Kuaishou (Trans): Core Business Under Significant Pressure in H2, Full-Year Positive Cash Flow Expected

Compiled by Dolphin Research: $ KUAISHOU-W.HK FY26Q2 Earnings Call Trans.

I. Key Financial Highlights

1. Shareholder returns: Approx. 43.3mn shares repurchased YTD, about 1% of outstanding shares, with total consideration of HK$2.0bn. In addition, cash dividends of HK$3.0bn were paid.Combined returns have nearly reached last year's full-year level, and the company expects FY26 shareholder returns to exceed last year.

2. 2H outlook: Management expects revenue pressure alongside ongoing AI investment to weigh on margins in 2H. On e-commerce, continued traffic and commission support for merchants implies pressure on e-commerce ad services and commission income.Outside e-commerce, last year's high base in AI applications and on-demand retail plus potential shifts in client budgets amid competitive and strategy changes may also add pressure.

3. Cash flow and capex cadence: Capex was front-loaded into 1H, and the group delivered positive cash flow in Q2, targeting positive FCF at the group level in 2H. Following Kling AI's independent financing, incremental compute can be addressed via leasing and other flexible arrangements, improving cash outlay elasticity.As of Jun 30, 2026, cash, cash equivalents, time deposits and near-cash financial assets totaled RMB 121.3bn; Q2 operating cash inflow was RMB 5.9bn.

4. Key quarterly metrics

a. Aggregate: Revenue of RMB 35.5bn (+1.4% YoY); core commercialization revenue (online marketing services plus other services) rose +7.4% YoY. COGS was RMB 17.2bn (+10.7% YoY), 48.4% of revenue, mainly on higher revenue-sharing costs and related taxes as revenue grew.GP was RMB 18.3bn (vs. RMB 19.5bn a year ago), with GPM at 51.6% vs. 55.7% last year.

b. Segments: Online marketing services revenue RMB 20.6bn (+4.4% YoY; vs. RMB 19.8bn last year). Other services (primarily e-commerce and Kling AI) RMB 6.2bn (+18.5% YoY; vs. RMB 5.2bn), driven mainly by continued expansion of Kling AI.Live-streaming revenue RMB 8.7bn.

c. Opex and profit: S&M expenses RMB 9.9bn (vs. RMB 10.5bn last year), ratio down to 27.9% from 30% on reduced promotions. R&D RMB 4.6bn (+34.2% YoY), 12.9% of revenue, reflecting AI-related investments including model training costs.G&A RMB 895mn (vs. RMB 897mn), roughly flat; net profit RMB 3.2bn, Adj. net profit RMB 3.9bn, Adj. NPM 11%.

II. Earnings Call Details

2.1 Management remarks – key messages

1. Kling AI

a. Commercialization: Q2 revenue exceeded RMB 850mn, up over 200% YoY, maintaining global leadership in AI video monetization.b. Models and products: The 3.0 series launched the industry's first native 4K video output for film and advertising professionals, enabling one-click, cinema-grade 4K without complex post-production. 3.0 Turbo was released concurrently to enhance creation efficiency and reduce production costs while retaining high-quality motion and precise audio-video sync.c. Ecosystem interfaces: Officially launched Kling MCP and Kling CLI, enabling AI agents to orchestrate Kling for batch content creation and expanding workflow automation and intelligent orchestration.d. Industry recognition: At Cannes Lions 2026, Kling-generated works won 1 Silver and 2 Bronze; multiple Kling creations (including 'Paper Smartphone') were shortlisted in the AIGC unit at the 2026 Beijing Int'l Film Festival. A Kling-participated work 'Stroke of Genius' won an annual micro-series recommendation award.

