Fu Weigang/Wen

Over the past few months, Tencent's stock price has been difficult to understand.

From a high of HKD 639 at the beginning of 2026, it fell to around HKD 472 by the end of April, a decline of over 25%. During this period, Tencent did not have any obvious bad news: game revenue remained stable, advertising business recovered, and WeChat's monthly active users remained stable at over 1 billion. The financial reports kept shining, but the stock price continued to decline. The explanations given by analysts are varied, but there is always a feeling of scratching the surface: they are all right, but they haven't touched on the root.

Until I saw a screenshot of my friend's phone and suddenly had a new perspective on thinking.

The screenshot shows the battery usage record of the iPhone, with data from two apps being combined on the same day: WeChat, screen active for 2 hours and 1 minute, consuming 50% of power; Claude? The screen is active for 2 hours and 11 minutes, consumes 26% of power, and has almost the same usage time. My friend added a sentence: 'I spend more time using Claude now than I do on WeChat.'. ?

This screenshot certainly does not illustrate any macro issues, but it provides an observation perspective: in a specific user's day, an AI application is competing with WeChat. Not Tiktok, not Weibo, not any product that has challenged WeChat in the past 15 years, but a completely different type of application.

The pricing in the capital market is precisely due to this structural instability. The problem with Tencent is not how much money it makes today, but that users' time is flowing in a direction that WeChat has never truly defended. To understand what this means, we need to start from a more fundamental proposition: what are the truly scarce resources in an era of information overload?

Attention as a scarce resource

In 1971, Herbert Simon wrote in a frequently cited but often misunderstood paper that the abundance of information inevitably leads to a lack of attention. His logic is concise and powerful: information consumes the attention of the receiver, therefore, the richer the information, the scarcer the attention. This insight was put forward before the birth of the Internet, but its full historical weight was really found in the era of mobile Internet.

In 1997, Michael Goldhaber further proposed the concept of "attention economy" based on this: in the information age, attention itself is currency. Whoever can capture users' attention can monetize it - whether through advertising, subscriptions, or data. The business models of Internet platforms are essentially monetization of attention.

This framework provides the first key to understanding the competitive history of WeChat. Over the past 15 years, every competition WeChat has faced has been a struggle over product functionality on the surface, but in essence, it has been a redistribution of attention share: what operators have lost is the time and money users spend on text messages; Weibo loses the idle time that users spend in public squares; Tiktok competes for the fragmented sensory time of users before waiting, commuting and sleeping.

But attention is not homogeneous. Herbert Simon himself is a cognitive scientist who is well aware of the hierarchical nature of human cognitive resources - shallow perceptual attention and deep reflective attention, which are fundamentally different in quality. This distinction has become crucial in the era of AI, and we will return to it later.

From SMS Killer to National Operating System

To understand the competitive history of WeChat, one must first understand how fierce its emergence was back then.

When WeChat was launched in 2011, the Chinese mobile communication market was in the golden age of operators. SMS is the backbone of daily communication, MMS is the prototype of social media, and Feixin is a failed attempt by China Mobile to build a closed ecosystem. WeChat, a minimalist instant messaging tool, has disrupted this order. It quickly gained popularity based on QQ's user base, transforming a communication tool into a social platform with the three pronged approach of "shake", "nearby people", and "circle of friends".

This downgrade strike on operators took less than three years.

The first true competitors that WeChat faced afterwards were similar instant messaging products. Momo was launched in 2011, focusing on socializing with strangers, and was once quite aggressive in user growth. Mi Chat was a contemporaneous product of Xiaomi, and Lei Jun personally supported it back then, which was seen as WeChat's most dangerous early competitor. But neither Momo nor Mi Chat has been able to break through the network effect barriers that WeChat has already formed.

Network effect is an extremely cruel competitive mechanism. Economists Carl Shapiro and Hal Varian systematically explained the operational logic of this mechanism in their 1999 book "The Rules of Information": when the value of a product increases nonlinearly with the number of users, the user base formed by the first mover itself constitutes an entry barrier.

