Kuaishou has essentially become the financial lifeline for Kling AI. As traffic-driven narratives lose their appeal across the internet platform landscape, Kuaishou is leveraging Kling to orchestrate a new wave of revenue reshuffling.

On August 19, after the Hong Kong stock market closed, Kuaishou delivered its Q2 2026 earnings report: total revenue reached 35.535 billion yuan, a modest year-on-year increase of 1.4%; adjusted net profit was 3.913 billion yuan, a 30.3% decline year-on-year. A closer look reveals that traditional core businesses are under comprehensive pressure: advertising growth has stalled, live streaming has downgraded, and e-commerce has ceased disclosing GMV figures. However, beneath the lackluster fundamentals, Kling AI emerged as a rare bright spot, generating over 850 million yuan in revenue for the quarter, a 240% year-on-year surge.

Over the past quarter, Kuaishou executed a highly publicized strategic pivot. It spun off Kling AI for independent fundraising, securing a post-investment valuation of $18 billion. Research and development investment surged 34.7% year-on-year, with the company willingly sacrificing profit margins to go all-in on AI. Consequently, the market has begun repricing Kuaishou, with Kling AI substantively reshaping the commercial foundation of this short-video company.

AI Highlights Amid a Low-Growth Landscape

The Q2 2026 earnings report for Kuaishou reveals a stark dichotomy. On one hand, overall revenue growth has slowed and traditional businesses are collectively under strain. On the other, Kling AI is racing ahead, with its user base surpassing 100 million, crafting a new growth narrative. Only by considering both aspects can one grasp the true commercial picture of Kuaishou today.

From a total volume perspective, quarterly revenue of 35.535 billion yuan, up just 1.4% year-on-year, not only falls far below the average for the internet industry but also marks the lowest growth rate since the company's listing. Core commercial revenue, encompassing online marketing and other services, grew 7.4% year-on-year, which appears acceptable but masks the weakening momentum of the broader business.

The contraction on the profit side is more direct. Adjusted net profit for the second quarter was 3.913 billion yuan, down 30.3% year-on-year, while gross margin slipped from 55.7% in the same period last year to 51.6%. Of course, the primary reason for the profit shrinkage isn't a deterioration in core operations, but rather increasingly aggressive AI research and development spending. Quarterly R&D expenses reached 4.581 billion yuan, a 34.7% year-on-year increase, with the R&D expense ratio jumping from 9.7% to 12.9%. The incremental investment has been almost entirely channeled into large model training, video generation, and AI Agent development. In essence, Kuaishou is proactively sacrificing short-term profits to secure a strategic position in AI technology.

Kuaishou's management made it clear during the earnings call that they remain cautious about the macroeconomic environment for the second half of the year and will not cut AI investment to protect short-term profits, accepting 1-2 quarters of profit pressure. This contrasts sharply with adjustments in other expense lines: sales and marketing expenses fell 5.5% year-on-year, with the expense ratio compressed from 30% to 27.9%, down 4% quarter-on-quarter. Administrative expenses in the first half also dropped to 1.661 billion yuan from 1.725 billion yuan in the prior year. This essentially means marketing budgets saved are being fully redirected to R&D, with resources overwhelmingly tilted towards technology investment.

In terms of resource allocation priority, AI has ascended to Kuaishou's top strategic focus. It is no longer confined to a single standalone product like Kling but is now embedded throughout the entire operational chain, including advertising creative production, merchant operations, live streaming interaction, content creation, and internal office workflows. Specifically, on the content production front, AI is involved in creative development, with short-drama supply increasing more than fivefold since the start of the year, boosting advertising spend. On the merchant operations side, over 850,000 merchants are using free AI operation tools covering product selection, creative assets, and customer service. Internally, over 92% of employees use self-developed AI Agents, and AgentX has improved recommendation experiment efficiency to eight times that of manual processes.

These figures demonstrate that AI is indeed enhancing Kuaishou's operational efficiency. However, the current impact of AI is primarily seen in cost reduction and efficiency gains; the portion that directly translates into revenue growth remains limited. For instance, the RaG technology contributed a 1.87% lift in advertising revenue, which is relatively insignificant compared to the overall scale and insufficient to reverse the declining trend in ad growth.

As a result, Kuaishou's revenue structure is undergoing a shift. Online marketing remains the largest revenue source, generating 20.6 billion yuan in the quarter, up 4.4% year-on-year, falling short of the market's prior expectation of 6% growth and sliding into the low single-digit range. Over the past three quarters, Kuaishou's advertising growth has consistently declined, weighed down by two factors: first, a weakening traffic base, with DAU, highly correlated with ad monetization, nearly stagnant, and MAU expansion primarily driven by low-frequency users briefly returning, which fails to convert into advertising value; second, a sluggish macro consumption environment, with budget pressure in key sectors like e-commerce. Although short-drama marketing spend has doubled year-on-year and AIGC marketing creative consumption has grown significantly, these gains are insufficient to fill the gap left by the slowdown in traditional advertising.

