Kuaishou (Minutes): Short-term AI investments are expected to impact margins by 1-2pct

The following is the minutes of Kuaishou's 4Q24 earnings conference call. For financial report commentary, please refer to Kuaishou: Can AI Boost the "Lao Tie Economy"?

I. Core Financial Data

Our DAU has exceeded 400 million, marking an important milestone in our user ecosystem.

In 2024, our total revenue reached RMB 126.9 billion, a year-on-year increase of 11.8%. Our adjusted net profit grew by 72.5% year-on-year to RMB 17.7 billion, with an adjusted net profit margin of 14%, reflecting a rapid improvement in profitability.

In the fourth quarter, our total revenue grew by 8.7% year-on-year to RMB 35.4 billion, mainly driven by the growth of our online marketing services and e-commerce business. In the fourth quarter, online marketing service revenue increased by 13.3% year-on-year, rising from RMB 18.2 billion in the fourth quarter of last year to RMB 20.6 billion, accounting for 68.3% of total revenue.

Quarterly revenue surpassed RMB 20 billion for the first time, contributing to our annual online marketing service revenue growth of over 20% year-on-year. This growth was primarily due to our continuous improvements in algorithms and models, as well as our smart delivery products.

Revenue from other services, including e-commerce, reached RMB 4.9 billion in the fourth quarter, a 14.1% increase from RMB 4.3 billion in the same period last year. This growth was mainly driven by the increase in e-commerce GMV, which boosted e-commerce commission revenue.

We further advanced our holistic operations by leveraging our rich content ecosystem, continuously improving the collaborative efficiency of monetized traffic, and expanding the number of merchants in Kuaishou's business.

At the same time, we provided a richer product supply and more diversified marketing activities, continuously enhancing the user shopping experience, consolidating e-commerce user loyalty, and promoting repeat purchases. All these measures led to an increase in the number of monthly active paying users and monthly active merchants in e-commerce.

In the fourth quarter, our live streaming revenue was RMB 9.8 billion, a decrease of 2% from RMB 10 billion in the fourth quarter of last year, with the year-on-year decline continuing to gradually narrow.

We are continuously developing diversified leading categories in live streaming, deepening our expertise in segmented verticals, accelerating the discovery and support of local small and medium-sized streamers, and continuing to cultivate a healthy and sustainable live streaming ecosystem. We effectively increased the number of streamers and agencies and their engagement.

The cost of revenue in the fourth quarter grew by 6.5% year-on-year to RMB 16.3 billion, accounting for 46% of total revenue. This increase was mainly due to the rise in revenue-sharing costs and related taxes that aligned with revenue growth, including depreciation of property and equipment, depreciation of right-of-use assets, and amortization of intangible assetsGross profit increased by 10.6% year-on-year to RMB 19.1 billion. The gross profit margin was 54%, an increase of 0.9 percentage points year-on-year.

Next are the expenses. Sales and marketing expenses increased by 11% to RMB 11.3 billion, accounting for 32% of total revenue. The main reason for the increase was the rise in spending on online marketing services and e-commerce business promotions.

Research and development expenses were RMB 3.5 billion, a year-on-year increase of 4.7%, accounting for 9.8% of total revenue, down from 10% to 10.1% in the fourth quarter of last year. Management expenses increased by 15.2% year-on-year to RMB 866 million, with the proportion of total revenue remaining the same as the same period last year. The increase in R&D and management expenses was mainly due to higher employee benefits costs, including related share-based payment expenses.

Net profit for the fourth quarter increased by 10% year-on-year to RMB 4 billion. Adjusted net profit at the group level increased by 7.8% year-on-year (RMB 4.7 billion), reaching a new quarterly high, with an adjusted net profit margin of 13.3%.

As of December 31, 2014, our balance sheet remained robust, with cash and cash equivalents, time deposits, restricted cash, and wealth management products totaling RMB 92.8 billion. By enhancing liquidity and efficient working capital management, we generated RMB 8.6 billion in positive operating cash flow in the fourth quarter.