More importantly, communication products face "bilateral lock-in": not only do you need to persuade users to switch platforms, but you also need to persuade all their contacts at the same time. This coordination cost is almost equivalent to an impossible task in practice. Mi Chat ultimately ceased service in 2019; Momo found her own living space, but never threatened the core position of WeChat.

The second wave of challenges comes from content platforms. The rise of Weibo once led the public to believe that the main battlefield of social media was in public squares rather than private message spaces. WeChat Moments was launched in 2012, which was largely a strategic response to Weibo - using a "semi closed" circle of friends to counter Weibo's public broadcasting logic. The result proves that this judgment is extremely correct. The intimacy and information quality of Moments are more suitable for the social habits of mainstream Chinese users than the information flood of Weibo. Weibo gradually transformed into a public platform for celebrities and media, coexisting with WeChat in a differentiated manner rather than engaging in direct competition.

The third challenge comes from short videos. Tiktok was launched in 2016, and Kwai was earlier. This is the first real time diversion crisis that WeChat has encountered. Short videos do not compete for communication or social relationships, they directly compete for every minute that users stare blankly in front of their phone screens. Tiktok's algorithm recommendation mechanism is extremely accurate, which is actually the ultimate engineering of attention economy logic: use reinforcement learning algorithm to continuously optimize the "attention capture rate per minute", so that users can brush for an hour without expectation.

The two-sided market theory of Jean Tirole, a platform economist, has an interesting application here: WeChat and Tiktok actually compete for both sides of the same group of users - users as content consumers and users as communication nodes. Tiktok only challenges the former, and can't challenge the latter without it. This explains why although WeChat has fallen in the era of short videos, it has not shaken its foundation.

WeChat's response is video accounts. Launched in 2020, the development was slow, but with Zhang Xiaolong's continuous strategic bet, the daily activity and duration data of the video account have been increasing year by year. By 2024, video accounts will have become an indispensable third pole in China's short video industry. WeChat's response this time cannot be considered a failure, but it is by no means a complete victory. The duration advantage of Tiktok has not yet been completely shaken.

Have you ever lost on WeChat before the emergence of AI?

To be frank, before AI, WeChat had never fundamentally lost any competition.

It is a complete victory for operators. It is a complete victory over early instant messaging competitors. Weibo is strategically unbeatable (separating the two, but the core social function of WeChat is unshakable). The short video is a partial failure. The time users spend on Tiktok is really eroded, but WeChat does not rely on the length of entertainment, but on the irreplaceable status of communication infrastructure.

This is the true moat of WeChat: it's not an app you like to use, it's an app you have to use.

The work group is on WeChat, and the payment code is WeChat Pay. Parents can contact them through WeChat, and school notifications can be made through WeChat. Ordering takeout, registering, and checking housing provident fund can all be done through WeChat mini programs. WeChat is no longer a communication tool, it is the underlying operating system of China's digital life.

There is a theoretical concept worth distinguishing here: conversion cost. Economists distinguish between two types of conversion costs - one is functional, which means the cost of relearning and adapting to a new product; Another type is relational, meaning that switching platforms means the cost of disconnecting from existing social networks. WeChat's moat is mainly the second type. Users often use WeChat not because it is the best, but because the social network is already in practice. The cost of leaving is not losing a tool, but losing a set of social connections.

David Evans and Richard Schmalensee have pointed out in their research on platform competition that once this relational lock is formed, there are almost only two ways to break through: either the entire social network collectively migrates (extremely difficult), or a new demand scenario emerges that does not require the migration of old relationships at all (possible).

AI? It just provides the second possibility.

AI: The first challenger to truly change the structure of time

Returning to Herbert Simon's hierarchical classification of attention. In the past, all competitors of WeChat competed for users' shallow attention - entertainment time, fragmented browsing time, social chat time. These times can be divided by Tiktok and occupied by games, but the "just needed communication time" of WeChat is always unbreakable.

The emergence of AI assistants has triggered a different type of time for the first time: deep attention time.

