Tencent Earnings Call: Work Buddy Acts as "AI Orchestrator," Hunyuan Continues Push Toward SOTA

Tencent is entering a critical phase where AI investment and commercialization are advancing in parallel: core businesses continue to contribute profits, providing financial support for high-intensity investments in AI infrastructure and native applications.

On August 12, during Tencent's 2026 Q2 Earnings Call, Tencent President Martin Lau addressed the increase in the company's capital expenditures in the second quarter. He stated that Tencent is indeed making large-scale investments in computing power and has already seen clear upside potential for returns, with several new applications performing well. In addition, using computing power for cloud leasing business is expected to bring substantial revenue growth and improve the return on capital expenditures. For some of the previously placed computing power orders, if sold, they could yield profits more than 30% higher than the purchase price from a few months ago.

Tencent is building AI infrastructure with unprecedented intensity. Capital expenditures in the second quarter rose to RMB 52.8 billion, nearly doubling from RMB 19.1 billion in the same period last year, hitting a record high; free cash flow recorded a negative value of RMB 13.8 billion. The company stated that if prepayments for AI computing power purchases were excluded, free cash flow would still be RMB 37.6 billion. In other words, the core reason for the current pressure on cash flow is not the deterioration of traditional businesses, but Tencent's proactive front-loaded investment of substantial funds into AI computing power and infrastructure.

This round of investment is mainly used for the upgrade of the Hunyuan model, inference demands for Work Buddy and CodeBuddy, WeChat's AI layout, and meeting the continuously growing cloud service demands of external customers. Management believes that AI infrastructure is a front-loaded investment for future business growth: first achieving leading intelligence levels through top-tier models, then leveraging market-leading AI applications to translate model capabilities into user demand, ultimately forming long-term economic returns.

In terms of commercialization, Tencent has begun to explore monetization methods such as selling Tokens through AI applications. Its goal is to gradually establish a complete closed loop of "model capability – AI applications – user demand – commercial monetization," and the current rapid growth in computing power investment is the upfront cost that must be borne while this system is still under construction.

Martin Lau stated, "If we proceed in the order of first using computing power to build our own models, then developing applications, and finally leasing out computing power externally, we can build a highly scalable native AI business, which will bring very substantial profits."

Work Buddy Acts as AI "Orchestrator," Hunyuan Is Not the Only Model

At the AI application level, management highlighted Work Buddy. Its positioning is not as a single model, but as an AI "orchestrator" capable of integrating different models and skills, selecting appropriate solutions based on user tasks, and balancing task quality with cost efficiency.

Hunyuan will serve as one of the models supported by Work Buddy, but not the only one. If Hunyuan leads in performance and ratings in actual tasks, it is expected to become the mainstream model on the platform. Meanwhile, more models and skills provided by developers will be integrated into the platform in the future to meet diverse user needs.

Hunyuan Continues Push Toward SOTA, Collaborative Development of Models and Products

Regarding models, management revealed that Hunyuan 4 has reached a phased milestone and will subsequently upgrade to Hunyuan 5, continuing to move closer to the State-of-the-Art (SOTA) level in the industry.

Management stated that once SOTA is achieved, Tencent will form a model matrix covering different scales, solving various user problems at different cost and performance levels. This means Tencent is not just pursuing a single "large model," but is simultaneously laying out models of different sizes to adapt to different application scenarios.

At the same time, Tencent is promoting "collaborative design" between models and products. As Hunyuan 4 and Hunyuan 5 continue to iterate, the capabilities, feature richness, and execution speed of related products will also improve synchronously. The deep integration of model capabilities and application experience will become an important path for Tencent's AI business to establish competitive advantages.

Continuously updating

Risk Warning and Disclaimer

The market carries risks; investment requires caution. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investment based on this content is at your own risk.