2. Foundation model and internal AI apps

a. Keye-VL 2.0: An upgraded multimodal model supporting 256K context for deep perception, enabling near-lossless reasoning for long-sequence video understanding, and for the first time embedding agent collaboration with potential for code parsing and tool invocation.b. AgentX: A self-evolving AI agent for industrial-grade recommender systems, autonomously handling recommender model and strategy design, evaluation and knowledge deposition, materially improving iteration efficiency.c. Marketing scene agents: Generating marketing creatives tailored by industry and client needs, with AIGC-related ad spend up over 70% YoY in Q2. Generative recommendation and intelligent delivery have expanded to live-streaming, search and shelf-based e-commerce, improving content recommendation and unlocking more client budgets.d. Organizational efficiency: The in-house general agent 'MyFlicker' integrates core internal systems and is open to all employees, with usage above 92% by Jun. Among R&D engineers, AI contributed 60% of new code.e. Wanqing (enterprise-grade model platform): Integrates high-performance inference, low-cost model customization and fully managed services, supporting internal AI scenarios and exporting foundation-model infrastructure to enterprise clients. On OpenRouter, multiple Wanqing-hosted open-source models ranked among the top by API consumption.

3. Users and content ecosystem

a. Scale: Kuaishou App Avg. DAU 412mn and Avg. MAU 797mn in Q2. AI-driven smart acquisition improved paid traffic efficiency and boosted retention for new and returning users, while ongoing traffic governance focused on highly active core users.b. Social: Private messages between mutually followed users grew over 15% YoY. Playback smoothness and smart interaction experiences were also optimized.c. Content ops: Riding the World Cup momentum in Jun–Jul, the proprietary IP 'Kuaishou World Cup Fans Cup' plus self-hosted events like Kuaishou XYStyle FC and Dream-Chasing Youth football generated 58.2bn exposures and 360mn live views cumulatively.d. Rights and paid content: Secured broadcasting rights for the 2026 CPA league via an e-commerce live co-op model. In Apr, piloted paid online music performances (TLP concert) with sales exceeding RMB 10mn; the paid model also extended to grassroots sports events, achieving regional scale in Northwest China.

4. Online marketing services

a. Non-e-commerce: Content consumption, local services and AI application clients were the main contributors to YoY growth in non-e-commerce ads, under an all-traffic coordination strategy and brand-merchant initiatives.b. Short dramas: Lower production costs and easier creation are driving rapid supply growth; by Jun, live-action and AI-generated short drama supply rose sharply vs. Jan, with Q2 ad spend on short dramas up over 100% YoY.c. Local services: Deepening comprehensive and in-store sub-verticals, enhancing deep-conversion capabilities along the delivery path and influencer marketing solutions, and launching a crowd exploration AI agent to identify high-intent users and improve lead quality and conversion.d. AI app industry: Integrating ad delivery with in-app conversion to improve retention and conversion efficiency, strengthening competitiveness to capture AI app budgets.e. E-commerce ads: Tightening all-traffic coordination between e-commerce and commercialization to improve merchant-traffic matching, with finer merchant stratification and operations. On creatives, incentivizing first-release content and diversifying recommendations; net-transaction ROI product penetration rose from 45% in Q1 to 55% in Q2, helping merchants reduce return rates.The T2000 brand program launched in Q4 last year showed initial results, with brand advertisers' ad performance outpacing overall e-commerce ads in Q2.

5. E-commerce

a. Strategy: Scale paid users, broaden supply, and deepen traffic integration between e-commerce and commercialization. Focused on high-quality buyers during the quarter, with active paid users broadly stable QoQ.b. Scenario synergy: Strengthening private-domain advantages and aligning multi-scene traffic to form a virtuous loop across content recommendation, shelf conversion and store repeat purchases, while optimizing subsidy efficiency to balance content and shelf-based commerce.c. Supply: Upgraded the Starlight program to provide tiered support for brands, large merchants, industry belts and SMBs. New merchant onboarding grew YoY and by nearly 10% QoQ, and new merchants achieving scale by the next month rose nearly 30% YoY, with quality improving.d. Brands and verticals: T2000 brands' self-operated GMV delivered strong YoY growth, contributing more to overall GMV, while brand advertisers' marketing spend grew rapidly. Content-commerce strengths in tea, alcohol and health, beauty and personal care, and fresh food continued to expand.e. Influencer ecosystem: Deepened partnerships with top influencers, increased support for mid-tier talent in advantageous verticals such as agriculture and ACG, and expanded the talent base via in-app incubation, guild partnerships and external sourcing.Q2 saw YoY growth in creators with 10k+ followers going live, with higher frequency, rising influencer distribution penetration YoY, and over 20% YoY growth in merchant–influencer matching.

