Written by | Hao Xin, Huang Mengyu; Edited by | Hao Xin
On November 18, Tencent released its latest quarterly earnings report.
In Q3 2026, Tencent achieved revenue of CNY 207.336 billion, an increase of 8% year-on-year. Under non-IFRS standards, profit attributable to equity holders was CNY 66.519 billion, a decrease of 6% year-on-year. The three main business lines—gaming, advertising, and fintech & business services—continued to show steady growth.
While the core stronghold remains stable, Tencent has found itself in unprecedented confusion regarding its next steps.
At the Tencent shareholders' meeting in May this year, when discussing whether Tencent's AI efforts were lagging, Pony Ma stated: "A year ago, we thought we had boarded the ship, but later discovered it was leaking. Now we feel we are on board, but cannot yet sit down comfortably. We still hope the ship can accelerate."
On November 11, during the LinkTime session at Tencent's executive office, Pony Ma once again discussed AI.
"The waves have already crashed over us. Some core businesses appear solid, but AI is advancing rapidly. Distant tides are swallowing individual hills one by one; you cannot just sit and watch. Land warfare is becoming naval warfare, and infantry must become sailors. Everyone needs to transport supplies, shift skills, and quickly board the ship to fight," he said.
For Tencent, the critical issue is not whether it has AI, but whether AI can form a new closed loop.
Tencent has always possessed strong C-end (consumer-facing) DNA, meaning its AI path naturally bears the mark of "C to B"—leveraging C-end product habits and entry-point momentum to unlock B-end scenarios and revenue.
According to Nomura Securities data, in April 2026, among the Top 50 apps covering approximately 93% of China's mobile internet usage time, ByteDance's share of usage time rose from 32.3% in the same period last year to 38.3%, while Tencent's share stood at 30.8%. A year prior, ByteDance and Tencent were nearly equal at 32.3% and 32.4%, respectively.
After leveraging WeChat's massive user base and driving 114 million monthly active users through a CNY 1 billion red packet campaign, Yuanbao (Tencent's AI app) dropped to 37.33 million monthly active users in September this year, no longer in the same league as Doubao's 419 million monthly active users.
In the AI era, the situation Tencent faces appears more severe than ever before. As C-end growth slows, where will Tencent find the closed loop for the AI era?
The "Thick" and "Thin" of the C-End Matrix
Multiple financial metrics indicate that the walls of Tencent's C-end business remain "thick." The core foundation has not only withstood impact but also seen profit growth following the AI transformation of its products.
On the revenue side, Tencent's latest earnings report highlighted three major growth drivers.
The report shows that Value-Added Services (VAS) revenue grew 3% year-on-year to CNY 98.3 billion. Domestic market game revenue reached CNY 47.6 billion, up 11% year-on-year, driven by increased revenue from games such as Delta Force, VALORANT Mobile, and Honor of Kings.
Marketing services revenue increased 19% year-on-year to CNY 43.0 billion. Tencent leveraged integrated closed-loop marketing capabilities within the WeChat ecosystem to support short dramas, WeChat Stores, and other integrated WeChat services, placing ads across WeChat and other Tencent-managed ad inventory.
In fintech and business services, Q3 2026 revenue rose 9% year-on-year to CNY 63.7 billion. Fintech service revenue growth stemmed from increased income from consumer loans and wealth management services. Business services revenue growth was primarily driven by a 35% year-on-year increase in cloud service revenue.
However, financial data reveals that AI-native applications exhibit relative weakness. The essence lies in the fact that Tencent's C-end advantage is a distribution advantage, not an AI-native advantage.
Leveraging technologies such as AI ad recommendation models and AIM+ upgrades, Tencent's advertising business has achieved end-to-end delivery capabilities—from targeting and bidding to creative generation and execution. In gaming, the integration of AI with gameplay continues to advance; for example, in Game for Peace (PUBG Mobile), AI NPCs are used to enhance player experiences.
From the perspective of Tencent's AI implementation in advertising and gaming, the company has moved beyond technical pilots into operational validation. However, this path remains logic-based on enhancing existing businesses with AI, rather than creating new entry points or incremental growth through AI-native applications.