In addition, we actively implemented a shareholder return plan in 2024. During 2024, we repurchased approximately HKD 6 billion worth of shares, about 123 million shares, accounting for approximately 2.8% of our total share capital at the beginning of 2024.

II. Management Report

Over the past year, under our technological guidance, we have improved our user-centric business philosophy, leveraging technology to empower content and business ecosystems. This has created greater value for our users and partners and driven our strong financial performance in the fourth quarter.

Our average daily active users remained above 400 million, with total revenue in Q4 increasing by 8.7% year-on-year to RMB 35.4 billion. Our adjusted net profit reached a new quarterly high of RMB 4.7 billion in the fourth quarter.

For the full year of 2024, our total revenue increased by 11.8% year-on-year to RMB 126.9 billion, with adjusted net profit increasing by 72.5% year-on-year to RMB 17.7 billion, and an adjusted net profit margin of 14%. These earnings reflect our steady improvement in profitability. Today, with the continuous advancement and breakthroughs in large models and applications, AI is increasingly pushing the boundaries of video creation, user experience, and the broader business ecosystem.

Kuaishou is at the forefront of this critical intersection of technology and large video models. While we are driving transformative changes across the industry landscape, we firmly believe that artificial intelligence is not just a tool for improving efficiency; it is the core engine for creating greater value in the business ecosystem while continuing to drive traffic growth on the platform. Importantly, our work actively participates in and drives profound transformation across the entire industry1. Business Development Highlights

(1) User Ecosystem: In the fourth quarter, Kuaishou's average daily active users reached 401 million, with monthly active users hitting 736 million, representing a year-on-year growth of 5%. The average annual usage time per DAU on the Kuaishou App was 125.6 minutes. Although the total user usage time increased by 5.8% year-on-year, our optimized user growth strategy has continuously improved the engagement and retention rates of new users.

We focus on optimizing features and enhancing the end-to-end viewing experience for users. By improving resolution and ensuring smooth streaming playback, we continuously enhance the video sharing and communication experience. Additionally, we introduced various innovative features in private messaging, which drove a nearly 5 percentage point year-on-year increase in daily private messaging penetration among urban users in the fourth quarter. We also found that our common ranking strategy led to a more than 40% year-on-year increase in the time users spent on the comment feature in the fourth quarter.

In terms of traffic distribution, we ensure that high-quality content with unique sequential characteristics reaches the right audience through more recommendations and higher exposure, anticipating deeper engagement and stronger connections.

For example, in the general knowledge sector, offline travel and online live streaming sparked extensive user participation and discussion, with travel performances during the New Year period accumulating over 140 million views, including live broadcasts and short videos. This faster connection between online and offline regional users and creators reinforced core user stickiness while promoting the inheritance of traditional Chinese arts and cultural heritage.

(2) Advertising: In the fourth quarter, revenue from online marketing services grew by 13.3% year-on-year, reaching RMB 20.6 billion. This marks the first time this segment's quarterly revenue exceeded RMB 2 billion. Furthermore, this revenue is expected to grow by over 20% year-on-year in 2024. Our quarterly revenue growth was primarily driven by a high single-digit year-on-year increase in cost per thousand impressions (CPM).

We captured incremental opportunities, such as commercializing word games and online marketing services, and enhanced our marketing service recommendation model through artificial intelligence. This improved our ability to predict accurate results and increased the conversion efficiency of marketing maturity. Additionally, intelligent marketing solutions like UAS delivery solutions and animation platforms significantly enhanced the advertising performance of our marketing clients.

In the fourth quarter, external circulation advertising continued to be the main driver of our online marketing services. Particularly, the content consumption industry, including short videos and games, experienced faster growth. Notably, marketing spending from commercialization or short videos to search increased more than threefold year-on-year.

On the product side, we upgraded the UAS delivery solution from rule-based decision-making to model-based decision-making. As a result, marketing spending in the U.S. accounted for over 55% of total external marketing spending in the fourth quarter, and closed-loop marketing services continued to support merchants in leveraging high-quality traffic on Kuaishou to enhance operational efficiency.

