing expenses was mainly due to increased spending on business promotions, including online marketing services and e-commerce business. While the decline as a percentage of total revenue was attributable to our improved operational efficiency, research and development expenses were CNY 3.4 billion, rising by 21.2% year-over-year and accounting for 9.7% of total revenue. The increase in R&D expenses was mainly due to increased investments in AI. Administrative expenses increased by 13.3% year-over-year to CNY 897 million, accounting for 2.6% of total revenue.
The increase in administrative expenses was mainly due to higher employee benefit expenses, including related share-based compensation expenses. Group-level net profit for Q2 was CNY 4.9 billion. Group-level adjusted net profit rose 20.1% year-over-year to CNY 5.6 billion, with an adjusted net margin of 16%. As the company's profitability and cash management capabilities further improved, our cash and reserves continued to increase, with cash and cash equivalents, time deposits, restricted cash, and wealth management products totaling CNY 101.9 billion. As of June 30, 2025, we generated a positive operating net cash flow of CNY 8.5 billion in Q2. Additionally, we actively delivered on our commitment to shareholder returns based on the market conditions. As of June 30, we had repurchased an aggregate of approximately HKD 1.9 billion, or around 38.8 million shares, which accounted for about 0.9% of our total shares outstanding within 2025.
I'd like to explain the changes we are planning to make in our GMV data disclosure. As we further diversify our e-commerce business model and revenue mix, quarterly GMV is no longer a core indicator of our e-commerce business performance. We began to release quarterly GMV figures after the company went public because we wanted to make it easier for the market to follow Kuaishou's growth trajectory as an up-and-coming e-commerce service platform. Now that this year's expected GMV exceeds CNY 1.5 trillion, our e-commerce business will be driven by a more nuanced combination of factors going forward. These include more diverse e-commerce scenarios, AI-empowered merchant solutions, and improved comprehensive experiences for e-commerce users, which cannot be fully captured by the quarterly GMV metrics alone. As such, beginning in the first quarter of 2026, we will no longer disclose quarterly and yearly GMV separately.
Hence, everyone can focus on a more comprehensive set of indicators that can better reflect the long-term value created by the company. This adjustment is aligned with our industry peers' practice and will help investors gain a more accurate understanding of our operational priorities and long-term strategies. We will still disclose GMV data in our third and fourth quarter reports this year and annual report of 2025, as we have been doing to ensure the market has sufficient time to process and adapt to this adjustment. In summary, looking ahead, we will continue to focus on our user needs while remaining committed to investing in AI. We will further build our content and business ecosystem mode, maintaining solid financial performance improvements that drove the company's sustainable profitability and create long-term value for users, partners, and shareholders. That concludes our prepared remarks. Operator, now let's open the call for questions.
[Foreign language]
We will now begin the question and answer session. To ask a question, please press star one on your telephone keypad. To withdraw your question, please press star two. Please ask your question in Chinese first and repeat your question again in English.
[Foreign language]
Your first question comes from Lincoln Kong of Goldman Sachs.
[Foreign language]
[Foreign language]
Thank you, management, for taking my question and congrats on the very solid second quarter. My question is about the Kling AI. We have seen Kling AI second quarter revenue accelerating. Could management elaborate what the major use cases for Kling AI users are at the moment? In the future, how should we think about the potential breakthrough and the upgrade areas? What's the growth strategy for Kling AI for the future? Thank you.
[Foreign language]
Thanks for your question. Currently, Kling AI's users include both mass creators with an interest in AIGC and professional creators, such as shelf-based e-commerce, self-media users, designers, and artists, e-commerce/advertising industry professionals, and film television production studios. At the moment, Kling AI's major paying users are professional creators. In terms of application scenarios, AIGC-driven mass creators primarily use Kling AI to produce engaging, creative images or videos and share them on social networking or content platforms. Pan self-media users adopt Kling AI to quickly create short video content, such as popular science content visualization. They efficiently gain traffic and achieve monetization by taking advertising orders. Designers and artists can leverage Kling AI to inspire creative ideas and explore artistic styles, helping them to create all new artwork.
For e-commerce/advertising industry professionals, they use Kling AI to produce high-quality product displays and create advertising materials at significantly lower cost, elevating operating efficiency. As for film and television production studios, they use Kling AI for pre-production concept design, special effects, and establishing shots.
