By Wang Huiying

Edited by Ziye

At the 2017 Apsara Conference, Jack Ma established the DAMO Academy and pledged to invest over 100 billion yuan in technological R&D within three years.

Over the next two years, the DAMO Academy recruited 13 top-tier scientists to lead key areas such as machine intelligence, data computing, robotics, fintech, and quantum computing. Dubbed the "Thirteen Sweeper Monks" of the DAMO Academy, they formed a star-studded team.

In Jin Yong's novels, the Sweeper Monk is a hidden master in the Scripture Depository, quietly sweeping fallen leaves in the corner but capable of defeating multiple opponents single-handedly when necessary. Jack Ma, with a passion for martial arts, hoped that the DAMO Academy would outlast Alibaba itself.

Nearly nine years have passed, and the DAMO Academy remains vibrant, but most of these thirteen individuals have moved on. Some have pursued other paths, while others have stepped out of the core spotlight. Apart from Li Feifei, who joined Alibaba in 2018, only Zhou Jingren has stayed throughout.

Image sourced from DAMO Academy's WeChat official account

Like a steadfast city guardian, he has seen off batches of comrades and stands alone at the forefront of technological waves.

From the foundational era of cloud computing to the data intelligence era of e-commerce and finance, and now to the large model era, he has consistently made the right moves and withstood the pressure with each technological shift. His appointment as an Alibaba partner in 2025 was a recognition of this steadfastness.

However, in 2026, Alibaba AI underwent a dramatic restructuring. In April, Lin Junyang, the technical visionary behind Tongyi Qianwen, departed. In June, after three role adjustments within three months, rumors surfaced that Zhou Jingren was also set to leave.

Although Alibaba subsequently denied these rumors, a deeper question arose: As Alibaba AI's strategic focus shifts from technological breakthroughs to commercialization, and as its decentralized R&D system moves towards centralized governance, does Alibaba still need Zhou Jingren? Or, will Zhou Jingren choose to align with the organization's needs?

To understand this, we must look back at the past decade of Zhou Jingren and Alibaba AI.

Three Adjustments in Three Months: Is Zhou Jingren Being Promoted in Name but Demoted in Reality?

In just seven days, the rumored departure of Alibaba's chief scientist, Zhou Jingren, became the most talked-about topic in China's AI circle.

The story traces back to early June.

On June 8, Alibaba announced that Zhou Jingren, head of the Tongyi Large Model Business Unit, would serve as Alibaba's chief scientist.

However, within Alibaba, Zhou Jingren did not appear where he was most expected. On the same day, Alibaba announced another significant move: merging the Tongyi Large Model Business Unit and the Future Living Lab to form the Token Foundry Business Unit, directly overseen by Group CEO Wu Yongming.

In simple terms, the core model R&D team previously under Zhou Jingren's control was entirely transferred to this AI business unit directly managed by the CEO. Meanwhile, he was tasked with exploring cutting-edge technologies.

Shifting from leading the core R&D of the Tongyi Large Model Business Unit to focusing on cutting-edge exploration at the AI Future Research Institute, many industry insiders believe this means he no longer holds concrete operational power.

Speculation about a promotion in name but demotion in reality reached a fever pitch on June 12. Just four days after his new appointment, rumors swirled that "Zhou Jingren was leaving."

Interestingly, Alibaba did not respond promptly. Instead, its reply came two days later, dismissing the "Zhou Jingren resignation" rumors as false: "We have noticed that someone has been organizing the spread of this rumor online in recent days and call on everyone not to disseminate false information."

Despite Alibaba's denial, the dramatic sequence of events continues to unsettle the outside world.

The first domino in this personnel upheaval fell quietly three months earlier.

In early March, Lin Junyang, the core architect of Tongyi Qianwen and the mastermind behind the Qwen model, confirmed his departure. Along with him left key personnel from multiple teams, including post-training and code models. This technical team, which had propelled Tongyi to the pinnacle of global open-source platforms, lost nearly half its members overnight, leaving Alibaba's large model R&D frontline leaderless.