Consequently, the traditional C-end traffic moat upon which Tencent relies is being continuously eroded by the wave of AI-native applications. Its original distribution advantage is no longer a guaranteed "amulet" for victory.
In QuestMobile's September ranking of AI-native applications, the top three apps by monthly active users were: Doubao (419 million), Qwen (158 million), and DeepSeek (123 million). None of Tencent's AI-native products made it into the first tier.
This "thinness" in the AI-native sector keeps Tencent in a perpetual catch-up position.
Before WorkBuddy emerged, Tencent lacked a benchmark product capable of establishing a foothold in the C-end AI track. Therefore, Tencent not only used revenues from gaming and entertainment sectors to sustain AI development but also leveraged its C-end ecosystem advantages to provide entry-point scenarios for its own AI-native applications.
Overall, however, the results have been less than satisfactory. Over the past two years, several AI products under Tencent—including QClaw, Tencent Yuanqi, Tencent Translator, and Zhiying—have announced shutdowns or integrations. Some capabilities have migrated into mainline products like Yuanbao and WorkBuddy, ceasing operations as independent C-end AI applications.
Facts prove that incubating AI-native applications with self-growth capabilities using Tencent's familiar C-end traffic funneling tactics is highly difficult. The social relationship chain, representing the first layer of value in its C-end DNA, cannot guarantee long-term product stickiness.
In contrast, WorkBuddy, part of the Cloud and Smart Industries Group (CSIG), successfully ran through a closed loop via free exploration in edge projects, gradually growing into a strategic AI product for Tencent.
Tencent's Q3 financial data indicates that, measured by large language model token consumption, WorkBuddy has become the most-used AI product in China, with daily active users exceeding 10 million. These tens of millions of users collectively consume over 100 trillion tokens daily.
The second layer of value in Tencent's C-end DNA lies not in traffic entry points, but in how to synergize years of accumulated users and products to extend and implement solutions in B-end productivity scenarios.
This is precisely why WorkBuddy has been pushed to the forefront, carrying Tencent's expectations for an AI breakthrough.
WorkBuddy: The Only Validated Sample in the "C to B" Closed Loop
"We believe WorkBuddy has developed into the largest AI application in China. Driven by WorkBuddy, along with our allocation of more computing power for Model-as-a-Service and GPU rentals, the growth rate of cloud service revenue has further improved," Tencent stated.
In the AI era, Tencent has once again achieved a C-to-B closed loop, and this time, the sample is WorkBuddy.
WorkBuddy's predecessor was CodeBuddy, an internal efficiency tool aimed at developers. Early on, AI Coding was not a hot track. To validate product value, the CodeBuddy team directly entered enterprise customer sites, participating in meetings, deployments, feedback collection, and product modifications.
This meant that the early CodeBuddy team accumulated not just model invocation capabilities, but also the working capability to integrate models into real-world scenarios.
When WorkBuddy lowered the barrier to entry and shifted toward office scenarios for non-technical users, it became a work partner for everyone, capable of autonomously understanding goals, breaking them down, running code, calling tools, and ultimately generating deliverables.
Simultaneously, within Tencent's internal ecosystem, WorkBuddy connected with C-end products such as Tencent Docs, QQ Mail, and Tencent News, aggregating the application momentum of the Tencent ecosystem. QuestMobile data shows that WorkBuddy ranked first in the AI office category, reaching 32.35 million monthly active users in September—more than triple that of the second-ranked app.
In October this year, WorkBuddy deeply integrated with Enterprise WeChat, achieving unified account interoperability and further connecting with Tencent's office ecosystem.
In this process, WorkBuddy serves as the task execution layer, utilizing Enterprise WeChat as a customer acquisition entry point to directly reach 14 million enterprises. It successfully transplanted C-end product habits into B-end scenarios. The conversational interaction habits employees cultivated on the C-end can be directly translated into enterprise workflows. Enterprises can pay directly for delivered results, rather than just for models and computing power.
In the second half of this year, WorkBuddy secured an increasing number of enterprise orders, spanning transactions in insurance, banking, hospitals, fast-moving consumer goods (FMCG), construction, media, and other fields.
