Tencent's Chief AI Scientist Says AI Moat Lies in Products, Not Models

At a 2026 Future Science Prize dialogue, Zhang Yaqin, a chair professor at Tsinghua University, asked Yao Shunyu, Tencent's chief AI scientist, what the moat of model companies really is. Yao said the moat lies more at the product level than the model level. What is truly hard to replace is an AI product that understands users, uses private data, and earns user trust.

He gave WeChat as an example. An AI inside WeChat may not need the strongest model, but it knows the user's schedule, friends, and activities. That makes it hard to replace. Even if a rival has a smarter model, it may not know who the user is. Zhang Yaqin joked that by this logic, Anthropic might someday be worth less than one or even two trillion dollars, as models could become commodities.

Yao graduated from Tsinghua's Yao Class and earned a computer science PhD from Princeton. He worked on AI agent products at OpenAI. In December 2025, Tencent upgraded its AI research structure, and he became chief AI scientist, leading the foundation model department and reporting to President Martin Lau.