The contraction in the live streaming business is even more pronounced. Quarterly live streaming revenue was 8.7 billion yuan, down 13.5% year-on-year, with its revenue share dropping to 24.5%. The company officially describes this as a "proactive adjustment of high-income streamer structure to promote a healthier live streaming ecosystem." However, this appears more like a passive adjustment driven by industry saturation and stricter regulation: user willingness to tip is declining, and the risk associated with top streamers' income models is rising, making the growth ceiling for live streaming tips increasingly apparent. Hence, Kuaishou is proactively reducing reliance on top streamers to mitigate revenue volatility and reallocate resources towards e-commerce and AI.

The only segment maintaining high growth is "other services," with quarterly revenue of 6.206 billion yuan, up 18.5% year-on-year, and its share rising to 17.4%. This segment includes e-commerce revenue and Kling AI's commercialization revenue, with Kling AI contributing 850 million yuan, serving as the core incremental driver. However, changes within the e-commerce business are more intriguing. The company has stopped disclosing GMV, instead shifting focus to merchant quality, ROI, return rates, and the proportion of brand merchants as key metrics, claiming a transition from pursuing transaction scale to becoming a service and tools platform. But it's evident to observers that ceasing to disclose core metrics signals the e-commerce industry's entry into an intense phase of competing for existing market share.

Overseas operations are also entering a contraction cycle. Initially, Kuaishou's overseas strategy pursued rapid DAU expansion through subsidies and cash burning to gain scale. In Q2 2026, it shifted definitively towards high-quality growth, prioritizing e-commerce monetization, increasing average order value, and optimizing product mix, abandoning blind user subsidies. Following this adjustment, overseas revenue in the second quarter was 1.179 billion yuan, accounting for less than 3.4% of total revenue, making it difficult for this segment to shoulder the responsibility of opening up new growth space in the near term.

Meanwhile, subtle changes are occurring in the user base. In Q2, Kuaishou's average MAU reached 797 million, up 11.5% year-on-year, hitting an all-time high. Average DAU was 412 million, up only 0.8% year-on-year, nearly at what appears to be Kuaishou's ceiling. The significantly higher growth rate of MAU compared to DAU suggests that sports rights, paid performances, and expanded short-drama supply are driving periodic user returns. However, these new users are mostly low-frequency MAU returning across different scenarios, whose time spent, engagement depth, and monetization value are far lower than core DAU users. On a positive note, community attributes are strengthening: two-way private message users grew over 15% year-on-year, and social relationship chains are becoming a differentiating foundation for Kuaishou compared to pure short-video consumption platforms.

Three Hurdles Facing Kling AI's Breakthrough

Undoubtedly, Kling AI is the highlight of this quarter's earnings and a core variable supporting Kuaishou's valuation. It achieved 850 million yuan in quarterly revenue, 240% year-on-year growth, surpassed 100 million global users, and attracted nearly 50,000 enterprise clients. These figures prove the commercialization potential of AI video, but Kling AI's path forward is far from smooth, with persistent concerns over losses, competition, and talent retention.

First, Kling AI's commercialization speed has exceeded market expectations. From surpassing $100 million in annualized revenue in March 2025, to exceeding $20 million in monthly revenue by the end of 2025, and then reaching over 650 million yuan in Q1 2026 and over 850 million yuan in Q2, the revenue curve remains steeply upward. Product iterations continue: it launched the industry's first native 4K direct output feature, bridging the professional imaging delivery gap; released the Turbo version to enhance generation speed, catering to high-frequency production scenarios like short dramas and ads. Viral content like "Paper Phone" and "Baseball Live Special Effects" has validated its deep narrative capabilities and mass appeal. Is AI video now directly delivering finished films?

The flip side of high growth is expanding losses. According to financial data, Kling AI lost 500 million yuan in 2024; losses widened to 1.9 billion yuan in 2025 against revenue of just 1.1 billion yuan. In 2026, Kuaishou has intensified investment further: full-year capital expenditure is approximately 26 billion yuan, exceeding Kuaishou's total net profit of 20.6 billion yuan for the entirety of 2025, with AI computing power consuming the lion's share. Additionally, Q2 2026 R&D expenses increased by 1.18 billion yuan year-on-year, mostly directed towards large model training. In other words, Kling AI's revenue is growing quickly, but its cash burn rate is even faster. The gap between revenue and investment needs to be narrowed over the coming quarters.

In July this year, Kling AI completed a financing round of nearly $3 billion, seemingly replenishing its arsenal, but it's far from sufficient when competing alongside tech giants like ByteDance and Tencent. According to reports, ByteDance plans to increase its 2026 capital expenditure by about 25% to 200 billion yuan, with a significant portion allocated to AI computing infrastructure. Kuaishou's planned 26 billion yuan capex for 2026 is less than one-seventh of ByteDance's, let alone capable of altering the long-term competitive landscape of the global AI video sector. Moreover, there's a clear temporal mismatch between computing power investment and commercialization returns. The 26 billion yuan capex, once deployed, requires multiple cycles of model iteration, market promotion, and paid conversion before gradually generating revenue.