Image sourced from Alibaba Cloud's official website

Zhou Jingren was called upon to stabilize the situation. At the time, he was still Alibaba Cloud's CTO and vice president of the DAMO Academy. As the technical leader in charge, he urgently took over management of the Qwen core technology team, restoring stability to the R&D front. This was his first role adjustment in 2026, more akin to a firefighting mission than an expansion of power.

A month later, on April 8, Wu Yongming issued an internal letter to all employees, initiating the second round of AI organizational restructuring: a new Technical Committee was established at the group level, with Wu Yongming personally serving as its leader and Zhou Jingren appointed as chief AI architect. Simultaneously, the Tongyi Lab was upgraded to an independent Tongyi Large Model Business Unit, with Zhou Jingren assuming full responsibility.

This time, he officially stepped down as Alibaba Cloud's CTO, transitioning from the cloud business's technical leadership to AI technology strategic design at the group level. His role evolved from technical leader of the cloud business to Alibaba's top AI figure, marking a significant promotion.

Elevating Tongyi from within Alibaba Cloud to a group-level strategic priority meant that Alibaba was concentrating its resources on AI. Zhou Jingren, as the technical core, was pushed to an even more central position.

However, few realized that this was merely an intermediate step in the centralized adjustment. Ultimately, the Token Foundry was established, completing Alibaba's three-step centralization of AI at the group level. As AI reporting lines shortened, power further consolidated under Wu Yongming.

This determined that while Zhou Jingren's titles continued to rise, his actual power clearly contracted. The rapid pace of adjustments and the obvious shift in power boundaries left Zhou Jingren effectively sidelined from the core battlefield.

Throughout this process, unlike Lin Junyang's highly personal responses, Zhou Jingren himself has rarely been seen responding. Perhaps he has long grown accustomed to such twists and turns.

Last April, as Alibaba launched its AI offensive, Zhou Jingren remarked in a conversation with LatePost, "After all these twists and turns, it's been a decade."

Zhou Jingren stated that his multiple role transitions at Alibaba have broadened his horizons and exposure. Involvement in such a wide range of business scenarios has helped him gain clarity and make better-informed technical judgments.

Regarding technical judgment, Zhou Jingren early on proposed the classic concept of "Model as a Service." He also astutely pointed out that the core paradigm of large models lies in their integration into enterprises.

As someone so forward-thinking, he cannot be unaware of his current awkward position or the trend of the group's R&D priorities giving way to commercialization.

Zhou Jingren likely understands better than anyone that this is a seismic shift in the era.

Zhou Jingren's Decade at Alibaba AI

Those caught in the midst of change are often the most perceptive to shifting winds.

However, what truly defines Zhou Jingren is not his sensitivity to these winds but his ability to stand firm in his own way during each transition.

This ability stems from a contradictory yet unified aspect of his personality: the prudence and rule-following nature of a foreign corporate manager on one hand, and the directness and stubbornness of a technologist on the other.

Within Alibaba, there are many versions of who Zhou Jingren is.

Some say he carries a hint of Microsoft's office culture, is well-versed in workplace rules, possesses a certain degree of upward management skills, and is typically unsmiling and inscrutable—a classic example of a foreign corporate-trained manager.

Image sourced from Alibaba Cloud's WeChat official account

Regarding downward management, several former Alibaba employees commented that Zhou Jingren is not only capable of research and product development but also team leadership. He is willing to boldly empower young people and delegate authority. Both Zhou Chang and Lin Junyang, the two successive leaders of the Qianwen model team, were post-90s campus recruits.

However, some have witnessed a completely different side of Zhou Jingren. A colleague who worked with him for many years lamented, "Jingren can be quite straightforward at times."

During a conference, Alibaba's CPO, Tong Wenhong, encouraged everyone to speak their minds and point out existing issues at Alibaba. After a moment of silence, Zhou Jingren said something that instantly quieted the room: "Then I'll be honest—the code written by Taobao in the past was terrible."

