<img src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c5de3ab6-f9b7-4b6e-b6dc-8205fcfb32a6.jpeg?x-oss-process=image/auto-orient,1/interlace,1/resize,w_1440,h_1440/quality,q_95/format,jpg" alt="" original-src="https://imageproxy.pbkrs.com/https://wpimg-wscn.awtmt.com/c5de3ab6-f9b7-4b6e-b6dc-8205fcfb32a6.jpeg"/> Author | Huang Yu
Editor | Zhou Zhiyu
In the wave of large model applications, Tencent has welcomed the opportunity to reach new heights in the AI era.
Based on Tencent's &#34;Blitzkrieg&#34; approach in the AI application battlefield over the past month, the capital market expects Tencent to publicly declare its determination to fully embrace the AI era, and the latest financial report undoubtedly serves as an important observation window.
On March 19, Tencent released its performance report for the fourth quarter and the whole year of 2024, showing that due to the AI strategy entering a reinvestment period, its total capital expenditure for the previous year exceeded 76.7 billion yuan, a year-on-year increase of 221%, setting a historical high, accounting for about 12% of the annual revenue of 660.3 billion yuan.
During Tencent's earnings conference call on the same day, the AI strategy undoubtedly became the focus.
Tencent President Liu Chiping revealed that the increase in capital expenditure in the fourth quarter of last year was very significant because Tencent purchased more GPUs in that quarter to meet inference demands. &#34;We plan to further increase capital expenditure in 2025 and expect capital expenditure to account for a low double-digit percentage of revenue.&#34;
In other words, Tencent's capital expenditure this year is likely to reach the hundred billion level.
For Tencent, focusing on AI not only helps it seize the next super traffic entrance but also serves as a multiplier for Tencent's business development. According to Goldman Sachs, in the Chinese artificial intelligence industry, &#34;AI infrastructure looks to Alibaba, while AI applications look to Tencent,&#34; but it is clear that these two Chinese internet giants are now both releasing the momentum they desire.
However, to take the lead in this global AI arms race, &#34;financial capability&#34; is crucial. After the transformation of &#34;cost reduction and efficiency enhancement, convergence and focus&#34; in the past two years, Tencent's performance last year has already achieved a rebound from the bottom, with total revenue in the fourth quarter recovering to double-digit growth, which is the confidence behind Tencent's strong investment in AI.
Standing at the critical juncture of the explosion of AI applications and the revaluation of Chinese assets, Tencent, with its large user base and rich AI application scenarios, is expected to deepen the moat of its traditional business while seizing new opportunities brought by AI, leading its market value back to 7 trillion yuan or even higher.
Reshaping AI Strategy
From cautious implementation last year to an all-out offensive now, Tencent is reshaping its AI strategy.
Tencent Chairman and CEO Ma Huateng stated: &#34;Months ago, we reorganized the AI team to focus on rapid product innovation and deep model research and development, increased capital expenditure related to AI, and intensified our research and marketing efforts for native AI products.&#34;
Tencent believes that these increased investments will bring sustained returns by enhancing the efficiency of its advertising business and extending the lifecycle of its games, and will create longer-term value as its personal AI applications accelerate in popularity and more enterprises adopt Tencent's AI services.
On the same day the financial report was released, at the 2025 Tencent Global Digital Ecosystem Conference in Shanghai, Tencent's system elaborated on its AI strategic thinking
According to reports, Tencent will adopt a multi-model strategy of &#34;firmly investing in self-developed models + openly embracing advanced open-source models,&#34; comprehensively laying out its capabilities at the foundational, model, and application levels to provide users with a rich array of AI application products, aiding in industrial innovation and growth.
This means that while actively integrating the DeepSeek large model, Tencent will continue to increase its investment in the research and development of self-developed large model technology.
Tencent's self-developed Hunyuan large model is the first in the industry to adopt the MoE architecture, with flagship model parameters reaching the trillion-level, and it has already been integrated into over 700 products and business scenarios at Tencent.
Integrating the DeepSeek large model also pushes the Hunyuan large model to upgrade more quickly. Recently, Tencent's self-developed reasoning model Thinker (T1), which excels in complex tasks and deep reasoning, has been launched on Tencent Yuanbao. Additionally, Tencent has released a new generation of fast-thinking model Hunyuan Turbo S, which emphasizes faster task processing capabilities, and has launched and open-sourced several models such as &#34;Image-to-Video Model&#34; and &#34;3D Generation Model.&#34;
Furthermore, to further enhance organizational collaboration efficiency, Tencent has adjusted more AI products and applications, including Tencent Yuanbao, QQ Browser, Sogou Input Method, and ima, to CSIG (Cloud and Smart Industry Group).
This move undoubtedly allows the AI product matrix, including Yuanbao, to focus more on commercialization, while TEG (Technology Engineering Group), responsible for the development of Tencent's Hunyuan large model, focuses on improving model performance.
Tang Daosheng, Senior Executive Vice President of Tencent Group and CEO of the Cloud and Smart Industry Group, stated that with the open-sourcing of DeepSeek and breakthroughs in deep thinking, AI large models are crossing the threshold of industrialization and reaching a new node of widespread application. The industry has shifted from being model training-driven to being more application and Agent-driven today.
