Inside Tencent’s AI Pivot: From a “Leaking Ship” to a $53 Billion Bet on WeChat Agents

Yao Shunyu’s debut: the 28-year-old handed the AI keys

The most-watched figure at the conference was Yao Shunyu, the 28-year-old former OpenAI researcher whom Tencent appointed as Chief AI Scientist on December 17, 2025 . Reporting directly to president Martin Lau, Yao is a product of Tsinghua University's elite Yao Class and holds a PhD from Princeton, where he proposed foundational AI agent frameworks like ReAct and Tree of Thoughts . At OpenAI, he worked on core agent products including Operator and Deep Research before joining Tencent .

This was Yao's first in-person public appearance since joining nearly six months earlier, and the packed venue underscored the weight of expectations on him . In March 2026, Tencent had dissolved its decade-old AI Lab — home to more than 70 PhDs and 300 engineers — effectively consolidating its AI future under Yao's leadership . The conference was his opportunity to publicly frame the road ahead.

“The second half is just beginning”

In a joint dialogue with Tang Daosheng, Tencent's Senior Executive Vice President and CEO of Cloud & Smart Industries Group, Yao addressed the persistent narrative that Tencent had fallen behind in AI . He described AI as a "long-term endeavor" and argued that the industry is only at "the beginning of the second half" — a phase where the focus shifts from finding methodologies to solving meaningful, real-world problems .

Crucially, Yao rejected the idea that a single AI model or product will dominate the market. He stated plainly that neither ChatGPT nor Claude would become the sole dominant application, a thesis that underpins Tencent's strategy of embedding AI across its vast ecosystem rather than racing to win a single benchmark .

Tang Daosheng was equally direct. He acknowledged that "people often like to pick on a specific point to criticize Tencent," and admitted the company's portfolio is so diverse that "there are indeed areas where it can perform better" . But he challenged critics to "set even higher expectations" and framed AI as a marathon, not a sprint . The message was clear: Tencent had taken "some detours and engaged in some experimentation," but was not rushing to commercialize prematurely .

Most of Tencent’s code is now AI-generated

One of the most concrete disclosures came from Tang, who revealed that most of Tencent's code in 2026 is generated by AI . The company's engineers have shifted from writing code themselves to a supervisory role — reviewing architectural design and directing AI coding agents that produce the actual code . It is a significant productivity transformation that signals how deeply AI has penetrated Tencent's own operations, even as the company faces external doubts about its competitive position.

WeChat agents and a $53 billion market surge

The conference took place against the backdrop of a dramatic market event. On June 2, the Financial Times reported that Tencent was testing a prototype AI agent embedded inside WeChat — an agent that would allow users to access the platform's ecosystem of 3.8 million mini-programs through a simple swipe-and-command interface . The news triggered a 10.46% single-day stock surge to HK$481.60, adding approximately $53 billion in market capitalization .

The rally reflected investor conviction that WeChat's 1.4 billion users and its deep integration of payments, commerce, and services give Tencent a distribution advantage no model alone can replicate . Analysts framed the move as a bet not on closing the model-quality gap, but on leveraging the ecosystem .

At the conference, WeChat confirmed agent-to-agent partnerships with Huawei, Xiaomi, OPPO, and vivo — extending its AI capabilities directly onto device-level assistants from China's largest smartphone makers . The partnerships suggest Tencent's AI strategy is not confined to its own apps but aims for ambient presence across the hardware layer, too.

Pony Ma’s “leaking ship”

Underpinning all of this is an unusual moment of executive candor. At Tencent's annual shareholder meeting on May 13, 2026, CEO Pony Ma Huateng was asked directly whether the company was falling behind in AI. His response has become the defining metaphor of Tencent's AI journey :

“A year ago, we thought we had boarded the ship. Later, we found the ship was leaking. Now, we feel we are standing on it, but we cannot sit down yet. We still hope the ship can sail faster.”

The metaphor acknowledged overconfidence, ongoing instability, and incomplete recovery — but also signaled that Tencent is no longer in denial. By June 5, the message from the conference was that the company had stabilized the vessel, charted a long-term course, and was betting its AI future on ecosystem depth rather than short-term headline wins.