A scientist who dares to point directly at code quality issues is, at his core, a technologically idealistic person. However, this trait alone does not fully define Zhou Jingren. If he were merely a technical purist, he could not have navigated Alibaba's system for a decade, undergone multiple role transitions, and remained steadfast at the table.

What truly sets Zhou Jingren apart is his rare dual nature: he understands underlying computing architectures and can also delve into the messy business of e-commerce search and recommendation systems.

How did he develop this "dual" capability? We must return to the autumn of 2016 when he joined Alibaba Cloud from Microsoft.

That year, Zhou Jingren was 42.

His previous background was a standard success story: enrollment in the University of Science and Technology of China's Youth Class, a Ph.D. in computer science from Columbia University, and a partnership in Microsoft R&D. Upon joining Alibaba, Zhou Jingren's first major task was to lead the establishment of iDST, the precursor to the DAMO Academy.

This was Alibaba's most challenging period in technological development. Zhou Jingren built Alibaba Cloud's AI computing foundation from scratch. He led the team in iterating the Feitian operating system, implementing the Shenlong computing architecture, and establishing the underlying infrastructure for Alibaba Cloud's AI computing power.

What truly transformed him from a technical expert into a rare breed was a seemingly off-course shift. After laying the foundation for Alibaba Cloud, he ventured deep into the business realm, transferring to the Taobao Search and Recommendation Business Unit to oversee data intelligence operations for Taobao and Tmall, including search, recommendations, and advertising.

There, he mastered the full suite of commercial logic behind search ranking, personalized recommendations, and advertising bidding. Later, he moved to Ant Group to lead the Intelligent Engine and Data Middle Platform, further refining his technical perspective in fintech scenarios.

This experience marked a crucial identity shift for Zhou Jingren. He was no longer just a technologist versed in distributed systems but a multifaceted individual who understood both underlying computing power and how algorithms could translate into GMV.

This rarity made Zhou Jingren invaluable when the AI era arrived.

In late 2022, ChatGPT burst onto the scene, and Zhou Jingren was entrusted with leading the R&D and iteration of the Tongyi Qianwen large model. With his dual technical and business mindset, the Tongyi series rose rapidly. By the 2025 Apsara Conference, Tongyi had released over 300 open-source models, with downloads exceeding 600 million.

In September of the same year, Qwen3-Max ranked among the top three globally in performance. By year-end, Zhou Jingren was selected as an Alibaba partner, the sole new addition as the partnership shrunk from 26 to 17 members. Alibaba management described Zhou Jingren in just five words: "Extremely not easy."

Image sourced from Alibaba Cloud's official website

During this process, Zhou Jingren also insisted on one thing: open-sourcing. While the industry increasingly chose closed-source monetization, he advocated for open-sourcing the Tongyi models and led the development of the ModelScope community. This strategy earned Tongyi widespread recognition in the global developer community, making Qwen one of the world's largest open-source model families.

From joining Alibaba Cloud in 2016 to becoming a partner in 2025, Zhou Jingren spent nearly a decade transforming himself into an indispensable piece of Alibaba's AI puzzle.

However, with Alibaba's AI strategic balance rapidly tilting, whether his decade-long technical reputation and irreplaceable dual capabilities serve as assets or burdens may no longer be in Zhou Jingren's hands.

Advancement or Retreat: Is Zhou Jingren Still Indispensable to Alibaba?

During his ten-year tenure at Alibaba, Zhou Jingren has seldom made public appearances.

Unlike some technical leaders who frequently participate in industry forums, media events, or post on social media, his name primarily surfaces on product launch PPTs during Alibaba Cloud Summits or in the technical speech agendas at Apsara Conferences.

Even during his high-profile appointment as an Alibaba partner in 2025, all the outside world witnessed was an internal memo and a few standard evaluations.

He is more akin to a "幕后推手" (power broker) working behind the technology scenes. When systems experience instability, Zhou Jingren personally leads teams to revamp technical modules, showcasing a certain degree of technical purity.