Turning Defense into Offense
In the wave of AI driven by ChatGPT, Tencent has seen products that have not turned the tide despite efforts over the past two years. After DeepSeek's explosive popularity during this year's Spring Festival, AI applications have reached a critical juncture, and the AI model equity brought by DeepSeek has reshuffled the entire AI industry, seemingly returning the competition of AIGC products to the same starting line.
Tencent has keenly captured this point and has changed its usual approach, actively shifting from defense to offense in its AI strategy.
In February, Tencent fully launched the &#34;Lightning War&#34; plan, becoming one of the most proactive companies among major firms to embrace DeepSeek. Not only has it integrated the DeepSeek model into multiple products, but it has also become the first leading company to introduce DeepSeek into its main C-end products.
As of now, products such as WeChat, Yuanbao, QQ Browser, QQ Music, ima (Tencent Smart Workbench), Tencent Docs, and Tencent Maps have all integrated DeepSeek and simultaneously support the self-developed Hunyuan large model.
In this opening battle of AI applications, Tencent has two tactics that have attracted significant industry attention: one is the integration of WeChat search with the DeepSeek large model; the other is the concentration of resources to enhance Yuanbao, with Tencent disclosing that it has updated 30 versions in 35 days, demonstrating significant effort
Since the release and open-source of the DeepSeek-R1 model nearly two months ago, many internet products have integrated it, essentially aiming to capture market share by leveraging this recognized capability. However, none of these integrations compare to WeChat, which has attracted attention with its super traffic pool of over 1.3 billion monthly active users.
On February 15th, Wall Street News confirmed from Tencent that WeChat Search officially began a gray-scale test integrating DeepSeek-R1. User enthusiasm for the DeepSeek model far exceeded the WeChat team's expectations, resulting in significant computational consumption.
Unable to meet the AI search demand of all users in the short term, three days later, WeChat had no choice but to redirect massive traffic to its &#34;brother app&#34; Yuanbao. Users who have not yet been included in the gray-scale testing of AI search will see an opportunity to &#34;download Yuanbao (experience DeepSeek-R1)&#34; when they click on the search bar at the top of the WeChat dialogue box.
As the core entry point of the Tencent ecosystem, WeChat's traffic allocation strategy has always been cautious, focusing more on optimizing its own functions rather than actively directing traffic to other applications. This action of redirecting downloads of Yuanbao through the WeChat search entry reflects Tencent's significant shift in its AI strategy, marking Yuanbao's more proactive traffic generation and Tencent's ambition in C-end AI scenarios.
Yuanbao Pioneer
Tencent's strategy of using a &#34;Mixed Yuan + DeepSeek&#34; dual-engine approach has emerged, and among the AI product matrix it has created, Yuanbao has already positioned itself at the core of Tencent's AI strategy, serving as the pioneer in capturing the next generation of super traffic entry points.
Tencent hopes Yuanbao can become an &#34;all-purpose AI assistant&#34; in everyone's pocket. From the perspective of AI search capabilities, Yuanbao also acts as a connector, linking all of Tencent's internal ecosystems, including WeChat Official Accounts, WeChat Work, and more.
Shifting from last year's low-key approach, Tencent is both iterating Yuanbao through late-night efforts and intensively boosting traffic. After integrating DeepSeek, Yuanbao's advertisements quickly spread across all platforms within the Tencent ecosystem, while external channels such as Bilibili, Zhihu, and Xiaohongshu also became important battlegrounds.
According to DataEye data, starting from February 18th, the advertising volume for Tencent Yuanbao surged, rising from about 2,000 daily materials to over 17,000 by the end of the month, jumping from the bottom position among Doubao and Kimi to the top. In contrast, Kimi's advertising efforts significantly declined, dropping from about 20,000 daily materials to nearly stagnation.
It is evident that Yuanbao aims to leverage the popularity of DeepSeek to launch a blitzkrieg. In addition to substantial advertising investments, it is also rapidly iterating its product features to capture the market, including launching the deep thinking model &#34;Mixed Yuan T1,&#34; the fast thinking large model Mixed Yuan Turbo S, a desktop version, and supporting the upload and export of Tencent documents, with updates occurring almost every three days.
As a result of these operations, on March 3rd, the Tencent Yuanbao APP rose to first place in the free app download rankings in the Chinese Apple App Store, surpassing DeepSeek, while Doubao ranked fourth. It is noteworthy that on February 13th, Yuanbao was outside the top 100 on this list
However, it is worth mentioning that download volume can only reflect short-term promotional effects, while retention rate and active user count can directly reflect the long-term value of the product and user recognition.
Liu Chiping revealed that from February to March, the daily active users (DAU) of Yuanbao increased by 20 times, becoming the third-ranked AI-native mobile application in China by DAU.
According to Xsignal, from February 17 to the end of February, Yuanbao's DAU showed a linear upward trend. On February 17, Yuanbao's DAU was approximately in the million range, nearly 20 million less than Doubao; by the 25th, Yuanbao's DAU had already surpassed 10 million.
However, within a month of the release of DeepSeek R1, DeepSeek's DAU had already exceeded 100 million at its peak, and remained stable at around 80 million on average daily. If Tencent's Yuanbao aims to become the number one AI super application in the future, DeepSeek will be a stronger competitor compared to Doubao.