f. Merchant AI tools: Over 850k merchants used the platform's free AI ops tools in Q2, covering selection and listing, creative production, ops analytics, smart delivery and AI customer service. AI is evolving from efficiency tools to full-fledged execution solutions.

6. Live-streaming

a. Revenue and supply: Q2 live-streaming revenue was RMB 8.7bn. The 'Chuanliu' program boosts new-anchor supply and early-stage growth via recruitment incentives and guild support, while strengthening independent anchor ops, identifying high-value independent anchors, and encouraging top anchors to expand into group streaming.b. AI enablement: Customizable AI gift formats continued to expand, increasing users' willingness to pay, with over 6mn users sending AI gifts in Q2. AI content understanding improved live recommendations and anchor–user matching, supporting paid user growth.Real-time multimodal gift recommendation and ranking enhanced paid experiences and efficiency, while AI interaction assistants and digital humans further improved anchor service efficiency.

7. Overseas

a. Traffic and content: Maintaining a high-quality growth strategy, with precise paid acquisition, local content production and broader creator networks, deepening content consumption among core users around real-life scenarios.b. Monetization: Leveraging key events like Brazil's Festa Junina and the World Cup, combining AI capabilities, ROI deep analytics, audience insights, innovative products and industry strategies to help clients capture marketing windows and grow quickly. Monetization potential in formats like short dramas is also being unlocked.c. E-commerce: Orders grew solidly YoY in Q2, with higher AOV via product-mix optimization and better supply, while sustaining operational efficiency and profitability.

2.2 Q&A

Q: With multiple video-generation foundation models updating recently, how do you view the competitive landscape in AI video generation, and what is Kling's strategy?

A: The global market opportunity tied to video content generation is around US$150bn, and AI video-generation models still have ample room to penetrate. The industry is in an active phase where players leverage platform ecosystems, vertical scenarios and technical strengths to compete in differentiated ways, with commercialization expanding into ads, e-commerce, film/TV, short dramas and games.Compared with the more fragmented LLM market, AI video generation is structurally more concentrated, with higher requirements for compute, data, talent and technology.Over the past two years, Kling has remained in the first tier of AI video generation, with forward-looking judgment and strong execution: launched the world's first commercial video model based on DiT in Jun 2024; introduced the first multimodal visual–language interaction architecture (MVL) in Apr 2025; and in Dec 2025 released the first multimodal video-generation model based on an Omni model. Kling's recent independent financing further strengthens its competitive position.Product positioning focuses on professional content creators—those who create video professionally and can pay sustainably—using video-generation foundation models to boost productivity. Kling excels in prompt understanding and controllable shot planning, delivering precise and coherent creative expression.It meets professionals' demands for video quality and volume, and is the world's first to support native 4K output, enabling one-click 4K without upscaling or post-processing for clearer, richer, more cinematic visuals. As model capabilities iterate, product experience improves, and professional creator ecosystems and commercialization scenarios expand—amid rapid industry adoption—Kling's long-term competitiveness and growth potential should continue to unfold.

Q: Beyond Kling, what other AI progress did you make this quarter?

A: This quarter's AI progress also showed up in organizational efficiency, AI applications in online marketing, and iterations of our core recommender systems.In marketing scenarios, AI is deeply embedded in creative generation, smart bidding and generative recommendation. For creatives, we connected user-interest modeling with video production, shifting from 'finding users for videos' to 'creating videos for users'.For smart bidding, agent-based systems learn from clients' historical delivery and conversion goals to form self-optimizing auto-bid strategies, improving long-term value and performance with stronger ROI. For generative recommendation, models truly understand ad content and user needs, translating products, live streams, search queries and industry information into semantic representations to improve matching efficiency.

