From increased spending to the WeChat Agent, the market is waiting for Tencent AI's double strike.
```
Author | Huang Yu
After more than three years of competition in AI large models, the first half of the global AI contest has come to an end.
At this stage, Tencent, which sat steadily on the “throne” during the internet era, has been questioned by the outside world for “falling behind” due to its failure to produce phenomenon-level large models and AI applications.
Unwilling to be marginalized, Tencent quietly began to accelerate since the end of last year, recruiting top talents such as former OpenAI researcher Yao Shunyu, and carrying out a series of AI team reorganizations.
Amid the “lobster” craze sweeping the globe, Tencent is attempting a major AI counterattack through intensive product planning.
With Tencent currently in a critical period of valuation recovery, investors are waiting for the management to present a vision that can lead the market to re-evaluate.
On March 18, Tencent Holdings (00700.HK) released its Q4 and full-year 2025 financial report. In Q4 last year, Tencent’s revenue increased by 13% year-on-year to 194.37 billion yuan, Non-IFRS operating profit rose 17% year-on-year to 69.52 billion yuan. Full-year revenue was 751.77 billion yuan, up 14% year-on-year, while Non-IFRS operating profit was up 18% to 280.66 billion yuan.
This is a financial report that exceeded market expectations. However, investors believe that Tencent’s sustained growth isn’t difficult; the real challenge is whether it can maintain its industry position in the AI era.
At the earnings conference that day, Tencent President Liu Chiping revealed that Tencent’s 2025 capital expenditure of 79.2 billion yuan was indeed lower than the intended target, mainly due to GPU supply. “In 2026, we hope to significantly increase capital expenditure, intensify flexible leasing, and boost overall computing power acquisition in both areas.”
He pointed out that Tencent currently has a relatively clear strategic plan for AI. For the next two or three quarters, it is important to continue doing very important and correct things. When these things are accomplished, quantifiable progress will naturally emerge. Reportedly, Hunyuan 3.0 is now in internal testing and will gradually open to the public in early April.
Spending Difficulty in Buying Cards
After bold reforms over the previous two years, Tencent resumed double-digit revenue growth in Q4 2024, and in 2025 is expected to maintain this growth level.
Tencent management stated that the growth in 2025 was thanks to AI tech enhancing advertising targeting and player interaction in games, as well as accelerated revenue growth from cloud business and scaled profitability.
As Tencent’s core business, games are the key driver behind steady revenue growth.
Financial reports show that Tencent’s game business revenue for 2025 reached 241.6 billion yuan, up 22% year-on-year, accounting for 32.1% of total revenue. Domestic game market annual revenue grew 18% year-on-year to 164.2 billion yuan. International game market revenue hit 77.4 billion yuan, up 33% year-on-year—a new record.
In the second half of the AI “application competition,” Tencent Games became an important battleground. Tencent’s financial report specifically mentioned that game AI is now effectively accelerating content creation, improving user experience, and boosting marketing effectiveness.
Despite its strong moat businesses, Tencent is very cautious about expansion, and has indicated that the only area now truly worth early investment is AI.
Compared to the doubling of capital expenditure in 2024 to 76.7 billion yuan, Tencent’s 2025 capital expenditure is about 79.2 billion yuan—growth is smooth and presents a sharp contrast to Alibaba and ByteDance’s announcements of multi-hundred-billion-level infrastructure investments.
Liu Chiping explained that last year, because Tencent couldn’t buy “cards,” it spent more on share buybacks. If this year it can buy “cards,” it will reduce buybacks and increase capital expenditure.
Meanwhile, Liu Chiping pointed out that last year Tencent invested 18 billion yuan in AI new products, and this year it plans to at least double that. “Our stable and sustained growth in core businesses enables increased AI investment.”
Tencent was the absolute winner in the mobile internet era, solidly holding the title of “King of Tech Stocks in China” thanks to its moat businesses of social and gaming.
However, because of the perceived “falling behind” in this round of AI competition, Tencent’s stock price has been continuously falling since October last year, closing at about 550.5 HKD per share on March 18, with its market cap dropping to 5 trillion HKD.
