New developments have emerged regarding the previously refuted claim that former OpenAI researcher Shunyu Yao joined Tencent with an annual salary of over 100 million yuan.

Recently, Tencent issued an internal announcement stating that Shunyu Yao has been appointed Chief AI Scientist of the CEO/President’s Office, reporting directly to President Martin Lau. He also concurrently serves as Head of the AI Infrastructure Department and Head of the Large Language Model Department, reporting to Lu Shan, President of the Technology and Engineering Group.

Back in April, Yao sparked discussions by proposing the concept of the "second half of AI." He argued that AI development can be divided into two phases. The first half centered on models and methodologies, focusing on building foundational models and training approaches. As AI technology matures, the second half shifts to defining practically meaningful tasks and effectively evaluating AI system performance. At that time, he believed the industry was in an "intermission" stage.

Yao’s connection with Tencent dates back earlier. In a conversation in May, when asked how he would develop an AI Agent within WeChat if he were in charge of the app, Yao stated he would adopt a cautious approach, prioritizing observation of new interaction methods over hasty moves. "When holding a defensively strong position, why rush to launch an offensive?" he questioned, noting that disruptive innovation poses the greatest threat.

Yao’s High-Profile Recruitment Reflects Tencent’s Talent Drive

A graduate of Tsinghua University’s prestigious Yao Class for computer science and a PhD holder from Princeton University, Yao joined OpenAI after graduation before moving to Tencent over a year later, reporting directly to President Martin Lau. The 27-year-old’s career path embodies that of a top academic prodigy.

During his time at OpenAI, Yao contributed significantly to core research projects such as the Agent product Operator and Deep Research. His academic achievements include influential works like ReAct and Tree of Thoughts, which have each accumulated over 4,000 citations, with his total citations reaching nearly 16,000 on Google Scholar.

In September, rumors claimed Yao joined Tencent with a nine-figure annual salary, which Tencent officially denied via its WeChat public account. Since his recruitment, Tencent has intensified efforts to attract AI talent, reportedly offering double salaries to poach researchers from ByteDance and 50% above-industry pay to new PhD graduates. Yao has also helped Tencent’s Hunyuan team recruit core staff from ByteDance, Alibaba, and major Chinese AI startups including Kimi, MiniMax, Zhipu AI, and others.

From Slow Starter to Accelerated AI Push

Alongside Yao’s appointment, Tencent announced a major upgrade to its large model R&D structure, establishing three new departments: AI Infrastructure, AI Data, and Data Computing Platform.

The AI Infrastructure Department will build core capabilities for large model distributed training and high-performance inference services, providing technical support for algorithm R&D and business deployment. The AI Data Department will focus on data and evaluation system development, while the Data Computing Platform Department will construct integrated data intelligence platforms for big data and machine learning. Wang Di continues as Deputy General Manager of the Large Language Model Department, reporting to Yao, while Liu Yuhong and Chen Peng lead the AI Data Department and Data Computing Platform Department respectively, reporting to Vice President Jiang Jie.

Previously perceived as a latecomer in AI, Tencent has accelerated its efforts in 2025. In February, its cloud services and consumer-facing AI assistant Yuanbao quickly integrated DeepSeek, boosting Yuanbao’s downloads to top rankings. WeChat has also rolled out AI search features and enabled users to interact with Yuanbao directly within the chat interface. At the 2025 World Artificial Intelligence Conference in July, Tencent unveiled its Hunyuan 3D World Model and embodied intelligence platform Tairos, showcasing its full product ecosystem for the AI era.

In a discussion with industry figures in May, Yao and others debated WeChat’s AI strategy. Li Guangmi argued that WeChat risks losing its edge if it fails to aggressively develop Multi-Agents and Long-Term Memory capabilities, while Yao emphasized the need to monitor how human-Agent interaction evolves, noting the importance of infrastructure development regardless of the outcome.

Zhu Xiaohu, Managing Director of GSR Ventures, commented that Tencent typically avoids early-stage trial-and-error spending, preferring to wait for market clarity before scaling up, a strategy that has helped it overtake competitors in the past. He believes domestic AI competition has just begun, with Alibaba newly ramping up consumer-facing AI efforts and Tencent yet to fully exert its strength.

