AI investments lead the way, with profit returns following. Kuaishou's AI value realization period has arrived earlier than expected.

Financial report data shows that in 2025, Kuaishou's total revenue reached RMB 142.8 billion, a year-on-year increase of 12.5%. Among this, revenue from online marketing services was RMB 81.5 billion, up 12.5% year-on-year; revenue from live streaming services was RMB 39.1 billion, a 5.5% year-on-year increase; and revenue from other services, including e-commerce and Kling AI, was RMB 22.2 billion, a 27.6% year-on-year rise.

In 2025, Kuaishou's annual adjusted net profit was RMB 20.6 billion, a 16.5% year-on-year increase, with the adjusted net profit margin rising to 14.5%.

This financial report reflects a high input-output ratio for AI, with AI large models becoming the core engine driving growth, the most prominent being Kling AI.

Cheng Yixiao, founder and CEO of Kuaishou, revealed that as of January this year, Kling AI's annualized revenue run rate (ARR) has exceeded $300 million, with monthly revenue growing by 25% compared to December last year.

Cheng Yixiao specifically mentioned that in the recent hit drama "Peaceful Years" produced by Huace Film & TV, Kling AI has been deeply involved in virtual scene and special effects production, significantly reducing production costs while ensuring high-quality content output.

Cheng Yixiao stated, "This year, the company has strong confidence in achieving more than a 100% year-on-year increase in revenue for Kling AI."

Notably, Kuaishou's AI investments have been both rational and precise, with a strict benchmark set for this purpose.

Kuaishou's CFO, Jin Bing, shared several key metrics during the conference call. In 2025, despite investing approximately RMB 15 billion in computing infrastructure, Kuaishou still achieved a free cash flow inflow of nearly RMB 12 billion, with cash reserves further increasing to RMB 104.9 billion.

In 2026, Kuaishou's AI investments will increase significantly. CFO Jin Bing stated that the group's overall Capex investment is expected to reach around RMB 26 billion, an increase of approximately RMB 11 billion compared to 2025, with continued investments in AI computing power and infrastructure. Although the investment scale will continue to expand, Kuaishou remains focused on achieving healthy free cash flow growth. It is this meticulous investment strategy that has enabled Kuaishou to realize a value closed loop (closed loop) from technological breakthroughs and scenario applications to commercial returns.

Kuaishou is efficiently converting AI potential into tangible performance growth.

Kuaishou's AI Landscape: Dual-Track Driven

Kuaishou's AI strategy currently operates along two main lines.

Kling serves as the fulcrum for leveraging AI, influencing the development of the AI video sector through continuous technological iterations. From Q4 2025 to Q1 2026, Kling successively launched the Kling O1, Kling 2.6, and Kling 3.0 series models. Its technological approach has evolved from single-point capabilities to a unified architecture and complete workflow. Kling is increasingly seen as Kuaishou's new calling card in the AI era, driving user growth and enhancing brand recognition.

Financial report data shows that by the end of 2025, Kling AI's global user base had surpassed 60 million, with over 600 million videos generated and API services provided to more than 30,000 enterprise clients and developers.

On the B2B front, Kling AI is entering deeper waters, evolving from a video generation tool into a creative infrastructure capable of addressing pain points and adapting to professional scenarios such as film and television, advertising, short dramas, and gaming.

As a result, Kling AI's commercialization progress has accelerated significantly, showing a consistent upward trend. In Q4 2025, its total revenue reached RMB 340 million; in December 2025, monthly revenue exceeded $20 million, with an ARR of $240 million; by January 2026, ARR continued to grow, surpassing $300 million.

The other main line focuses on leveraging the main platform business as the foundation, allowing AI capabilities to flow and integrate across various Kuaishou businesses, achieving cost reductions, efficiency gains, and even providing incremental value. From content understanding and recommendation distribution to advertising placement, e-commerce search, merchant operations, and R&D collaboration and organizational efficiency optimization, AI has become an integral part of Kuaishou's ecosystem.

The advantages of parallel operation along these two tracks are evident. The main business enhances efficiency and stability, "supplying blood" to AI R&D investments. Kling, as the vanguard, explores new markets and creates new growth curves, injecting fresh vitality into the core business.

AI investment, returns, and continued escalation have formed a virtuous cycle at Kuaishou. Financial report data shows that in 2025, Kuaishou's R&D expenditure was RMB 14.5 billion, an 18.8% year-on-year increase, reflecting its long-term commitment to technological investment to some extent.

Another key investment is in computing infrastructure, aligning with the broader trend of expanding Token consumption.

Jin Bing stated that computing investments in 2026 will reach approximately RMB 26 billion, an increase of about RMB 11 billion compared to 2025. This "gift package" includes computing support for Kling large models and other foundational large models, as well as routine server procurement, data storage and processing, and the construction of data and computing centers.

In the long run, Kuaishou's AI strategy stands out. On the one hand, it explores technological limits; on the other, it establishes business performance floors. Kling's technological limits lead toward multimodal fusion and world models, while its floor involves reshaping the efficiency of people, goods, and scenes with AI. The limits determine how high Kuaishou can fly, while the floor determines how steady it can stand.

If we were to summarize the characteristics of Kuaishou's AI, it might be described as "walking on two legs," deeply binding cutting-edge technologies with core businesses. This approach avoids the dilemma of pure technology companies being "acclaimed but unprofitable" and the risk of homogenized competition in AI application layers.

If this system proves successful, Kuaishou will not only be defined as a "short video" company but as an infrastructure platform with AI as its core productivity.