In the fourth quarter, merchants using our full-platform marketing solutions and intelligent hosting products accounted for approximately 55% of total closed-loop marketing spending in e-commerce and marketing. By focusing on improving the advertising effectiveness and performance of small and medium-sized merchants on our platform, we drove these merchants to achieve over 30% year-on-year growth(3) E-commerce: Leveraging our advantages in content scenarios, our e-commerce GMV grew by 14.4% year-on-year in the fourth quarter, reaching RMB 462.1 billion. We provided synergistic efficiency across all platform traffic, enabling us to better meet the needs of e-commerce users.

In the fourth quarter, the number of monthly active paying users in e-commerce increased by 10% year-on-year, reaching 143 million, with a monthly active user penetration rate of 19.5%. We also launched targeted programs to optimize coupon incentives through key promotional events to acquire new users from South China and enhance their activity levels.

During the Double 11 shopping festival, we gained over 7 million net new users, consolidating user loyalty to promote repeat purchases. Looking ahead, we will continue to adhere to a user-centric strategy and collaborate with merchants and KOLs to optimize the consumer shopping experience.

Regarding merchants, in the fourth quarter, merchants continued to thrive in the Kuaishou e-commerce ecosystem. The average number of monthly active merchants increased by over 25% year-on-year. The GMV of small and medium-sized merchants saw significant growth in the fourth quarter, mainly benefiting from our three core policies: improving new merchant recruitment policies, optimizing existing merchant policies, and leveraging diversified scenarios. To encourage new merchants to use Kuaishou, we launched the Golden Bounty Program and provided traffic through targeted scenarios. These programs helped early-stage merchants increase traffic, reduce operational costs, and align incentives with their key growth cycles and transition points.

At the same time, we further stimulated content-based scenarios through diversified activities and marketing tools, including KOL competitions to incentivize broadcasters. During 11 promotional events, over 39 million users joined KOL fan groups, and more than 2,500 live-streaming rooms achieved GMV exceeding RMB 1 million.

In the diversified scenarios of the fourth quarter, the GMV of short video e-commerce grew by over 50% year-on-year. As an important competitive advantage of our content-based scenarios, the interaction between short videos and e-commerce, as well as between short videos and live streaming, played an incremental role in helping merchants with their businesses. Additionally, Kuaishou's general merchandise GMV contributed 30% of our total e-commerce GMV in the fourth quarter. Its growth was primarily driven by strong supply and demand, consistently outperforming overall GMV growth.

In the fourth quarter, the average number of daily active merchants increased by over 50% year-on-year, and the average daily paying users in the mall grew by nearly 40% year-on-year. As our e-commerce increasingly complements our content, we strengthened the platform's control over popular products, enhancing the overall operational efficiency of merchants.

(4) Live Streaming Rewards: The revenue from the live streaming business in the fourth quarter was RMB 9.8 billion, with the year-on-year decline continuing to narrow compared to the previous quarter. We are committed to building a healthy and sustainable live-streaming ecosystem in the long term and driving diversified growth through high-quality content. By the end of the fourth quarter, the number of partnered guilds increased by over 30%, and the number of broadcasters in the guilds grew by over 60% year-on-yearIn terms of supply, leading categories continue to create value, such as multi-host live streaming, group live streaming, and large stage live streaming. In addition, by expanding user engagement in rural towers, we have effectively enhanced our ability to discover and support local small and medium-sized streamers.

In the fourth quarter of game live streaming, we explored comprehensive cooperation with key games, including Peacekeeper Elite and Crossfire, in areas such as streamer growth, content co-creation, and event promotion, while deepening our expertise in fighting games and other niche verticals.

Moreover, as our fast-hiring business developed in the fourth quarter, the average daily number of users submitting resumes was quite high, with a year-on-year increase of over 100%. The number of matches increased by over 270% year-on-year. For Li Auto, the average daily lead generation surged by over 260% year-on-year.