This unwavering commitment to technological idealism left him caught in the crossfire between Alibaba's AI strategy and its core technical staff during the March 2026 controversy surrounding Lin Junyang's departure, making him the most tormented intermediary.

He and Lin Junyang share a similar ethos, both favoring a pure and unrestricted academic environment in the laboratory. They are also cognizant that the vertically integrated model is the linchpin behind Qwen's meteoric rise. However, as an Alibaba partner and the technological linchpin of the group, Zhou Jingren's perspective cannot be one-dimensional. When the call for a strategic pivot arises and the organization's will must be executed, Zhou ultimately opts to stand in support of maintaining Alibaba's strategic unity.

This is a compromise that a technology manager within a power structure must make. Zhou Jingren is acutely aware that the duration of his tenure as this 'ballast' is not contingent upon him but rather on the trajectory of Alibaba's AI strategy.

Image source: Alibaba's official website

Over the past three years, Alibaba AI's primary mission has been technological catch-up and the construction of its proprietary large model foundation. Commencing from 2026, Alibaba's AI strategy is unequivocal: prioritizing the rollout of scalable products and full-scale commercialization.

At this juncture, the shortcomings of fragmented AI R&D resources and inadequate business collaboration across Alibaba Cloud, Damo Academy, Taotian, and Local Services have become increasingly apparent. Consequently, management has taken measures to consolidate AI capabilities.

In March, the ATH Business Group was established to integrate multiple AI business lines. In April, a unified Group Technology Committee was formed to coordinate technical roadmaps. In June, the Token Foundry Business Unit was established under the direct supervision of the CEO to streamline the entire chain of model R&D, product iteration, and full-group scenario monetization.

This rationale is not difficult to comprehend. Large models are an incredibly resource-intensive endeavor, necessitating astronomical investments in computing power, data, and talent. If each business unit operates independently and reinvents the wheel, it will not only be inefficient but also result in substantial resource wastage.

Commercialization endeavors require not only technical prowess but also group-wide resource coordination, cross-business unit collaboration, budget allocation, and the opening up of business scenarios—all of which can only be driven at the CEO level.

From this vantage point, reassigning Zhou Jingren to the Frontier Research Institute appears more like a return to specialized division of labor. Entrusting mature business lines to a CEO skilled in coordination to lead the commercialization battle, thereby freeing top technical talent from daily management and short-term KPIs to tackle the formidable challenges of next-generation technology.

This is also a prevalent organizational strategy among major AI companies. Google has DeepMind focusing on frontier research, while OpenAI has a Superalignment team exploring long-term directions. The conventional path for global leading tech companies is to gradually free top technical talent from daily operations to invest in longer-term technological reserves.

This also seems more aligned with Zhou Jingren's own technological philosophy. During a group interview following the 2025 Alibaba Cloud Yunqi Conference, when queried about the core elements of achieving ASI, he stated, "The upper bounds of AI models have yet to be reached. To truly move towards ASI, we must continue to make breakthroughs in technical pathways, addressing a series of underlying challenges at the architectural, system, and algorithmic levels."

Image source: Alibaba Cloud's official website

Since 2026, major domestic companies have also been transitioning from laboratory-based models to industrialized approaches. Key members of ByteDance's Seed team have been reassigned, multiple major companies have downsized their AI labs to focus on business implementation, and startups have shifted from competing on leaderboards to competing on commercialization.

This brings us back to the most fundamental question. Alibaba undoubtedly requires Zhou Jingren, but the nature of its need for him has fundamentally evolved.

Alibaba needs Zhou Jingren's technical reputation more than ever to maintain its standing at the AI table, but it no longer requires those capabilities that once made him renowned. In a company like Alibaba, which has transitioned from a technology-first to a business-first approach, there has never been a place for a technical idealist at the center of the table.

For Alibaba and Zhou Jingren, whether their decade-long partnership will extend into another decade is merely the tip of the iceberg to outsiders. What truly determines the outcome has always been the direction beneath the surface.

(The header image for this article is sourced from the Alibaba Cloud WeChat official account.)