[Foreign language]
Looking ahead, as our model's overall performance improves, we hope to expand Kling AI's use in game production, professional film and television production, and other professional large-scale creative industry scenarios. This will expand our large-scale production user base. In the gaming industry, Kling AI's current cooperation with the Sword of Justice mobile game is largely focused on image-generated views of game characters and interactive special effects that have enriched the game's social features. We believe that Kling AI can play a bigger role in game concept design, scenario generation, character design, and other stages of game development. For the professional film and television production industry, Kling AI participated in the entire production process of the world's first AI-generated anthology series, New World is Loading, from script formulation to completion, demonstrating Kling AI's potential in large-scale professional film and television content production.
In particular, Kling AI can overcome the limits of traditional film and television production, which depended on physical set construction by generating scenes and images that are difficult to create in reality. This capability has brought more creative possibilities to film and television production, enhancing the visual impact and artistic value of films.
[Foreign language]
Regarding our strategy going forward, our core direction is rooted in our vision of empowering everyone to craft captivating stories with the help of AI. We are concentrating our resources on iterating Kling AI's foundational models and making the generated results more controllable. On the product level, we will serve the needs of our core user demographics by continuously elevating the user product interaction experience. Recently, we launched the Kling Lab, integrating text to image, image to video, and other features to form a streamlined operational process. It also empowers creators with real-time collaboration capabilities, offering a seamless and efficient one-stop creation experience. For industrial users in game production and professional film and TV production, among other areas, we will drive technological innovation through industry-specific solutions, create representative cases of applications, and extend the upper limit of our model and product capabilities.
As for mass creators, we will engage users with more creative features, improving product usability while broadening our user case by unveiling innovative, in-demand features. In conclusion, we are fully confident in the application scenarios and monetization prospects of our larger video generation models. We also believe Kling AI will maintain its market share leadership.
Thank you, operator. Next question, please.
Your next question comes from Felix Liu of UBS.
[Foreign language]
Thank you, management, for taking my question and congratulations on the strong second quarter results. My question is also on AI. In addition to Kling AI, could management share the AI use cases in your overall business, including content and creator ecosystem, online marketing, and e-commerce? What are the major progresses so far? Thank you.
[Foreign language]
Thanks for your question. We have consistently focused on deeply integrating AI technology throughout our existing businesses. In Q2, we rolled out the OneRec end-to-end generated recommendation large model and used it for recommending short video content, improving both user time spend and retention. OneRec has also been fully applied to recommendations in local services scenarios, propelling notable growth in local services GMV.
For online marketing service scenarios, in Q2, our large AI models made significant progress in AIGC marketing material generation, marketing placement bidding, and marketing recommendations. In terms of AI-driven marketing material generation, in Q2, we began to explore AIGC marketing material product solutions that better align with the customer's industry attributes and application scenarios. Take digital human live streaming in the apparel industry, for example.
We launched a streamer model with dual-person live streaming features, supporting real-time outfit changes and displays by models according to streamers' commentary. Once the feature was released, a menswear client changed all of their live streaming sessions from real-person to digital human live streaming and increased their marketing spend by almost two-fold. For marketing placement bidding, we continued to optimize generative bidding, relying on the quality of our datasets, leveraging reinforcement learning. By modeling long-term value-driven strategies, we explored rich sample data to heighten marketing conversion rates. In marketing recommendations, we harnessed the content understanding and inference capabilities of large language models for generative retrieval, uncovering potential relations between users' omni-domain behaviors and advertising conversions. This enabled end-to-end generation of ads that users are interested in, and with the generated candidate, results sent through the subsequent pre-ranking and ranking stages, increasing marketing material click-through rates.
This improvement drove low single-digit growth in online marketing services revenue. In e-commerce business scenarios, we continued to invest in building e-commerce AI capabilities centered on multimodal understanding large models, image and video generation large models, and successfully applied AI technology to search, recommendation, content generation, and other application stages, achieving good business results. In AIGC content generation, our self-developed advanced AIGC capabilities are fully empowering the e-commerce content ecosystem. Through automated high-quality material generation and optimization, we directly improve product conversion efficiency in product cards, short videos, and other fields. By optimizing product images with AI intelligent expansion and other technologies, the conversion efficiency of related product cards increased by more than 10%.
In summary, we have already seen the value of AI empowering Kuaishou's content ecosystem and business ecosystem gradually being released, especially bringing marginal improvements to the growth of online marketing and e-commerce businesses. In the future, we also
[Foreign language]
For the e-commerce business scenarios, we have consistently invested in building AI capabilities centered on multi-model understanding large models and image to video generation models. We've also seen remarkable results from applying AI technology to search-induced recommendations, content generation, and other business steps. In terms of AIGC content generation, our self-developed advanced AIGC tool is comprehensively empowering the e-commerce content ecosystem. The automated generation and optimization of high-quality materials have fully amplified product conversion efficiency in product cards, short videos, and other scenarios. Optimization of product images through AI-powered intelligent cropping and expansion technologies increase the transaction conversion efficiency of related product cards by over 10%. To conclude, it's clear we are gradually unlocking AI's value in empowering our content and business ecosystems, particularly AI led to incremental improvement in the growth of our online marketing and e-commerce business.