(5) Overseas Business: We are deeply engaged in Brazil, continuing to invest in local content operations and brand marketing. We have made breakthroughs in innovative user acquisition channels, with daily active users in Brazil increasing by 9.3% year-on-year. Benefiting from an optimized traffic distribution mechanism and collaboration with top local IP resources, we have gradually achieved a diversified content ecosystem, with user engagement steadily increasing. The average daily usage time exceeded 75 minutes year-on-year and quarter-on-quarter. Thanks to these improvements, our total overseas revenue maintained a rapid year-on-year growth of 52.9% in the fourth quarter. Notably, online marketing revenue increased by 83.5% year-on-year.

At the same time, due to our effective control of costs and expenses, the operating loss of our overseas business narrowed by 57.2% year-on-year. Additionally, we have preliminarily validated our e-commerce business model in Brazil, achieving continuous growth in order volume and improving subsidies and operational efficiency. These early successes have unleashed the potential for healthy and sustainable development in the Brazilian market.

(6) Local Life: The GMV of local life services in Q4 increased by more than double year-on-year. We focus on urban clusters with a strong user base and continuously pay attention to user needs by further optimizing new applications for price consumption capability and comparison capability. These efforts have enhanced our value-for-money consumption experience, driving a 52.4% year-on-year increase in average monthly paying users in Q4.

We are also committed to improving content quality and optimizing user experience, which has steadily improved conversion efficiency and monetization, with Q4 local service revenue increasing by 2.6 times year-on-year.

As we further optimize the infrastructure of our commercialization products, we have also strengthened cooperation with more high-quality local operators and tiered leading merchants by leveraging differentiated traffic resources, driving merchants to achieve incremental growth.

At the same time, to pursue a higher return on investment for healthy and sustainable growth in local services, we have increased subsidies and operational efficiency, continuing to narrow our operating losses in the fourth quarter.

Looking back over the past few years, despite facing numerous challenges, we have achieved systematic growth, offsetting the impact of partners, relying on a strong content ecosystem, continuously improving structures, and expanding business scenarios. In this new era, we are committed to advancing our AI strategy and consistently adhering to a technology-driven, user-centered business philosophy, deeply understanding user needs, continuously expanding current products, and nurturing an AI content and business ecosystem based on trust communities to create long-term value for our partners and shareholders.

(7) Development of Kling's Applications

As a major content and consumption platform for global users, it should be deeply embedded in every industry of users' daily lives. The demand for returns and more diverse top-tier content from users is becoming increasingly strong. Through elements of large video models, we can significantly lower the entry barriers for video creation, especially the cost obstacles.

Users can now generate high-quality videos through simple prompt images. This allows us to unleash the huge creative potential of non-professional creators. We believe that "Kling" is becoming the new foundational infrastructure for creation in the AI era. Currently, it is a powerful resource for PGC and PDC creators, helping them efficiently produce high-quality videos and images.

Secondly, Kling AI will evolve into a storytelling tool with film-level video generation capabilities, empowering people to create engaging visual stories and become directors based on generative AI, forming an interactive community driven by AIGC while creating large-scale applications in commercial scenarios.

Looking ahead, through the empowerment of "Kling," we hope to transform and upgrade our existing business while opening a new track for AI-driven video content production.

By doing so, we will strengthen and consolidate our position as a leading platform for video production and consumption communities. We launched "Kling" last June. Since its launch, it has quickly gained significant attention among global video creators and artists, attracting over 1 million applicants in just one month. Encouraged by positive user feedback, the Kling team rapidly improved the model, releasing new iterations approximately every two weeks.

By the end of last year's fourth quarter, we launched Kling AI 1.6, which performs excellently in response actions, temporary behaviors, and movements. It also significantly enhances visual quality, improving style consistency, color accuracy, light and shadow dynamics, and rendering details, keeping us at the forefront of global technological advantages.

Kling AI 1.5 pro is currently one of the top two large models globally, according to evaluations from the well-known AI assessment site Artificial Analysis. The AI ranking is tied with Google View 2, far ahead of other domestic video generation large models. In the fourth quarter of 2024, we officially launched the standalone Kling AI App, providing global users with multiple access methods, including through applications and web pages.

