Tencent’s AI Bet Pays Off: WorkBuddy Hits $500M Annualized Revenue, Cloud Growth Accelerates to 35%
TREE NEWS reports: Tencent reported third-quarter 2026 earnings on Tuesday, delivering a quarter reshaped by artificial intelligence. The Chinese tech giant’s cloud revenue growth accelerated to 35% from 21% in the prior quarter, driven by AI-related demand. Its flagship AI productivity tool, WorkBuddy, surpassed 10 million daily active users and generated more than $500 million in annualized revenue over the past seven days—more than double the revenue of Tencent Meeting. Management also warned that US chip restrictions will delay the next version of its Hunyuan model, though inference workloads remain largely unaffected.
AI Monetization Takes Center Stage
WorkBuddy, launched in June with an enterprise edition, has already signed up 60% of China’s top 100 companies as customers. Management compared its monetization model to China’s online gaming evolution, where subscription plus token-based usage allows users to start small and scale spending as they extract more value. “Flash models are an excellent customer acquisition tool, not a profit killer,” said President Martin Lau, addressing concerns that cheaper models would dilute pricing. He noted that some heavy users spend up to RMB 40,000 per month on tokens.
The company’s cloud segment is also benefiting. Model-as-a-service revenue more than tripled quarter-over-quarter, and GPU rental revenue grew over 30% year-over-year. Capital expenditure surged 257% year-over-year to RMB 63.6 billion, pushing free cash flow to negative RMB 69.1 billion. Excluding prepayments for compute, free cash flow would have been positive RMB 16.9 billion.
Chip Constraints Cast a Shadow
Lau acknowledged that both leased and self-built compute face constraints, which will inevitably affect model training. “In the short term, it may take longer to launch the next version of our Hunyuan model,” he said. However, he stressed that inference—the backbone of cloud revenue—can run efficiently on domestic chips. “Our cloud revenue will not be materially adversely affected,” he added, noting that WorkBuddy is already the largest customer for nearly all open-weight models in China.
The company also disclosed early progress on Xiaowei, its WeChat-embedded AI assistant, which has reached 30 million DAU. Management expects Xiaowei to eventually match Meta’s capabilities while serving all 1.4 billion WeChat users at a quarterly cost lower than Yuanbao’s early-2026 spending. Lau projected that Xiaowei could add 10-20% to WeChat’s existing revenue streams within six to twelve months.
Market Implications
Tencent’s results offer a template for how large-cap internet companies can monetize AI at scale. The acceleration in cloud growth, driven by AI inference and model-as-a-service, suggests that the AI infrastructure buildout is beginning to yield tangible returns. However, the heavy capex cycle—with cash capex expected to peak in 2026-2027—will pressure free cash flow and margins in the near term. Management guided that capex will decline after 2027 as infrastructure is reused and chip prices fall.
For investors, the key debate is whether AI investments will generate sufficient returns to justify the spending. Tencent’s operating profit excluding new AI products grew 8% year-over-year, while reported non-IFRS operating profit fell 6%, highlighting the drag from AI investments. The company’s ability to scale WorkBuddy’s unit economics—through inference optimization and higher user spending—will be critical.
The chip restriction overhang remains a risk for all Chinese AI players. Tencent’s pivot to domestic chips for inference is a positive, but training constraints could widen the gap with global leaders. The company’s emphasis on building AI-native businesses with small teams, while using AI to streamline existing operations, is a strategic shift that could yield long-term efficiency gains.
Key Takeaways
- AI monetization is real: WorkBuddy’s $500M annualized revenue run-rate and rapid enterprise adoption demonstrate that AI can drive meaningful revenue for Tencent.
- Cloud acceleration: 35% growth, driven by AI inference and model-as-a-service, signals that Tencent’s cloud business is re-accelerating after years of slower growth.
- Capex peak: Heavy spending on AI infrastructure is weighing on free cash flow, but management expects capex to decline after 2027 as efficiency improves.
- Chip constraints: US restrictions will delay next-gen Hunyuan training, but inference—the key revenue driver—remains resilient on domestic chips.
- WeChat AI opportunity: Xiaowei could add 10-20% to WeChat’s revenue, with commercialization clarity expected within 6-12 months.




