Summary of Key Points
Tencent's second-quarter financial report showed revenue (204.8 billion yuan, up 11%) and profit (68.4 billion yuan, up 9%) that exceeded expectations, yet its stock price fell by 4.46%. The reason is that Tencent invested over 100 billion yuan (52.8 billion yuan in capital expenditure + 51.4 billion yuan in prepaid computing power) to enhance its AI capabilities, resulting in a negative free cash flow of -13.8 billion yuan. The market is not interested in the “story” around AI investment; instead, they want to know when Tencent will start making money. Tencent's hopes for AI success lie with WorkBuddy, an office AI assistant that addresses the issue of AI being capable of only chatting but not performing actual tasks. WorkBuddy is currently the leading office AI solution in China, but Tencent still faces challenges such as weaker own models and limited computing power. Tencent’s strategy is to “catch up after others” (to invest when the business model is clear), which may prevent it from taking a leading role in the future.
1. Financial Report “Looks Good,” but Met with Cold Reception – Where’s the Return on Such High Spending?
Tencent’s second-quarter performance was actually quite good, with both revenue and profit exceeding market expectations, and its three main businesses (advertising, gaming, and fintech) were stable. However, the capital market was unimpressed because Tencent suddenly became very generous: it spent 52.8 billion yuan on capital expenditure (such as building data centers and purchasing equipment) and prepaid 51.4 billion yuan for computing power, totaling over 100 billion yuan.
The negative free cash flow indicates that Tencent spent more than it earned in this area. The market is tired of the slogan “AI will change the world” and is more concerned about when the investment will pay off. As a result, the stock price declined.
2. WorkBuddy – Is It Tencent’s “Lifesaver” for AI?
WorkBuddy is an AI office assistant that can perform practical tasks, such as organizing sales data, analyzing trends, creating PPTs, checking for spelling errors in articles, and verifying data sources. Unlike previous AI solutions that could only chat, WorkBuddy solves the problem of being capable of only talking but not doing actual work. It has performed exceptionally well, with over 20 million monthly PC visits in June, making it the leading office AI solution in China. Government officials have even used it in official apps, demonstrating its practical value. Tencent’s advantage lies in its ecosystem, as tools like WeChat, Enterprise WeChat, and Tencent Docs already have a large user base, and WorkBuddy can integrate these to create new traffic channels.
Of course, other companies are also competing in this market: Alibaba has Qianwen Office, ByteDance has DouBao Office and TRAE Work, and all are aggressively advertising their products in elevators and subways, making the competition fierce.
3. Tencent’s AI Weaknesses – Models and Computing Power
Despite being useful, WorkBuddy has two major drawbacks:
1. Weak Own Models: Tencent’s Hunyuan model previously ranked low globally, but the latest Hy3 version has surged to first place after becoming free. However, free access cannot sustain this success in the long term. The model is the “brain” of AI, and if it is not powerful enough, WorkBuddy’s capabilities will be limited.
2. Limited Computing Power: Training and running AI models require a lot of computing power (e.g., GPU chips). With chip prices rising in recent years, Tencent had to prepay 51.4 billion yuan for computing resources; otherwise, it might have faced difficulties.
However, Tencent is working to address these issues: it has hired AI expert Yao Shunyu from OpenAI to train the Hy4 model and still has over 200 billion yuan in cash available for further investments.
4. Tencent’s AI Strategy – Smart, but Lacking Ambition?
Tencent’s approach is to “catch up after others,” waiting until the technology is mature and the business model is clear before making significant investments. This strategy is prudent and avoids initial waste, but it also shows that Tencent is less proactive in defining the future of AI.
In contrast, companies like SpaceX, founded by Elon Musk, are willing to invest in ambitious projects like rocket reclamation and Mars exploration, even though they haven’t yet made a profit. WorkBuddy, while useful, seems more like a combination of existing technologies rather than a breakthrough innovation. Tencent is good at turning mature technologies into profitable businesses but lacks the innovative spirit to be the first to try new things.
Ma Huateng admitted, “We thought we were on board the ship, only to find it leaking; now we’re on the ship, but we can’t sit comfortably.” This reflects that Tencent recognizes its lag in AI development. Whether Tencent can move from being a follower to a leader in the AI era will determine how far it can go.
Conclusion
Tencent’s approach to AI is cautious and focused on stability. WorkBuddy is a highlight, but to win in the AI race, it must overcome challenges with models and computing power. More importantly, it needs to shift from focusing on current profits to creating opportunities for the future. After all, history remembers those who dare to venture into uncharted territories, not just those that make money on established routes.