AI Counterattack: Open and Hidden Cards
AI is now what the market cares about most for Tencent.
In 2025, Tencent adjusted its organizational structure and added AI reconstruction in multiple business scenarios.
Whether it’s merging AI product lines like Yuanbao into CSIG (Cloud and Smart Industries Group) to expedite commercialization, or deeply integrating large model capabilities into WeChat Search, Tencent’s strategic intent is clear: not to pursue an independent AI super app, but to make AI part of the Tencent ecosystem’s infrastructure.
However, this typical defensive counterattack is showing its limitations amid today’s wild growth on the AI battlefield.
Especially with the Agent era accelerating, there are worrying market concerns about Tencent’s development and investment in AI. Facing the double-pronged challenge from Alibaba and ByteDance, there is indeed a risk of falling behind.
The model side, which concerns basic capabilities, remains highly valued by the market.
Liu Chiping said that in recent months, Tencent has intensively upgraded the Hunyuan large model team’s organization and workflows, rebuilt the pre-training and reinforcement learning infrastructure, and further improved data quality.
Through a series of adjustments, Tencent hopes to build a smarter model and accelerate the overall iteration speed.
In Tencent’s AI efforts, applications have attracted more attention.
This Spring Festival, Yuanbao, Tencent’s pioneering AI-native application, spent 1 billion yuan to take the lead in the red envelope battle. Daily active users once exceeded 50 million, but dropped off noticeably after the holiday.
Liu Chiping believes Yuanbao has already amassed a fairly large user base, and Spring Festival promotion met company expectations.
Next, Yuanbao aims to grow users and improve retention through continuous experience iteration, including dialogue and search capabilities. “It’s clear that as these abilities improve, user activity and retention continue to rise.”
Meanwhile, Liu Chiping admitted that many users, whose Yuanbao experience wasn’t ideal before, still chose Yuanbao due to its tight integration with Tencent’s ecosystem. Yuanbao will further strengthen synergy with Tencent’s ecosystem.
From the Spring Festival push for Yuanbao, to today’s comprehensive OpenClaw layout, these are Tencent’s current “open cards”.
Starting March 2026, Tencent launched a matrix product plan around lobsters, calling itself the “Lobster Special Forces.” This product matrix covers both novice users and hardcore developers as well as large enterprises—aiming to break down usage barriers to “lobster.”
Pony Ma pointed out at the earnings conference that “lobster” is a decentralized entry point, helping Tencent leverage its resource advantages, unite various forces, and inspire the development of WeChat Agent.
Of course, Tencent’s biggest AI trump card is still WeChat.
Last year, Tencent management repeatedly mentioned it would create an AI Agent embedded into and operating on WeChat’s unique ecosystem.
Recently, multiple media reported that Tencent is secretly developing an AI Agent for WeChat, likely to start gray-box testing mid-year and possibly release to all users in Q3.
Liu Chiping confirmed this, saying the next step will be to develop an AI Agent inside WeChat.
This Agent will be closely connected to users via WeChat, offering a very rich ecosystem: mini-programs, content, transactions, social, and payments, so this agent can not only chat with users but also do many real tasks.
“But because WeChat has a massive user base, there are privacy and security issues, requiring further improvement in model capabilities to solve.” Liu Chiping said the WeChat AI Agent is moving forward, but there is no clear timetable yet.
Perhaps, after years of dormancy, as the focus of the AI industry shifts from “model parameter competition” to “landing capability competition,” Tencent has hit the best batting point and is acting swiftly.
Analysts from China Merchants Securities believe the market underestimates Tencent’s historical development and competitive barriers, and the appearance of “Little Lobster” may help Tencent find a “WeChat for the AI era.”
With WeChat’s 1.4 billion monthly active users and Tencent’s abundant cash flow, it’s far from the moment of elimination. Whether this full-scale AI counterattack comes at the right time and whether its “lobster” and WeChat trump cards can solidify Tencent’s core internet position remain to be tested by the market.
Risk Warning and DisclaimerThe market involves risks, investment needs caution. This article does not constitute personal investment advice, nor does it consider the special investment goals, financial situation, or needs of individual users. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their specific circumstances. Invest accordingly, at your own risk. ```