What are users doing when they open DeepSeek, Claude, or GPT? Write articles, search for information, analyze data, process legal documents, learn new knowledge, and conduct investment research. The common feature of these scenarios is that users are in an active cognitive engagement state, rather than a passive sensory consumption state. This is Herbert Simon's concept of 'critical attention', which has a scarcity and value density far higher than the fragmented time spent scrolling through social media.

From the battery screenshot, it can be seen that a user spent 2 hours and 11 minutes interacting with Claude in one day, exceeding his active time on WeChat. This is not entertainment substitution, this is workflow substitution. Workflow is a territory that WeChat has never truly competed for.

Tencent is certainly aware of this threat. The hybrid big model has been launched, and WeChat is also exploring the integration path of AI functions internally. But there is a fundamental tension here: WeChat's product philosophy is' use it up and leave '. Zhang Xiaolong emphasized on multiple occasions that good products should not make users addicted and should complete tasks as soon as possible before leaving. There is an inherent conflict between this and the product logic of AI assistants hoping to become deep work partners for users.

There is a more direct reality beyond the conflict of product philosophy: the actual performance of mixed elements is lackluster.

In the most intense global AI competition of 2024-2025, OpenAI, Anthropic, Google, Meta, and even China's domestic DeepSeek are iterating at a visible speed, with a steep upward trend in their ability curve. Hybrid elements have always been lukewarm, rarely appearing in the first tier of mainstream evaluations in China, and have almost no presence internationally.

This is a confusing result for the market for a company with massive user data, ample funds, and top engineers. Tencent is not losing in resources, but more likely to lose in priorities and organizational culture: whether a company with social and gaming genes can mobilize sufficient strategic will in an era where "all in" basic model development is needed is still an open question.

The deeper issue is that the competition among AI assistants is not about functionality, but about cognitive trust. In economics, there is a concept of "experience good" - the quality cannot be judged before consumption, and can only be determined after use.

AI assistants go further in this regard: they are highly personalized experience products, the longer they are used, the deeper the context, the more accurate their understanding of you, and the higher the replacement cost. Once this stickiness is formed, its mechanism is similar to the social relationship chain stickiness established by WeChat back then, but what is bound is not the user's social relationships, but cognitive habits and working memory.

The situation faced by Tencent is somewhat similar to China Mobile in 2011- it's not that the product is not good, it's not that the resources are insufficient, but that a new dimension of competition has been defined, and the definer is not you.

The End of Time

Returning to the original question: Where did time go?

Herbert Simon's proposition has been fully verified in the mobile Internet era: attention is indeed the most scarce resource. In the past 15 years, WeChat has won one attention war after another. It first snatched up communication time from the operators, then secured social time through Moments, and locked in service time through mini programs and payments. WeChat has built a super platform that covers the digital lives of Chinese users almost 24/7.

But the hierarchical structure of attention determines that this war has never truly ended.

The rise of AI assistants represents a new paradigm of time consumption - not entertainment style shallow attention consumption, but tool style deep attention investment. Every minute users spend on AI is actively completing one thing: writing, researching, analyzing, and learning. The value density of this kind of time is much higher than scrolling through social media or watching video accounts. This high-value time is precisely the territory that WeChat has never truly guarded before.

Shapiro and Varian make a prophetic judgment in "The Rules of Information": in the information economy, lock-in is the core source of sustained competitive advantage, but the basis of lock-in can shift with the change of technological paradigms. WeChat focuses on social relationships; Tiktok is locked in sensory habits; AI will focus on cognitive habits and working memory. These three types of lock-in target different levels of the human attention system, and there is no simple substitution relationship between them.

Tencent is not without the ability to compete: there is no shortage of funds, data, and user base. But the challenge it faces is at the level of product philosophy: the greatness of WeChat lies in knowing what it is and what it is not. The rise of AI assistants is redefining "what is the core of digital life". If the answer shifts from "social" to "thinking and working", WeChat's product positioning will require a difficult self reconstruction, which has no precedent to follow.

The stock price may reflect precisely this uncertainty. The market is not worried about how much money Tencent will make today, but about where WeChat will stand in the next level of user attention migration?

This issue cannot be clearly explained by a battery screenshot. But it asked very correctly.

(The author is a researcher at Shanghai Institute of Finance and Law)