On recommender iterations, we launched AgentX to build an agent-driven R&D loop. Previously, moving a recommendation idea to production required data analysis, solution design, code changes, experiment setup, A/B testing and monitoring, and attribution and review—heavily manual and algorithm-engineer dependent.Now agents act as the execution engine, freeing engineers from repetitive work to focus on goal-setting, key reviews and higher-level decisions, materially lifting both iteration speed and outcomes.

On organizational efficiency and enablement, in-house agent tools like MyFlicker have usage above 90%, covering tech R&D, data analysis and ops. In tech R&D, the Avg. delivery cycle in Q2 2026 shortened by over 10% vs. Q1, daily lines of code submitted grew over 70% QoQ under AI-assisted development, and AI contributed over 60% of new code.Wanqing not only supports internal AI scenarios efficiently but also provides foundation-model infrastructure to many enterprise clients, delivering strong revenue growth. We will continue to empower businesses and the organization with AI and expand application boundaries.

Q: Amid macro and competitive uncertainty, what measures are you taking to ease merchant pressure and drive merchant growth?

A: Macro demand remained under pressure in Q2, competition was still intense, and merchants need better visibility and certainty in their operations, which shapes our merchant strategy—tier merchants by operating profiles, and run layered operations.We provide more targeted support, policies and resource allocation to help different merchant types find appropriate growth paths.Starting in Q2, we restructured merchant service teams by business type, aligning differentiated strategies to core needs and returning policy and resource allocation to more rational levels. Previously, resources were mismatched: brand merchants lacked professional services; ad-driven merchants received e-commerce resources disproportionate to their revenue contribution; influencer/content-driven merchants had ample organic traffic but needed sustainability.We centralized brand merchants under specialized teams, rebalanced resources away from ad-driven merchants where mismatched, and used real platform followers to improve live sustainability for influencer/content-driven merchants. On that basis, we continued tiered, targeted support for brands, large merchants, industry belts and SMBs, keeping policies stable while making resource allocation more precise.We also deeply embed AI capabilities into daily merchant operations and cost management—covering creative production, smart marketing, intelligent after-sales and AI customer service—so merchants can reinvest saved resources into higher-ROI growth.On the supply side, our two strategic focuses this year remain brands and industry belts. Since Q4 last year, tighter compliance has narrowed cost differences across merchant types, raising ROI requirements for marketing spend; in this context, brands' advantages in operating capability and business stability are increasingly evident, while AI materially lowers content creation barriers.We further focused brand subsidies on key hero products, and brands on Kuaishou have entered a positive growth loop, showing genuine investment in building and growing on-platform businesses rather than treating Kuaishou merely as an exposure or off-platform conversion entry. In Q2, brands' self-operated GMV contributed more steadily to overall GMV, and their share of total marketing spend rose.For industry belts, via service providers we are penetrating belts nationwide and empowering local merchants. This model has rolled out progressively in Inner Mongolia, Yunnan, Shandong and Jiangsu, with tangible operational support and growth incentives for regionally distinctive supply, helping them establish and scale on Kuaishou.

Q: How do you view the e-commerce outlook for 2H?

A: We expect overall consumer demand to continue a moderate recovery, with spending shifting from discretionary to staples, implying merchants will prioritize predictability and the era of pure traffic-led growth is over. While we will keep supporting traffic and commission rates, we expect pressure on e-commerce ad services and commission income in 2H.For the platform, this is both a challenge and an opportunity. We will keep tilting coordinated traffic strategies toward brands and high-quality merchants with long-term operating capabilities and intent, and further strengthen smart delivery to improve predictability for merchants.By enhancing layered merchant operations and optimizing supply structure, we aim to capture structural opportunities and drive high-quality growth in e-commerce—key to navigating near-term cycles and achieving sustainable long-term growth.

