Tencent is navigating a critical phase where AI investment and commercialization are advancing in parallel. Its core businesses continue to generate profits, providing the financial backing for heavy spending on AI infrastructure and native applications.
During the Q2 2026 earnings call on August 12, Tencent President Martin Lau stated that the company is making large-scale investments in AI compute power and is already seeing significant upside potential for returns. On one hand, multiple AI-native applications are performing well. On the other hand, using compute power for cloud leasing is expected to generate considerable revenue, improving the return on capital expenditure. For some compute orders placed earlier, if sold now, the selling price could be over 30% higher than the purchase price from just a few months ago.
Tencent is building its AI infrastructure with unprecedented intensity. Second-quarter capital expenditure surged to RMB 52.8 billion, far exceeding the RMB 32.1 billion consensus estimate and nearly tripling from RMB 19.1 billion in the same period last year. Free cash flow was negative RMB 13.8 billion. The company noted that excluding prepayments for AI compute procurement, free cash flow would have been RMB 37.6 billion. This indicates that the current cash flow pressure stems primarily from upfront AI infrastructure investments, not from deterioration in traditional business operations.
These investments are primarily allocated to upgrading the Hunyuan model, supporting the inference needs of WorkBuddy and CodeBuddy, advancing the WeChat AI initiative, and meeting the growing cloud service demands from external clients. Management views AI infrastructure as a prerequisite for future business growth: first, build intelligent capabilities through a top-tier model; second, leverage AI applications to translate model capabilities into user demand; and finally, generate commercial returns.
Regarding the pace of AI investment, Lau said the situation remains highly dynamic. Tencent will remain relatively cautious until it sees a truly breakout opportunity, at which point it will ramp up spending, as seen with WorkBuddy. The overall strategy focuses on long-term returns. Over time, the economic benefits of the AI business will gradually become apparent, ultimately leading to profitability.
On the commercialization front, Tencent is exploring monetization through AI applications, such as selling tokens, while retaining compute leasing as a "fallback option." Lau stated that if the business model were directly shifted to leasing compute power, Tencent would not only avoid losses but could actually turn a profit. The company has always had this alternative, which gives it considerable confidence.
WorkBuddy: Clear Positioning as an AI "Orchestrator," Healthy Gross Margins in Paid Segment
At the application level, management highlighted WorkBuddy. Its positioning is not as a single model but as an AI "orchestrator" that integrates different models and skills, selecting the appropriate solution based on user tasks to balance task quality and cost efficiency.
Hunyuan will be one of the models supported by WorkBuddy but not the only option. If Hunyuan's performance and ratings lead in practical tasks, it could become the platform's dominant model. Meanwhile, more models and skills from developers will be added in the future to cover diverse user needs.
This means Tencent does not intend for a single model to dominate everything. Instead, it aims to improve the overall efficiency of AI services through a combination of models, skills, and applications.
From a profitability perspective, WorkBuddy is already showing healthy commercialization signals. Tencent Chief Strategy Officer James Mitchell stated that the gross margin for WorkBuddy's paying users is already comparable to the overall gross margin of Tencent Cloud. The overall gross margin for WorkBuddy is slightly lower, mainly due to free users, as Tencent continues to subsidize to expand market share. However, the paid segment has already generated "quite good" gross margins.
Hunyuan Strives for SOTA, Model and Product Iterate in Synergy
On the model front, management revealed that Hunyuan 4 has reached a milestone phase and will eventually be upgraded to Hunyuan 5, continuously moving towards state-of-the-art (SOTA) performance.
Management stated that once SOTA is achieved, Tencent will form a matrix of models of varying sizes, meeting diverse user needs at different cost and performance levels. This strategy emphasizes matching models of different sizes to specific application scenarios, rather than simply pursuing a "large model."
At the same time, Tencent is driving collaborative design between models and products. As Hunyuan 4 and Hunyuan 5 iteratively improve, the capabilities, feature richness, and execution speed of related products will also advance. For Tencent, the deep integration of model capabilities and application experience will be a key step in transitioning the AI business from technical investment to commercial returns.
Xiaowei: AI-Powered Reshaping of the WeChat Ecosystem
Regarding Xiaowei, the WeChat AI assistant currently in beta testing, Lau positioned it as a key vehicle for the WeChat ecosystem to enter the "AI era." He compared its potential leap to the "10x value amplification" WeChat brought relative to QQ.
In the longer term, Tencent hopes users can simply issue natural language commands, and Xiaowei will execute transactions and operations on their behalf. Simultaneously, merchants and mini-programs within the WeChat ecosystem will eventually deploy their own agents, enabling automated interactions between user agents and merchant agents.
On cost control, Lau stated that the ongoing inference cost for Xiaowei will be lower than Tencent's previous investment in Yuanbao. He emphasized that the design of its VLM (Vision-Language Model) is specifically intended to balance privacy protection and cost efficiency within the WeChat environment.
Addressing concerns that "agent transactions could erode advertising inventory," he argued that the AI-driven transformation of WeChat will create incremental value, not simply replace the existing business model.
Yuanbao: From Subscription Revenue to Token Commercialization
Commercialization for Yuanbao is also beginning to show a clearer economic model. Tencent management disclosed that the current gross margin for paying users of Yuanbao is already comparable to the overall gross margin of Tencent Cloud.
Management pointed out that due to the subscription model's revenue recognition mechanism, similar to the gaming business, there is a time lag between current cash inflows and revenue recognition in financial statements. Therefore, the cash revenue growth currently seen will gradually translate into reported revenue growth for Tencent Cloud within the year.
Among three cloud business opportunities—GPU bare metal leasing, Model-as-a-Service (MaaS), and Yuanbao token production—management believes token production holds the most enduring economic value and thus has the highest priority. For free users, the company continues to subsidize to expand market share, but paying users already have a healthy gross margin foundation.
Capital Allocation: Dynamic Balance Between AI Investment and Share Buybacks
Faced with a significant increase in AI capital expenditure, Tencent does not treat buybacks and AI investment as a fixed binary choice. Instead, it emphasizes that capital allocation will be dynamically adjusted based on return rates.
Management stated that if AI capital expenditure can demonstrate significantly superior returns compared to share buybacks, the company will correspondingly increase capital expenditure.
Lau emphasized that the AI infrastructure investment is essentially a finite "lump-sum initial investment" and is not expected to grow linearly every year. Therefore, evaluating this expenditure should not focus solely on current free cash flow but should also consider cash on hand, the value of the investment portfolio, operating cash flow, and reasonable debt capacity.
For Tencent, the core logic of current AI investment is shifting from "how much to invest" to "how to generate returns." As compute leasing, token production, and AI-native applications gradually commercialize, Tencent hopes to ultimately transform its upfront infrastructure investment into a new growth curve.
Comments