Second, Kling AI's former core advantage was its first-mover position in professional video generation: native 4K direct output, controllable storyboard capabilities, and image quality were once industry-leading. But entering 2026, competitors are rapidly closing the gap, and the technological moat is being eroded. Kling AI urgently needs to prove itself with new models. In the domestic market, Seedance accounts for over 80% of computing power consumption and nearly 95% penetration in the short-drama industry, backed by Douyin and TikTok's ecosystem of over two billion monthly active users globally. The Seedance 2.5 released at the end of July supports generating 30-second long takes in a single pass, while Kling currently remains at 15 seconds, already lagging on this metric.

Meanwhile, open-source pricing strategies are equally aggressive. On August 3, MiniMax H3 was open-sourced, becoming the industry's first top-tier video generation model to be released openly. Its official API is priced at 0.8 yuan per second for 2K resolution, but some third-party platforms, such as Metaso AI and OiiOii, offer prices as low as 0.09 yuan per second through points systems, limited-time subsidies, or package deals. This means competitors can conduct secondary development based on the model, further driving down creative costs and directly impacting Kling's subscription and API revenue.

Additionally, structural talent mobility is becoming a critical variable in the technological arms race. Zhang Di, who led Kling's creation, departed in August 2025. On August 6, 2026, Wang Xintao, a senior researcher at Kling and author of the open-source super-resolution projects Real-ESRGAN and GFPGAN, a key founder of Kling's image quality capabilities, also confirmed his departure. Whether these departures represent the loss of irreplaceable core figures remains uncertain, but the intense competition for top talent in the video generation field is an undeniable fact. As the AI video industry explodes, companies are aggressively poaching AI professionals, and Kling's core R&D personnel have long been prime targets for headhunters. To retain talent, Kuaishou has had to establish a separate option pool for the Kling team. According to our information, when Kling AI raised funds independently, it simultaneously adopted three share participation plans, with an equity incentive cap set at 15%. Kling CEO Gai Kun holds 3% equity with ten times voting rights. But the race between equity incentives and talent flow is ongoing; the speed of incentive mechanism construction must keep pace with the intensity of model competition. Otherwise, incentives risk becoming mere "pie in the sky," and if expectations aren't met, talent attrition will worsen.

Finally, the spin-off and independent financing of Kling AI is seen as a key step in Kuaishou's valuation restructuring. However, looking at the equity structure, Kuaishou still holds 68.33% economic interest after the financing, maintaining absolute control, and Kling's financial statements remain consolidated. This means Kling AI's losses will continue to appear on Kuaishou's income statement, and the capex investment pressure still falls on Kuaishou. The $3 billion raised sits in Kling's accounts and has no direct bearing on the parent company's profits or cash flow. In simple terms, this is a split without separation: Kling AI is nominally independent, but its losses are still attributed to Kuaishou, while the funds raised can only be used for Kling's own development.

Notably, this capital increase agreement includes clear IPO performance clauses. If Kling fails to complete an IPO before October 30, 2031, current external investors have the right to exercise a repurchase option, requiring the company to redeem their shares at the original principal plus 8% simple interest per annum, with Kuaishou bearing joint repurchase obligations. To some extent, this provides risk protection for financial investors, akin to a debt-like investment with an embedded call option. Even if Kling AI fails to list independently, investors can secure an 8% annualized return and exit unscathed. Additionally, the agreement includes a restrictive clause: for 5 years, Kuaishou cannot directly or indirectly control any other entity primarily engaged in video generation model businesses. This forces Kuaishou to commit resolutely to this investment, a do-or-die gamble. The 5-year performance period also confirms a fact: both Kuaishou and investors recognize that Kling is not yet ready for a high-valuation listing and needs sufficient time to deliver growth and digest its valuation.

Conclusion: A Negative Valuation for the Core Business

Following Kling AI's spin-off financing, the capital market's valuation logic for Kuaishou has genuinely changed. For instance, Tencent signaled its stance through a precise share swap. On July 2, when Kling AI's financing list was announced, Tencent invested 1.363 billion yuan as one of the lead investors. On July 6, Tencent reduced its stake in Kuaishou by 273 million shares through off-market block trades, lowering its holding from 15.68% to 9.37%. Although Tencent later claimed confidence in Kuaishou's long-term prospects, its actions clearly indicate skepticism about Kuaishou's overall valuation, opting instead to bet only on Kling AI, the asset with the most imaginative potential.

Based on the latest calculations: Kuaishou has approximately 121.3 billion yuan in available funds on its balance sheet. Kling's post-investment valuation is $18 billion, with Kuaishou holding a 68.33% stake, corresponding to an equity value of about $12.3 billion, roughly 96 billion Hong Kong dollars. Yet Kuaishou's current total market capitalization is only 163.5 billion Hong Kong dollars. The value of cash plus the Kling equity stake alone exceeds the total market cap, implying the market assigns a negative value to the short-video core business. A mature business generating nearly 4 billion yuan in quarterly net profit and 5.9 billion yuan in operating cash flow is being priced at zero or less by the market. The logic behind this is clear: when core traffic growth stalls, the market is only willing to pay for the AI narrative, effectively making Kuaishou a financial "blood bag" for Kling. So, viewing Kuaishou at this juncture, its predicament serves as a microcosm of all mobile internet legacy products facing growth saturation.