Applying AI Where It Matters: From Content Understanding to Commercial Transformation

In the fourth quarter, AI large model technologies began to show initial success in empowering Kuaishou's content and commercial ecosystems, with the core being the continuous upgrading of "content understanding, recommendation distribution, and interactive experiences."

In terms of content understanding, Kuaishou's self-developed Keye multimodal large model enhances content understanding capabilities in short video and live streaming scenarios, breaking through the limitations of traditional shallow recognition and introducing a tagging system, TagNex, for more detailed content structuring. On the recommendation front, Kuaishou further upgraded its end-to-end generative recommendation large model and launched the One Rec-V2 version, continuously enhancing recommendation precision. In terms of interactive features, products like "AI Vientiane " gifts continue to upgrade, supporting customizable special effect gifts with over 1 million cumulative generations.

This is no longer just an AI-driven improvement but a transformation from "guessing what you like" to "understanding what you love."

Under precise full-scenario distribution, enhanced user retention and interaction have solidified Kuaishou's user community foundation. As of December 31 last year, Kuaishou's average daily active users in the fourth quarter reached 410 million, with average monthly active users at 740 million, and users' average daily usage time remaining at 126 minutes.

AI has brought a second wave of prosperity to the user community and injected new momentum into commercialization and e-commerce business growth.

Material production has become one of the most mature fields for AI applications. It is understood that the cost of a single piece of content on the Kuaishou platform has dropped to a few cents, prompting advertisers to focus more on quality than price. Jiang Peng, Vice President of Kuaishou Technology and Head of Commercialization Algorithms, mentioned in a previous media interview that content produced through Kuaishou's AI outperforms client-produced content in terms of placement quality, with an overall conversion rate improvement of 20%-30%.

Commercialization breakthroughs are not isolated; last year, Kuaishou deployed end-to-end generative recommendation large models and bidding large models specifically tailored for its advertising system, upgrading its commercialization engine. After deployment, the platform can provide advertisers with a fully automated process for content generation, infrastructure setup, and intelligent bidding.

To date, this technology has achieved approximately 90% automation in the entire placement process, requiring clients only to set ROI targets and placement products. Expanding scenarios and innovative applications have driven double-digit growth in online marketing service revenue, which reached RMB 23.62 billion in the fourth quarter, a 14.5% year-on-year increase.

Fourth-quarter financial reports show that continuous upgrades to generative recommendation large models and intelligent bidding models have boosted domestic online marketing service revenue by approximately 5%. AIGC marketing material consumption reached approximately RMB 4 billion. This demonstrates that Kuaishou's AI is not just burning money but is already generating incremental value in its core advertising and e-commerce businesses.

"The earliest and most enthusiastic adopters of AI come from e-commerce," explained an industry insider. Intense industry competition has stimulated demand, while the high standardization of underlying data has created an ideal testing ground for AI implementation. During the 2025 Double 11 shopping festival, AI was fully implemented, and Kuaishou was no exception.

Short video-live streaming synergy has become the "standard answer" for Kuaishou merchants to achieve growth. As AI accelerates its penetration into content production, promotion, operations, customer service, and other aspects, merchants have more time and resources to achieve short video-live streaming synergy.

Leveraging its AI technological accumulations, Kuaishou E-commerce launched an AI image-to-video feature, enabling one-click generation and publication of short videos in line with the platform's style, helping merchants achieve dual-domain traffic coverage for "live streaming + short videos."

During the Double 11 promotion, merchants released over 300,000 AI-generated videos with product links through the Kuaishou Store merchant workstation; among them, 15,500 merchants chose to release AI-generated image-to-video content with product links, publishing over 261,000 such videos. The daily average number of releases increased by over 12% compared to the 30 days before the promotion, while daily average GMV increased by nearly 30%.

From Recommendation to Reasoning: Agents Reconstruct Kuaishou's New AI Engine

Looking ahead, the accelerated implementation of Agents will become the main theme of AI's in-depth transformation of the content industry. In online marketing scenarios, Kuaishou will develop AI Agents for e-commerce merchants to achieve ultra-automated placement, covering the entire process of intelligent product selection, creative editing, AI material generation, intelligent placement and dynamic pricing, AI customer service assistance, and post-placement data analysis, further lowering placement barriers and improving placement effectiveness and cost stability. In e-commerce scenarios, Kuaishou will also enhance user search experiences through search and recommendation Agents, boosting search order volumes, with further disclosures on related progress to be made at the appropriate time.

Additionally, as an independent AI product, Kling will also become a more significant growth engine by advancing Agent capabilities. In the future, Kling will continue to expand into more modalities such as actions and expressions, further enhancing the controllability of video generation and continuously addressing complex scenario settings and consistency issues. At the product level, the model will gradually acquire more complete full-process autonomous creation capabilities, including storyboarding, character and scene consistency control, audio-visual synchronization, lighting, and camera movement design, centered around user needs.

In 2024, Kling quietly gained popularity. During its initial launch, overseas users flooded the X platform with requests for testing qualifications, filling 500-person chat groups instantly. With the 2.0 version, queued testing and "pull-down creation" became trends. That year, Kuaishou's AI garnered significant attention.

A year later, as Kuaishou's AI began to permeate its entire business ecosystem, including content, commercialization, and e-commerce, financial report data validated the incremental value brought by new technologies.

This year, we are witnessing a qualitative shift in the driving forces behind Kuaishou's growth. With the iteration of foundational models and AI Agents, a new transformation is quietly unfolding: reasoning is reconstructing everything from the recommendation era.