Liu Chiping pointed out that as a startup, Tencent has indeed provided significant investment flow for Yuanbao, and is closely monitoring the retention rate issue during this process, which is currently quite good.
&#34;So we (investment flow) are still continuing, but in the future, it will definitely not rely solely on investment flow to acquire users. We will add various functions to Yuanbao to make it a smarter AI assistant and create organic connections between Yuanbao and our existing products.&#34;
Reassessing the Future
Tencent's past investments in AI have already been reflected in its financial reports.
Tencent's Chairman and CEO Ma Huateng stated, &#34;Benefiting from the upgrade of the AI-enabled advertising platform, increased user engagement on video accounts, and growth in evergreen games, we achieved double-digit revenue growth in the fourth quarter of 2024 and continued to improve operational efficiency.&#34;
Although DeepSeek has become popular, compared to national-level applications like WeChat, Douyin, and Weibo, which have hundreds of millions of users, it is still far from being a super application. However, Tencent is expected to quickly realize value monetization due to its large user base and rich scenario advantages, which is one of the biggest expectations of the capital market for Tencent today.
Capital expenditure is the most indicative measure of Tencent's commitment to investing in AI and is currently the focus of investors. Tencent's recent decision to further increase capital expenditure will undoubtedly help boost market confidence.
In comparison, not long ago, Alibaba announced that it would invest over 380 billion yuan in the next three years to build cloud and AI hardware infrastructure, exceeding the total of the past decade.
Tencent's budget this year is also aiming for the hundred billion level, and its investment intensity does not seem to be weaker than Alibaba's.
Regarding specific directions for capital expenditure, Liu Chiping revealed that in terms of R&amp;D, Tencent will continue to invest in self-developed models and accelerate AI application development across various business groups. At the same time, Tencent will also invest in marketing to enhance user awareness and adoption rates of new AI products, such as Yuanbao.
&#34;We believe these investments will bring good economic returns, but we also have the ability and willingness to continue providing capital returns to shareholders. We plan to repurchase at least 80 billion Hong Kong dollars of stock by 2025.&#34;
With the support of real money, Tencent's AI application landscape is fully opening up from B-end to C-end, and its AI strategy is expected to become its second growth curve.
However, some investors have pointed out that while AI technology can bring intelligent upgrades to Tencent's existing products, especially the huge commercialization potential of the WeChat ecosystem, whether it can achieve new growth through AI depends on technological breakthroughs and market applications.
In this global AI arms race, Tencent has obvious advantages.
Analysts from Huazhang Securities pointed out that for Tencent, the deep integration of its own products with DeepSeek reflects Tencent's &#34;internal and external&#34; technological integration capability. By embedding AI capabilities into core scenarios such as social networking, content, and finance, Tencent is building an intelligent service network that responds to user needs in real-time and circulates data value in a closed loop.
&#34;When the technological dividend is exponentially released through super entrances like WeChat, QQ, and Tencent Cloud, the resulting network effects may create a 'data-scenario-technology' flywheel that is difficult for other companies to replicate,&#34; said the analysts from Huazhang Securities.
Goldman Sachs believes that Tencent is a major beneficiary of AI applications, thanks to its e-commerce infrastructure for small shops, mini-program ecosystem, and WeChat Pay functionality.
HuaChuang Securities is optimistic about Tencent's top-down push for AI transformation of products by 2025, which will bring user experience innovation and open up incremental commercialization space, driving valuation increases, giving Tencent a target price of HKD 544-604.45 for 2025.
Since the beginning of the year, Hong Kong tech stocks have performed brilliantly, with Tencent's stock price continuously climbing, breaking the HKD 500 mark, and rising over 29% year-to-date. As of the close on March 19, Tencent's stock price was about HKD 540 per share, with a total market capitalization of approximately HKD 4.96 trillion.
In the mobile internet era, Tencent is the absolute winner, firmly holding the position of &#34;king of tech stocks&#34; in China with its two major moat businesses: social networking and gaming. At the beginning of 2021, its stock price once exceeded HKD 700 per share, with a market capitalization of over HKD 7 trillion.
Now, as the AI era rushes in, whether Tencent can seize the traffic foundation brought by the mobile internet era and capture the next generation of super traffic entrances will be key to whether it can return to its peak market value

Alibaba  earnings 

TCEHY

TCTZF

00700

Competing for dominance in the AI era

- Tencent is seizing opportunities in the AI era, as indicated by its recent financial report showing a significant increase in capital expenditure, reaching over 76.7 billion yuan, a 221% year-on-year rise.  
- The company plans to further increase its capital spending in 2025, focusing on AI applications to enhance its business growth and market position.  
- With a robust user base and strategic investments in AI, Tencent aims to reclaim its market value, potentially exceeding 7 trillion yuan, amidst a competitive landscape in the AI sector.

Tencent is set to invest 100 billion yuan in AI