Official WorkBuddy data indicates that its enterprise version has landed in over 50 industries, covering 50% of China's Fortune 500 companies, with the number of developers and collaborators exceeding 50,000.
From C-end user habits to B-end scenario payments, WorkBuddy has become the sole empirical proof of a closed loop for Tencent in the field of AI-native applications.
However, Tencent does not position WorkBuddy as a replacement for existing tools and software, but rather as a platform capable of hosting all tools, potentially serving as an operating system for the Agent era.
Buddy One, a cross-application AI collaboration system launched on October 29, uses WorkBuddy as a workspace to internally connect Tencent Docs, Tencent Meeting, Tencent Drive, and ima, while externally linking to existing enterprise systems and third-party Agents.
Under this strategic framework, Tencent has preliminarily achieved an AI closed loop on the B-end, possessing a fulcrum that does not rely on C-end traffic scale.
Where is the Next WorkBuddy?
The AI window of opportunity waits for no one, and the gap in the AI office domain is narrowing.
According to Analysys data, in September 2026, among desktop-side AI-native office agent platforms in China, Qwen Office recorded 46.55 million monthly visits, and Doubao Work recorded 30.39 million monthly visits. Alibaba and ByteDance are also racing to catch up through organizational restructuring.
Meanwhile, Tencent's proud C-end stronghold is facing erosion from ByteDance.There are two paths here: first, time substitution, where Douyin and Hongguo Short Drama continue to capture users' leisure time.
QuestMobile data shows that in May 2026, Hongguo reached 356 million monthly active users; in July, its daily active users hit 168 million, surpassing the combined DAU of four major long-video platforms: Tencent Video, iQIYI, Youku, and Mango TV.
The second path is habit migration, where AI applications like Doubao are establishing new interaction inertia. In QuestMobile's September 2026 statistics, Doubao saw a monthly per-user usage frequency of 72.0 times and a monthly per-user duration of 132.5 minutes.
These two paths point to the same result. Although Tencent's C-end entry points remain, the mindset of opening Tencent products to complete tasks is being diluted by opening ByteDance products to kill time or by diversion to AI services.
The leverage point of "C to B" is precisely this mindset. Once the fulcrum begins to loosen, the efficacy of the lever declines accordingly.
Pony Ma described the transformation, noting that some people act like "infantry," guarding core businesses like gaming and social networking while improving efficiency through AI; others act like "navies," exploring AI Native businesses and new growth spaces.
This metaphor itself explains the problem: if the C-end foundation were sufficiently stable, there would be no need to convert half the workforce from "infantry" to "navy." The premise of shifting from "land warfare" to "naval warfare" is that advantages on land are being bypassed by distant waves. The urgency of organizational adjustment stems exactly from the vulnerability of C-end DNA under AI disruption.
It is not that WeChat will disappear, but that the value weight of the "relationship chain entry point" represented by WeChat may no longer be sufficient to support Tencent in leveraging the next decade of B-end business via the C-to-B model in the AI era.
Tencent, previously accustomed to making money while "lying down," has begun taking decisive action against its organization. Tencent stated, "Last time was passive cost-cutting; this time is proactive efficiency enhancement." At this moment, what Tencent lacks most is an AI transmission mechanism at the organizational level to switch from C to B.
Tencent believes that AI development at the application level is just beginning, "similar to the early industrial age when electricity was just invented for lighting, but various appliances and industrial uses of electricity had not yet appeared." Future larger-scale intelligent applications will occur in production, integrating with thousands of industries, presenting many opportunities worth investing in.
Following this logic, Tencent has started replicating the Agent engineering capabilities accumulated by WorkBuddy, the reach capabilities of Enterprise WeChat, and Tencent's industry data into vertical scenarios.
Officially, two months after the launch of the "Buddy One" open platform, Buddy applications have covered 55 professional scenarios and over 30 vertical domains, including finance, law, healthcare, education, public welfare, design, and SaaS.
The next "WorkBuddy" may not be called "WorkBuddy," but it will certainly carry the same DNA. Starting from an existing C-end entry point or habit, it will grow into an Agent that B-end users are willing to pay for.
WorkBuddy is the first answer, but Tencent still needs a series of answers.