As Kling AI's capabilities continue to improve, its user base is accelerating growth, while Kling's commercialization has also made progress, thanks to personal user access channels such as membership subscriptions and enterprise API interface solutions.

As of February, since Kling AI achieved monetization, we reached the significant milestone of cumulative revenue exceeding 100 million RMB for Kling AI. We expect Kling AI to achieve another leap in revenue by 2025In the next phase, we will continue to optimize and iterate the technical capabilities of AI Kling through focused R&D efforts and rapid resource investment in the Kling large model, maintaining its industry-leading position.

(8) AI-Driven Marketing

With the continuous advancement of Kling AI, we firmly believe that technology and its applications are creating value for our partners on the platform, especially our marketing clients and e-commerce merchants. Therefore, we will upgrade AI algorithms and underlying recommendation technologies by closely monitoring and learning from the technological advancements of DeepSeek and other global foundational models, which will continuously improve the matching efficiency and conversion rates between users and products, thereby bringing higher returns on investment for our marketing clients.

We will also gradually upgrade our magnet engine to the next generation of AI-driven intelligent business engines. Through large multimodal models, we have generated more creative and user-friendly marketing solutions for marketing clients, significantly reducing their content production costs and freeing up more budget to focus on brand promotion and traffic conversion. At the same time, AI-generated virtual digital human live-streaming solutions will lower the entry barriers for marketing, helping clients achieve breakthroughs in sales.

In the fourth quarter, the daily expenditure on AIGC short video marketing materials and virtual digital human live streaming solutions exceeded RMB 30 million, with a daily peak reaching RMB 60 million. In addition, by using large models to analyze product semantics and user behavior, and integrating model architecture and sequence representation, we have achieved a model estimation capability of 45 hours.

We have enhanced the advertising performance of marketing materials, driving more marketing expenditure. AI recommendation technology has also provided us with the opportunity to redefine the e-commerce paradigm, shifting from a deep understanding of excellent content and quality products to a model where products can intuitively understand and find the right people (from people searching for products to products finding people).

For consumers, we have validated the sales funnel of AI trial tools in KOL live streaming tools, making personalized AI shopping easier and more interactive. For small and medium-sized streamers and merchants, we have launched an intelligent display tool that can perform intelligent product enhancement and generate sales recommendations, enhancing the e-commerce marketing capabilities of live streamers and merchants.

The era of AI has arrived. As a short video platform content community, we firmly believe that this is not a replacement for human creativity, innovation, or creation. We see it as a tool to help more people in these endeavors. AI will enable ordinary users to break through industry and cultural barriers, allowing them to become globally popular artists and entrepreneurs and to access better creative productivity tools equally. With our enduring commitment to inclusivity, we have been striving to create a community where everyone can showcase their talents and share their lives.

We recognize that this transformation requires time and patience. In the coming years, we plan to increase capital expenditures and new investments, which may affect the speed of our profit margin improvement. Nevertheless, I firmly believe that the expenses we plan today will shape the entire video production and consumption industry over the next decade and formally establish Kuaishou's unique position in the AI eraAlthough we look to the stars and envision the future, we remain grounded and move forward steadily. After outlining our vision, I would like to briefly review our key business highlights and developments in the fourth quarter of 2024 and for the entire year.

III. Q&A

Q: Regarding AI. We noticed that the company's large video model Kling has attracted widespread attention. It was even mentioned at the press conference during the Two Sessions in early March. Can management share the reasons for Kling's success and your plans to maintain the model's competitive edge?

A: First of all, I would like to thank everyone for their attention and support for Kling AI. Kling AI was launched as an internally incubated project at the beginning of last year. It opened for testing on June 6 and has rapidly iterated over 20 versions since its launch. On December 19, we officially released the Kling AI 1.6 model. This significantly improved instruction adherence, visual aesthetics, and dynamic quality, while supporting both standard and high-quality modes.