Q: What is the outlook for non-e-commerce ads (including local services) in 2H?

A: Growth in online marketing services depends on the macro backdrop and client budgets. Visibility remains limited heading into 2H; last year's rapid growth in AI apps and on-demand retail creates a high base this year, and client budgets may be influenced by competitive dynamics and strategy shifts.However, some sectors still offer structural opportunities.First, content consumption led by short dramas still has room to grow. AI can keep reshaping the supply side by lowering production costs and shortening cycles, supporting richer themes, formats and greater supply: Kuaishou's anime-format short dramas numbered 60k episodes in 2025, and are expected to reach 500k in 2026.The industry will gradually shift from pure volume to quality improvements. The value of short-drama ecosystems is becoming more evident, with daily exposures reaching 230mn in Jul 2026; monetization is evolving too, with IAA rising in mix, indicating short dramas are maturing into an ecosystem model with content value and incremental ad inventory.Going forward, formats may extend from traditional viewing to IP adaptation and derivatives, interactive short dramas and virtual companionship, and integration with culture/tourism and creative industries, opening new consumption and monetization space.We will capture these opportunities through revenue-sharing incentives, all-traffic support and partnerships with premium content rights-holders, trading high-quality content for more stable traffic and monetization returns.

Second, sub-verticals with still-low penetration. In local services, while macro-linked, the numerous sub-verticals differ widely in merchant operating models, user decision journeys and conversion goals, allowing us to keep mining growth in beauty, home renovation/furnishings, real estate and more via more refined industry operations.In gaming, content consumption on Kuaishou is strong but monetization lags; we are exploring tighter integration of game live-streaming and game content with marketing budgets to unlock monetization potential. In the AI era, the development threshold for mini-games is falling further, potentially boosting supply and ad demand.

Q: How did AI empower online marketing services this quarter, and what impact did it have?

A: As clients focus more on operating efficiency and ROI, AI has become a key lever for improving efficiency, lowering costs and unlocking incremental ad budgets.On creatives, AI materially lowers creation barriers. In local services and AI tools, for example, AI helps clients and service providers produce content and materials faster and at lower cost that better fit real users, driving ad demand in these sectors.On client operations, AI is applied to more workflows, including opportunity insights, product selection and site recommendations, delivery diagnostics and reviews, and customer service, standardizing and automating processes previously reliant on manual work to improve efficiency.On the platform side, we continue to use AI models to improve matching and conversion in the ad system. Better understanding of creatives and user interests, combined with smart bidding and pricing, helps advertisers match budgets more efficiently to the right users and scenarios.Overall, structural growth opportunities in online marketing services should keep emerging in 2H, and we will continue to use AI as a core capability to optimize marketing efficiency and ROI, driving client operating efficiency and commercialization growth.

Q: What is the impact of Kling's financing on Kuaishou's cash flow, and how will cash flow and shareholder returns be arranged in 2H?

A: With Kling AI's financing completed, we now have more flexibility in cash outlays, including meeting incremental compute needs via leasing and other means, further optimizing capital allocation and, to some extent, improving group cash flow.On cash management and capex, we will continue to adhere to prudent financial policies. In terms of cadence, the bulk of capex has been front-loaded into 1H, enabling positive cash flow in Q2, and we aim to maintain positive FCF at the group level in 2H.On shareholder returns, we remain focused on creating long-term value and consistently executing an active return strategy. YTD, we have repurchased shares with total consideration of approx. HK$2.0bn and paid cash dividends of HK$3.0bn, with combined returns close to last year's full-year level.While firmly investing in AI, we continue to lift shareholder returns, and expect FY26 combined returns to exceed last year. This is a tangible reward for shareholders and reflects the company's ability to generate cash flow and confidence in long-term development.We will drive cost optimization and efficiency through prudent financial discipline, maintain a solid cash position, and balance business expansion and shareholder returns on a resilient financial foundation.

Risk Disclosure and Statement:Dolphin Research Disclaimer and General Disclosure