Going forward, we believe AI will create even greater value for our ecosystem.
Thank you, operator. Next question, please.
[Foreign language]
Your next question comes from Brian Gong of Citi.
[Foreign language]
[Foreign language]
I'll answer myself. A very quick question on advertising. Our online marketing services revenue growth accelerated in the second quarter versus the first quarter. What verticals do we expect to have strong growth in the second half of this year, and how can we seize opportunities from those verticals? Thank you.
[Foreign language]
Thanks for your question. In Q2, thanks to our proactive customer outreach and our product capabilities and empowerment by larger AI model technology, our revenues from external and closed-loop marketing services both achieved a remarkable year-over-year growth. In the first half of this year, we saw year-over-year increases in average load rate and eCPM, two drivers of our growth. We believe these two metrics have room for further improvements in the second half of the year. In terms of industry opportunities, we believe regional lead-based industries such as local services and automotive, along with content consumption industries like short play and mini games, as well as native e-commerce, will continue to grow more rapidly.
[Foreign language]
In regional lead-based industries, in addition to iterating and optimizing targeted messaging and lead forms, as well as other products, we will expand our supply through various methods to increase our marketing client base. For example, in the local services industry, we will deepen operations that have mature SKUs and incubate new ones in key sub-industries, including medical care, home furnishing, and construction materials. At the same time, we'll explore growth opportunities in the high potential regions of our local services business. In the automotive industry, we'll help auto OEMs set up a matrix of sub-accounts under their official accounts and guide them to improve operational results through short videos and live streaming. We'll also partner with more automotive dealers in crucial regions through channel agents.
In the content consumption industry, we'll continue to encourage clients to align their campaigns with native in-platform content operations, increasing content value and user stickiness. In addition, we'll dynamically adjust user payment and nerves for in-app purchases and in-app ads promotional events using large model technologies to increase users' willingness to pay and time spent for content consumption, thereby boosting our clients' revenue. Over the past year, there has been replicated growth in marketing spend on short plays. Next, we are confident we can replicate the native sales funnel and product capabilities of short plays in question mini games and novels, leading to improved total benefits for clients.
[Foreign language]
For native e-commerce marketing services, we'll focus on reinforcing the synergies between our marketing services and e-commerce business. This will allow us to align revenue from closed-loop marketing services with e-commerce GMV, supported by e-commerce merchant expansion, intelligent marketing product upgrades, traffic distribution strategy integration, and stronger monetization of shelf-based e-commerce. Specifically, we will design a support system for key targeted merchants to ensure an early-stage performance in customer outreach through large-scale marketing events, thereby improving merchant retention. Meanwhile, we will strengthen merchant infrastructure capabilities through our omni-platform marketing, AIGC, and other intelligent marketing products, lowering the threshold for marketing placement materials. From a traffic distribution perspective, we will zero in on the long-term value of follow-up growth to strengthen merchants' commitment to sustained operations. In shelf-based e-commerce, our GMV has maintained fast growth for some time, but we have yet to monetize it on a large scale.
In the first half of this year, we achieved rapid year-over-year growth in closed-loop marketing services for shelf-based e-commerce. As a result of introducing more high-quality supply, optimized placement channels, funnels, and better algorithms, nevertheless, the advertising take rates for shelf-based e-commerce remain far lower than that of content-based e-commerce, leaving ample room for growth.
[Foreign language]
In conclusion, looking ahead to the second half of the year, we'll continue to concentrate on industry-specific customer outreach, iterating our intelligent marketing products, and optimizing our traffic strategy and algorithms to improve marketing placement conversion for clients and drive rapid growth in our online marketing services revenue ahead of the industry's growth rate.
Thank you, operator. Next question, please.
Your next question comes from Thomas Chong of Jefferies. [Foreign language]
[Foreign language]
Good evening, thanks management for taking my question and congratulations on a very strong set of results. We have seen our GMV growth in Q2 exceed expectations. In light of the intense competition in online shopping, how does Kuaishou maintain the momentum and outpace industry growth? In particular, can management talk about the progress for shelf-based e-commerce as well as the strengthening in monetization? Thank you.
[Foreign language]
Thanks for your question. The core driver behind our above-expectation e-commerce GMV growth in Q2 was the synergized ecosystem that we have built on both the supply and the demand sides, along with our efforts to accelerate shelf-based e-commerce commercialization.