In terms of instruction adherence, the Kling 1.6 model responds better to motion continuity tasks and camera movement descriptions, and it can better understand textual content while enhancing elements not included in the text. In terms of visual aesthetics, the style is better maintained, the colors are more appealing, the lighting is more reasonable, and the details are more realistic. In terms of dynamic quality, the rationality of motion has improved, and the expressions of characters are more natural. Notably, the internal evaluation of the image-to-video capabilities of the Kling 1.6 model has shown an overall improvement of nearly 200% compared to the 1.5 model. We are confident that the comprehensive effectiveness of Kling's image-to-video functionality is currently ranked first in the world.

The reason Kling AI has achieved and maintained its global leading position is closely related to our focused and resolute overall strategy in the research and development of multimodal large models, especially in quickly establishing Kling AI's industry advantages in organizational collaboration, computational power investment, algorithm enhancement, and data accumulation.

In terms of organization, Kuaishou has always placed great importance on its technology team, and we have a strong talent pool in the fields of algorithms and video understanding, with a high degree of collaboration among teams.

In terms of strategy, we have focused more strategic resources on the research and development of video large models, concentrating resources on key breakthroughs.

In terms of algorithm enhancement, the selection and combination of video generation models are more challenging, requiring more precise technical judgment, systematic selection, and a large number of algorithm innovations. We have always been at the forefront of algorithm innovation, solidifying our ability for further technological breakthroughs.

Finally, in terms of data accumulation, we have also formed significant technical barriers in video data processing and synthesis technology.

Regarding how to continue maintaining the model's advancement in the future, we will continue to adhere to a focused strategy and provide more support for Kling's development. We will maintain our investment in technological research and development for the Kling video large model and leverage the talent aggregation effect brought by Kling, combining computational power, algorithms, and data to strengthen our efforts. We are confident in our ability to maintain the technological leadership of video generation models globally.

Q: What is the breakdown of the long-term AI strategy? How to think about action direction and phased goals?

A: At the beginning of 2023, we launched a new AI strategy and set clear strategic goals to ensure Kuaishou maintains a leading position in AI technology breakthroughs. We believe that as AI technology develops and matures, the market will inevitably experience cycles of highs and lows.

For Kuaishou, the best choice is to rely on AI to upgrade existing businesses, quickly forming a positive cycle of R&D investment and returns, so that Kuaishou's AI can truly have the ability to transcend market cycles. At the same time, we will safeguard the continuous breakthroughs in technological innovation by the large model team through a healthy mechanism, maintaining the leading advantage of large model technology.

In the long term, we hope to upgrade existing businesses through the empowerment of Keling AI while creating a new track for AI-generated video content, further enhancing Kuaishou's leading position as a transformed production and consumption community platform.

Returning to the long-term strategy, we will focus on the following aspects:

Refining the underlying technical capabilities of large language models and multimodal large models, closely following industry development trends, and continuously achieving innovations in algorithm technology. In terms of large language model R&D, we shifted our focus to the Moe model with hundreds of billions of parameters starting in the third quarter of last year, significantly reducing model training and inference costs while maintaining comprehensive model performance, and better adapting to existing business scenarios.

This year, after the open-source release of the DEEPSEEK R1 large model, we are confident that we can closely follow and learn from the most advanced technology models in the market, continuously improving our large model performance.

Continuously upgrading the AI commercialization monetization model represented by Keling, achieving rapid revenue growth, especially striving for Keling AI to become the world's number one video generation AI application in terms of revenue scale.

As a tool for professional creators in the image and video ecosystem, as well as a visual storytelling tool for ordinary creators, we believe that Keling AI's powerful technology and product capabilities can significantly lower the barriers to creation, stimulating the creation, consumption, production, and interaction of AIGC content.

At the same time, we will leverage AI capabilities to empower Kuaishou's existing content and commercial ecosystem. For example, by upgrading content understanding and recommendation, we will enhance the scale and activity of Kuaishou community users. We will also comprehensively open AI capabilities to platform advertisers and e-commerce merchants, providing them with various capabilities including video marketing material generation, digital human live streaming, AIGC, and intelligent prizes, improving operational efficiency. We also look forward to these empowerments for commercial ecosystem partners bringing considerable increments to online marketing services and e-commerce services.

