Tencent Merges Multimodal and LLM Teams Into New AI Unit

Tencent Holdings has consolidated its multimodal and large language model teams into a single Foundation Model Department, placing the newly unified unit under the leadership of chief AI scientist Yao Shunyu, the company announced on July 23, 2026. The restructuring, reported by Caixin Global, signals the Chinese tech giant’s most aggressive move yet to centralize its AI capabilities and compete more effectively in an increasingly crowded market.

Background: A Year of Structural Change

The merger of the Hunyuan Multimodal Model Department and the Large Language Model Department represents the culmination of a year-long restructuring of Tencent’s AI operations. In March 2026, Tencent dissolved its AI Lab and reassigned core research personnel to the Hunyuan team under Yao Shunyu, who had formally taken charge as chief AI scientist in December 2025 after joining the company the previous summer.

According to TMTPOST, the structural consolidation brings together previously separate research and engineering teams under a unified organizational architecture designed to centralize technical talent and high-performance computing infrastructure. Tencent’s official rationale stated the move was intended to “improve model R&D and collaboration efficiency, and to explore the upper limit of full-modal intelligence.”

Why Tencent Is Restructuring Now

The reorganization comes at a moment of intense competitive pressure in China’s AI landscape. On July 6, Tencent released its Hunyuan Hy3 model, triggering a surge in usage that lifted the company’s shares by 11% over three trading days. But just ten days later, rival Moonshot AI launched Kimi K3, a model so successful that demand forced the company to pause new sign-ups within 48 hours. Tencent’s stock fell 4.63% on July 17 in response.

As noted by CryptoBriefing, the market’s dramatic reaction to product launches indicates that AI capabilities are now considered a core valuation driver for Tencent. The company plans to double its AI expenditure from RMB 18 billion in 2025 to over RMB 36 billion in 2026, one of the largest single-year increases in AI investment globally.

Talent and Computing Power: The Twin Challenges

Yao Shunyu, a Princeton PhD and former OpenAI researcher aged approximately 27–28, has brought an assertive, results-driven management style to Tencent’s AI operations. A Hunyuan researcher told 36kr that “Shunyu is more anxious than Tencent’s senior management,” referring to his persistent lobbying for greater computing power allocation. Yao’s strict management approach has included warning team members that data issues would lead to immediate dismissal.

The computing power constraints are real. Tencent’s 2025 capital expenditure of RMB 79.2 billion significantly lagged behind Alibaba’s RMB 126 billion, according to their respective financial disclosures. With limited resources relative to its peers, optimizing allocation has become a strategic imperative. The new unified department is designed to address this challenge by eliminating redundancies and concentrating computing resources on priority research.

Meanwhile, talent churn continues. Hu Han, head of Hunyuan Multimodal Understanding, resigned in late July 2026 to launch his own startup. His departure is offset by the arrival of Tian Yonglong, a former OpenAI researcher and MIT PhD, who joined in early July to lead vision-language model R&D under Yao’s direction.

Industry Significance

The merger reflects a broader industry trend across Chinese technology companies. As TMTPOST noted, structural realignments across major internet platforms reflect a transition toward native multimodal architectures as standalone text models face diminishing technical returns. By unifying specialized research talent and high-performance hardware, companies are attempting to reduce development friction while building complex models capable of processing vision, language, and spatial data simultaneously.

Tencent is also supplementing internal development through external investments. The company invested 1.4 billion yuan ($210 million) in Kuaishou’s Kling AI video model, taking a 1.1% stake in the startup, which raised 3 billion yuan in its latest funding round.

What to Watch

Several questions remain as Tencent executes its AI overhaul. Can Yao Shunyu secure sufficient computing resources to compete with Alibaba’s massive infrastructure budgets? Will the unified department produce a flagship omni-modal model capable of closing the gap with competitors? And how will the integration of Yao’s OpenAI-influenced culture mesh with Tencent’s broader corporate environment?

With AI capabilities now a primary driver of Tencent’s market valuation, the stakes could not be higher. The Foundation Model Department’s success — or failure — will likely define Tencent’s position in the next era of artificial intelligence.