[Foreign language]
On the supply side, we tailored enabling initiatives to different types of merchants, unleashing the vitality of omni-domain supply. For content-based merchants on Kuaishou, we helped them systematically utilize their live streaming and short video materials in shelf-based e-commerce scenarios. Aided by semi-hosting operation solutions, they achieved integrated omni-domain marketing with exceptional efficiency. Currently, semi-hosting operation solutions have been used by over 20% of merchants. For shelf-based merchants, we provided them with exclusive traffic support through the Blockbuster initiative and the new product growth initiative, facilitating their expansion fueled by marketing activities. As for supply-based merchants, we empowered them through our innovative Superlinks model, which integrates merchandise and mobilizes KOLs for omni-domain distribution.
These differentiated platform empowerment and capabilities drove a 30% year-over-year increase in the number of average daily active merchants in the shopping mall in Q2, along with a broader selection of high-potential categories such as fresh food, consumer electronics, and home furnishings.
[Foreign language]
On the demand side, we focus on growing our user base and increasing stickiness, tapping into users' lifetime value. We acquired customers with precision from shelf-based discounts, with monthly active paying users in our e-commerce business reaching 134 million in Q2. High-quality users continue to make up a larger share of the space. Meanwhile, we build a closed loop from product recommendations, conversion, and repeat purchases, elevating conversion rates through wishlists, mega crowd deals, and other scenario-based tools. These efforts led to a steady increase in repeat purchase frequency from e-commerce users.
[Foreign language]
At the same time, we made solid progress in commercialization of shelf-based e-commerce. In terms of commissions, we recommended products through Superlinks and helped successful bidders conduct omni-domain sales, adding incremental commissions to our e-commerce business. Furthermore, we helped more merchants achieve one-stop operation by providing them with a semi-hosting operation service solution. In particular, our optimized AI-powered materials have improved the efficiency of short video and live streaming content production, delivering completed AI-generated clips for merchants and boosting additional e-commerce commissions. For closed-loop marketing services, we improved marketing placement funnels to incorporate e-commerce merchants' top-tier content and product offerings into shelf-based scenarios. This, coupled with heightened matching efficiency from enhanced algorithm strategies, drove substantial increases in the advertising monetization rate for shelf-based e-commerce.
[Foreign language]
Looking ahead, as users' consumption needs become increasingly complex and layered, requiring more nuanced decision-making process, merchants must establish stronger connections with consumers. Kuaishou will continue to empower merchants across short videos, merchants' self-operated live streaming, KOL distribution, and shelf-based e-commerce, and tap into more predictable incremental opportunities through AI to unlock more business possibilities.
Thank you, operator. The last question, please.
[Foreign language]
Your last question comes from Yuan Liao of CITIC.
[Foreign language]
[Foreign language] Thanks management for taking my questions.
Congrats!
The strong results in the second quarter, my question is about the AI CapEx and net profit margin. As Kling AI develops further, are there any new updates for CapEx and AI expenses? What is your full-year profitability target for 2025, and will AI's impact on the full-year margin remain 1%- 2%? Thank you.
[Foreign language]
Thanks for your question. As Yixiao said, we are delighted to see the remarkable effects of AI on boosting the quality and efficiency of our operations. AI's value to Hua Su's content and business ecosystem is also becoming more visible. We can see this in the rapid ramp of Kling AI's revenue. We currently expect the Kling AI's full-year revenue to be double that of the target we set at the beginning of the year. This has reinforced our confidence in the value of our long-term investments in Kling AI. We also made additional investments mid-year in Kling AI's reference computing power. As such, we expect the Kling AI-related CapEx to double from the beginning of the year budget. On the expenses front, our budget from the beginning of the year already included expected investments for attracting and retaining AI talent.
We don't expect significant changes for those expenses at the moment.
[Foreign language]
Since Kling AI's growth margin has turned positive at the inference computing power level and has remained stable throughout product iterations and upgrades, the impact of additional investments in inference computing power on the group's profit will remain manageable. As such, we expect the impact of the group's overall AI investments on full-year profitability to hover around 1%- 2%. This means that despite our increased investments in AI, we remain confident in maintaining our full-year adjusted net margin at a stable year-over-year level in 2025. In the mid to long term, we believe there's further room to reduce Kling AI's unit training costs and inference costs. As we further unlock AI's value across our content and business ecosystem, we believe that the increment of AI investments will bring sustainable business growth opportunities. Thank you.
Thank you, Operator. That concludes the Q&A session.
[Foreign language]
[Foreign language]
Thank you once again for joining us today. If you have any further questions, please contact our Capital Markets and IR team at any time. Thank you.