With further breakthroughs in large model AI technology coming online, the barriers and costs of content creation will be drastically reduced, leading to an explosive growth in the quantity and quality of AIGC content production, potentially forming new business models and opening new sustainable growth pathways for the company's development.

Q: The company is empowering existing content and commercial ecosystems through AI technology. Can you elaborate on the application progress and planning? A: By empowering existing businesses with AI technology, we can achieve synchronized development of R&D investment and revenue, enabling Kuaishou AI to truly possess the ability to traverse market cycles. Since launching our new AI strategy, we have steadily made progress step by step, continuously achieving phased results. While continuously optimizing the comprehensive performance of the foundational large model, we are also strengthening AI technology to empower the existing content and commercial ecosystem.

In terms of content understanding and recommendation, Kuaishou leverages the large model's understanding of short video live comments and user interests to recommend more precise content to users, effectively enhancing user duration and activity. The large model changes the traditional content understanding system that requires a large amount of manual labeling, allowing for automatic learning of video content, live content, comment content, commercial content, and user preference attributes, enabling higher-level reasoning, recommendation, and matching.

In search scenarios, we also internally tested the integration of DEEPSEEK R1's large model capabilities in early March this year, optimizing intelligent search effectiveness and user experience, and further enhancing user activity while exploring the commercialization potential of search scenarios.

We expect that based on the large model's understanding of product semantic features and user representations, and leveraging model structures and sequence representations, we can continuously optimize model estimation capabilities and improve advertising effectiveness, thereby driving growth in marketing investments.

For example, regarding the content production costs for marketing clients, we expect to help most clients reduce the production costs of short video marketing materials by 60-70% or even more. At the same time, as the AI technology for virtual digital humans continues to mature, we can also provide merchants with new platforms for virtual human live streaming, helping them achieve breakthroughs in the content e-commerce field.

We anticipate that by 2025, AIGC, short video marketing materials, and virtual digital human live-streaming solutions will continue to contribute additional increments to online marketing revenue.

Q: Regarding the company's investment in AI this year, can management share the scale of this year's investment and its impact on our overall profits?

A: The AI strategy is very important for the company's future development, and we have also seen the tremendous value AI brings to our content and commercial ecosystem. Coupled with the recent acceleration of commercialization of Kuaishou, we are more confident in making firm and long-term investments in AI.

In terms of capital expenditures related to AI, we expect that in 2025, there will still be continued growth compared to 2024, but the scale of growth will be relatively controllable. The main reasons are as follows:

First, we have already conducted a certain amount of advanced training in 2023 and 2024, along with the procurement and reserve of computing power. We plan to continue strategic procurement in 2025, with the core goal still being to ensure sufficient training computing power for Kuaishou to maintain the technological advancement and leading position of the video large model.

Second, this year's capital expenditures related to AI will undergo structural adjustments. We will dynamically adjust and increase expenditures in potential reasoning computing power in line with Kuaishou's commercialization progress, to ensure a sufficient supply of computing power for Kuaishou's reasoning use. Therefore, although we expect a significant increase in reasoning computing power compared to last year, the overall increase in computing power expenditures will be relatively controllableFinally, we will continue to improve the efficiency of existing computing power and server usage through engineering innovation and operational optimization, maximizing the value of existing equipment.

Regarding related expenses, such as the costs of relevant R&D personnel, we expect a certain degree of year-on-year increase, mainly used for attracting and retaining AI technology talent and team building.

Based on the above capital expenditures and expenses, we expect that this year's investment in AI will impact the adjusted profit margin by about 1-2 percentage points. We will also dynamically adjust the scale of investment in AI based on business progress. If the progress of Keling's flexible realization continues to accelerate, we may further increase investment in inference computing power.

However, it is worth mentioning that since Keling has already achieved a breakeven point in gross profit margin for inference-generated videos, the additional investment in inference computing power will have a relatively small impact on profits.

In the long term, as AI continues to release its empowering value for content and the business ecosystem, we believe that incremental investment in AI can bring sustainable performance growth opportunities for the company's development.