Yao Shunyu: AI Moat Lies in Product Layer, Private Data and Trust Are Key

Tencent's Chief AI Scientist Yao Shunyu argued at the 2026 Future Science Prize dialogue that the moat for AI companies lies more in the product layer than the model layer. He believes that model capabilities are becoming commoditized, and what is truly irreplaceable are AI products that understand user context, leverage proprietary data and context, and win trust.

Citing WeChat as an example, Yao noted that a product that holds users' schedules and social relationships is hard to replace by a smarter model even if its model capability is not superior. He emphasized two sources of moat: private data and context that others cannot access, and users' trust in AI to handle data and take actions. Zhang Ya-qin joked about Anthropic's valuation logic, reflecting industry skepticism about long-term barriers at the model layer.

This view has important implications for the AI industry: as model capabilities homogenize, competition will shift from technology itself to product experience and data ecosystems. For developers, building product applications based on private data and user trust may offer more long-term value than merely pursuing model performance.