Inside Alibaba’s Token Foundry: Merging AI Labs for a Full-Stack Strategy

While ATH operates on a “create Token, deliver Token, apply Token” framework, Token Foundry is the engine that handles the first step: creating the foundational AI capabilities and packaging them for real-world use .

Leadership and Key Appointments

The restructuring came with clear leadership changes that signal Alibaba’s intent to separate long-term research from near-term product deployment.

Eddie Wu (Wu Yongming), Alibaba’s Group CEO, took direct oversight of Token Foundry, a sign of how central AI has become to the company’s strategy . Zheng Bo, an Alibaba vice president, was appointed to run day-to-day operations, overseeing model productization and the integration of Future Life Lab’s teams into the new unit .

Meanwhile, Zhou Jingren, who had previously led the Qwen model development, was named Alibaba’s Chief Scientist and tasked with launching a separate AI Future Research Institute. His institute will focus exclusively on long-term, frontier AI research, keeping it distinct from Token Foundry’s commercial mandate .

The Labs Being Consolidated

The consolidation brings together two teams that had at times overlapped in their work.

The Tongyi Large Model Business Unit is the group behind the Qwen series of large language models. Its latest release, Qwen 3.7 Max, was built with native agent capabilities and supports a 1 million token context window, designed for autonomous task execution over extended periods .

The Future Life Lab, originally part of Alibaba’s Taobao and Tmall Group before moving into ATH, had been building applied generative AI models. It developed two notable projects that are now key assets within Token Foundry .

Future Life Lab Projects: Happy Horse and Happy Oyster

Zheng Bo’s team at Future Life Lab created two AI models that have attracted industry attention .

Happy Horse is a multimodal video generation model that reportedly ranked first globally in its category. It was developed to serve as an AI-generated content tool for Alibaba’s e-commerce and entertainment businesses .

Happy Oyster is described as a real-time world model that allows interactive creation, functioning like a dynamic simulator where users can generate visual environments through natural interaction . Alibaba’s recent financial reports note that both models are now being commercialized in phases .

Strategic Purpose of the Restructuring

Token Foundry is not just a rebranding exercise. The restructuring addresses several strategic pressures Alibaba faces in the accelerating AI landscape.

Ending Internal Competition – Both Tongyi and Future Life Lab had been developing video generation models. By merging them, Alibaba aims to stop duplicated efforts and pool resources under unified leadership .

Forcing Commercialization – The creation of a dedicated AI Future Research Institute for Zhou Jingren explicitly frees Token Foundry to focus on productization and revenue generation. The unit is tasked with turning model capabilities, token-based services, and agent applications into standardized, deliverable products .

Building a Full-Stack AI Business – With Tongyi providing the foundation models, Happy Horse covering video generation, Happy Oyster exploring world models, and Alibaba’s Pingtouge chip division handling hardware, the restructuring completes a vertical integration from silicon to application . This aligns with a broader push to monetize AI, especially as Alibaba’s cloud computing revenue recently surged 38%, validating the market opportunity .

The formation of Token Foundry marks Alibaba’s third major AI reorganization in 2026, following the creation of ATH and earlier team integrations. It reflects a deliberate shift from experimental lab projects to a structured product pipeline, with CEO Eddie Wu firmly at the controls.