Zhou Jingren, the first Chief Scientist in the history of Alibaba

According to a report by Zhidx on June 8th, just now, Alibaba has added fuel to the fire of the popular domestic "Token Factory", with CEO Wu Yongming leading the charge.

Alibaba announced the merger of the Tongyi Big Model Division and the Future Life Laboratory to establish the Token Foundry Division, which will be directly led by Wu Yongming.

This adjustment involves multiple AI businesses. Zhou Jingren, a partner at Alibaba and the head of the Tongyi Big Model Division, will serve as the Chief Scientist of Alibaba and lead the establishment of the Alibaba AI Future Research Institute, focusing on the exploration of cutting - edge AI technologies. On April 8th, Li Feifei took over the position of CTO of Alibaba Cloud from Zhou Jingren. The title of Chief Scientist is the highest academic title in Alibaba's technology system. This appointment not only recognizes his contributions but also represents a strategic deployment for the future.

Meanwhile, Zheng Bo, the vice - president of Alibaba, will lead teams such as Happy Horse and Happy Oyster to join the Token Foundry Division. Zheng Bo joined Alibaba in September 2017. He worked at Google from 2006 to 2017 and was promoted to the M6 (vice - president) level in January 2022. He has served as the vice - president of technology at Taotian Group, the head of algorithm technology, and the CTO of Alimama.

After establishing the ATH (Alibaba Token Hub) Business Group in March this year, Alibaba has once again made organizational changes in the AI field within just three months. The ATH Business Group is led by Wu Yongming himself and integrates five major sectors, including the Tongyi Laboratory, the MaaS platform, Qianwen, Wukong, and AI innovation. It is a new organization with the core goal of "creating Tokens, delivering Tokens, and applying Tokens".

If the ATH Business Group focuses on horizontally coordinating the circulation and resource allocation of Tokens across the entire group, then the newly established Token Foundry Division is more like further focusing on the native R & D of models and the production of multi - modal models, and it is the "casting workshop" for Alibaba's core model production capacity.

By merging two core R & D units, Alibaba will more concentrate on scheduling computing resources to prioritize the efficiency of model training and iteration, thereby optimizing the allocation of computing resources for model R & D and avoiding redundant construction and resource waste.

This is Alibaba's continuous investment in the AI field. On May 20th, at the Alibaba Cloud Summit, Alibaba presented a full - stack Agent - based blueprint of "chip - cloud - model - inference". Zhou Jingren announced on that day that Alibaba had launched the latest Qwen - 3.7 - Max model. This model ranked second globally and first among domestic models in the Code Arena evaluation, and topped the LiveCodeBench v6, winning wide recognition.

The two models, Happy Horse and Happy Oyster, brought by Zheng Bo's team are also the "pioneers" of this integration. Happy Horse, as a large - scale video - generation model, has shown leading performance in multiple AI video evaluations. Happy Oyster is Alibaba's first open - world model. Based on a native multi - modal architecture, it supports the generation of a continuously running physical world from text or images. Users can interact in real - time and switch perspectives, and its consistency is better than similar products.

Alibaba's financial report for the previous quarter showed that the quarterly revenue of its AI products reached 8.971 billion yuan, accounting for more than 30% of its external commercial revenue for the first time, with an annualized revenue of over 35.8 billion yuan. Wu Yongming said in the earnings call: "AI has crossed the initial investment stage and entered the cycle of large - scale commercial returns." From this perspective, the significance of Token Foundry may not only lie in integrating computing resources but also in encapsulating Alibaba's base model capabilities into directly commercializable Token products. The trend is that models are becoming products.

Globally, the AI competition is shifting from a model race to a Token economy race. OpenAI, Google, and Microsoft are all exploring ways to transform model capabilities into scalable services. Tokens are becoming the core carrier connecting computing power, models, Agents, and applications. By concentrating R & D resources and optimizing computing resource allocation through Token Foundry, Alibaba is expected to drive the iteration of model capabilities and commercial transformation with a shorter organizational chain.

This article is from the WeChat official account "Zhidx" (ID: zhidxcom), written by Li Shuiqing and published by 36Kr with authorization.