Alibaba Qwen Tech Lead Junyang Lin Steps Down Amid Leadership Exodus

A Sudden Farewell to Qwen

On the afternoon of March 3, 2026, Junyang Lin, the technical leader of Alibaba's Qwen large language model project, submitted his resignation. Hours later, he posted a brief message on X: "me stepping down. bye my beloved qwen." The announcement sent shockwaves through the global AI community and triggered a 5.3% drop in Alibaba's Hong Kong-listed shares, the company's largest intraday decline since October.

The timing was striking. Lin's departure came just one day after Alibaba unveiled its Qwen 3.5 open-weight small models, a release he had been working on until the final hours. His exit was not an isolated event but the culmination of a broader leadership upheaval within Alibaba's AI division that raises questions about the future direction of one of China's most important open-source AI projects.

Who Is Junyang Lin

Junyang Lin joined Alibaba in July 2019 and became part of the Qwen team when it was established in April 2023. Over the next three years, he grew into the project's central technical figure, overseeing the development of the Qwen model family from its earliest iterations through to the Qwen 3.5 series.

His contributions extended beyond technical development. Lin was instrumental in connecting Qwen with the global developer community, building partnerships and ensuring that Qwen's open-weight releases reached international audiences. Hyperbolic CTO Yuchen Jin noted that Lin played a key role in bridging Qwen's technology with developers worldwide.

Wenting Zhao, a Qwen research scientist, described Lin's departure as "the end of an era" and acknowledged his contributions to advancing open-source AI work. According to sources cited by multiple outlets, some members of the Qwen team were deeply emotional upon hearing the news, with one colleague reportedly breaking down in tears.

The Restructuring Behind the Resignation

Lin's resignation was voluntary, but it was not unprompted. According to reporting from Geopolitechs and other outlets, the departure stemmed from Alibaba's plan to dismantle Qwen's vertically integrated team structure.

Under Lin's leadership, Qwen operated as a cohesive unit where pre-training, post-training, infrastructure, and training teams worked closely together. Lin's philosophy was that tight integration across these functions produced better models. Alibaba's leadership, however, decided to split these teams functionally and merge them with other divisions under the broader Tongyi Lab umbrella.

This organizational restructuring directly contradicted Lin's approach to AI development. Rather than accept a reorganization he disagreed with, Lin chose to leave. His resignation prompted an emergency all-hands meeting where Alibaba CEO Eddie Wu acknowledged: "I should have known about this earlier."

A Pattern of Departures

Lin's exit was not an isolated incident. He became the third senior leader to leave Alibaba's AI division in the first quarter of 2026:

Yu Bowen, the executive responsible for Qwen's post-training operations, left on the same day as Lin. Huibin, who led the Qwen Code team, had already departed in January 2026 to join Meta. The concentration of departures at the senior level suggests systemic organizational tensions rather than individual career decisions.

To fill the gap in post-training leadership, Alibaba brought in Zhou Hao, a former DeepMind Senior Staff Researcher, though his exact reporting structure remained under discussion at the time of the announcement.

Implications for Qwen Development

Qwen has established itself as one of the most significant open-source AI model families outside the United States. The Qwen 3.5 series, released just before Lin's departure, achieved benchmark results comparable to leading American AI systems. The project's open-weight releases under permissive licenses have made it a popular choice for developers and companies seeking alternatives to closed models from OpenAI, Google, and Anthropic.

The loss of three senior technical leaders in a single quarter creates real continuity risk. Pre-training, post-training, and code generation represent three of the most critical capabilities in modern LLM development. Losing the leaders of all three functions within months of each other could slow development velocity and disrupt the institutional knowledge that drives model improvements.

However, Alibaba's broader AI commitment remains substantial. The company has invested heavily in AI infrastructure and recently committed significant resources to competing with American AI labs. The question is whether the organizational restructuring that prompted these departures will ultimately produce a more efficient development process or whether it will fragment the cohesion that made Qwen competitive.

The Broader Context: China's AI Talent Competition

Lin's departure also reflects the intensifying competition for AI talent in China and globally. With Huibin joining Meta, the talent flow is not limited to movement within Chinese companies but extends to international competitors.

Lin had previously warned about the gap between Chinese AI labs and OpenAI, a candid assessment that drew attention from both the industry and investors. His willingness to acknowledge competitive realities made him a respected voice in the global AI community, and his departure removes that perspective from Alibaba's internal discussions.

The AI talent market in China has become increasingly competitive as ByteDance, Baidu, and numerous startups like DeepSeek and Moonshot AI all compete for the same limited pool of experienced researchers and engineers. Alibaba's organizational decisions may have inadvertently pushed top talent toward competitors at precisely the moment when retention matters most.

Pros

• Qwen 3.5 open-weight models were successfully released before the leadership transition
• Alibaba brought in Zhou Hao from DeepMind to manage post-training, bringing fresh external perspective
• The restructuring may produce more specialized and efficient teams in the long term
• Alibaba's overall AI investment commitment has not changed

Cons

• Three senior technical leaders departed in a single quarter, creating significant knowledge gaps
• The restructuring contradicts the integrated team approach that produced Qwen's competitive results
• Alibaba's shares dropped 5.3% on the news, reflecting investor concern about AI strategy execution
• The departures may signal deeper organizational tensions that could affect future talent recruitment

Outlook

The immediate challenge for Alibaba is stabilizing the Qwen team and maintaining development momentum. The Qwen 3.5 release demonstrates that the technical foundation remains strong, but sustaining that trajectory without the leaders who built it will test the organization's depth.

For the broader AI industry, the Qwen leadership exodus is a reminder that organizational decisions matter as much as technical ones. The best models come from teams that function well, and restructuring those teams carries risks that balance sheets do not capture. Whether Alibaba's reorganization proves to be a necessary evolution or a strategic misstep will become clear in the coming quarters as the next generation of Qwen models takes shape under new leadership.

Conclusion

Junyang Lin's departure from Alibaba Qwen marks more than one executive's career decision. It reflects a fundamental disagreement about how AI teams should be organized and managed at a moment when China's AI ambitions demand maximum execution capability. With three senior leaders gone in three months and a major organizational restructuring underway, Alibaba faces the task of proving that its new approach can produce results equal to what the departing team achieved. The global AI community will be watching closely.