New developments have emerged regarding the previously refuted claim that former OpenAI researcher Shunyu Yao joined Tencent with an annual salary of over 100 million yuan.

Recently, Tencent issued an internal announcement stating that Shunyu Yao has been appointed Chief AI Scientist of the CEO/President’s Office, reporting directly to President Martin Lau. He also concurrently serves as Head of the AI Infrastructure Department and Head of the Large Language Model Department, reporting to Lu Shan, President of the Technology and Engineering Group.

Back in April, Yao sparked discussions by proposing the concept of the "second half of AI." He argued that AI development can be divided into two phases. The first half centered on models and methodologies, focusing on building foundational models and training approaches. As AI technology matures, the second half shifts to defining practically meaningful tasks and effectively evaluating AI system performance. At that time, he believed the industry was in an "intermission" stage.

Yao’s connection with Tencent dates back earlier. In a conversation in May, when asked how he would develop an AI Agent within WeChat if he were in charge of the app, Yao stated he would adopt a cautious approach, prioritizing observation of new interaction methods over hasty moves. "When holding a defensively strong position, why rush to launch an offensive?" he questioned, noting that disruptive innovation poses the greatest threat.

Yao’s High-Profile Recruitment Reflects Tencent’s Talent Drive

A graduate of Tsinghua University’s prestigious Yao Class for computer science and a PhD holder from Princeton University, Yao joined OpenAI after graduation before moving to Tencent over a year later, reporting directly to President Martin Lau. The 27-year-old’s career path embodies that of a top academic prodigy.

During his time at OpenAI, Yao contributed significantly to core research projects such as the Agent product Operator and Deep Research. His academic achievements include influential works like ReAct and Tree of Thoughts, which have each accumulated over 4,000 citations, with his total citations reaching nearly 16,000 on Google Scholar.

In September, rumors claimed Yao joined Tencent with a nine-figure annual salary, which Tencent officially denied via its WeChat public account. Since his recruitment, Tencent has intensified efforts to attract AI talent, reportedly offering double salaries to poach researchers from ByteDance and 50% above-industry pay to new PhD graduates. Yao has also helped Tencent’s Hunyuan team recruit core staff from ByteDance, Alibaba, and major Chinese AI startups including Kimi, MiniMax, Zhipu AI, and others.

From Slow Starter to Accelerated AI Push

Alongside Yao’s appointment, Tencent announced a major upgrade to its large model R&D structure, establishing three new departments: AI Infrastructure, AI Data, and Data Computing Platform.

The AI Infrastructure Department will build core capabilities for large model distributed training and high-performance inference services, providing technical support for algorithm R&D and business deployment. The AI Data Department will focus on data and evaluation system development, while the Data Computing Platform Department will construct integrated data intelligence platforms for big data and machine learning. Wang Di continues as Deputy General Manager of the Large Language Model Department, reporting to Yao, while Liu Yuhong and Chen Peng lead the AI Data Department and Data Computing Platform Department respectively, reporting to Vice President Jiang Jie.

Previously perceived as a latecomer in AI, Tencent has accelerated its efforts in 2025. In February, its cloud services and consumer-facing AI assistant Yuanbao quickly integrated DeepSeek, boosting Yuanbao’s downloads to top rankings. WeChat has also rolled out AI search features and enabled users to interact with Yuanbao directly within the chat interface. At the 2025 World Artificial Intelligence Conference in July, Tencent unveiled its Hunyuan 3D World Model and embodied intelligence platform Tairos, showcasing its full product ecosystem for the AI era.

In a discussion with industry figures in May, Yao and others debated WeChat’s AI strategy. Li Guangmi argued that WeChat risks losing its edge if it fails to aggressively develop Multi-Agents and Long-Term Memory capabilities, while Yao emphasized the need to monitor how human-Agent interaction evolves, noting the importance of infrastructure development regardless of the outcome.

Zhu Xiaohu, Managing Director of GSR Ventures, commented that Tencent typically avoids early-stage trial-and-error spending, preferring to wait for market clarity before scaling up, a strategy that has helped it overtake competitors in the past. He believes domestic AI competition has just begun, with Alibaba newly ramping up consumer-facing AI efforts and Tencent yet to fully exert its strength.