Is Tencent’s AI Really Lagging? Yao Shunyu’s Long‑Term, Multi‑Modal Perspective

During the Tencent Cloud AI Industry Application Conference, senior executive Tang Daosheng asked Yao Shunyu whether Tencent’s AI was lagging. Yao responded with two core judgments.

First judgment: AI is a long‑term game, not a short‑term hype. He noted that while some Silicon Valley voices predict massive job loss within two years, Yao believes the real "second half" of AI is just beginning.

Second judgment: The AI landscape will become more diverse rather than following a single linear path. He criticized the current focus on a narrow pipeline—pre‑training, post‑training, reinforcement learning, agents, and coding agents—as a “gray” scenario where everyone copies the same approach.

Yao clarified his use of the term "second half," originally defined in his 2025 blog post as the stage where evaluation outweighs training. In the Tencent interview, he reinterprets it to mean that while foundational methods (pre‑training and post‑training) now act as a universal "hammer," the challenge is identifying valuable problems (the "nails") to strike.

He illustrated this with an example: early AI breakthroughs like AlphaGo solved a specific game, whereas modern foundation models can address a wide range of tasks, making problem selection the critical bottleneck.

Yao highlighted three factors that make Tencent attractive for AI development:

Context: Both personal and enterprise context provide the raw inputs that give models a competitive edge.

Culture: A low‑ego, trust‑based environment encourages honest feedback and long‑term commitment.

Resources: Sufficient infrastructure to iterate on foundation models.

He then described a three‑layer AI organization:

Foundation (Infra) layer: Build solid pre‑training and post‑training capabilities with ample resources.

Product layer: Apply models to create real‑world value, requiring strong product sense.

Frontier layer: Explore new research paradigms and opportunities, a space still under‑explored in China.

According to Yao, the most important goal is to establish a long‑term, AGI‑focused organization in China that balances these three layers, leveraging Tencent’s strengths in context, culture, and trust to drive the next phase of AI development.