Alibaba shifts Qwen strategy: from open source to paid AI services

Alibaba is changing how it develops Qwen: the company no longer wants to compete solely on the quality of open-source models, and is speeding up their move into paid cloud services. The Qwen family is not formally going away, but inside Alibaba the priority is shifting from the research race and benchmarks to products that can generate revenue now.

The turn toward monetization

In the first years of the generative AI boom, big companies readily invested in scaling models, even when the direct economics stayed murky. Alibaba now has a different focus: management wants the AI business to work harder for the cloud business and to stop being a showcase for technological achievements.

Alibaba chief Eddie Wu publicly promotes the MaaS model, in which customers pay for model use based on actual load. The problem is that, for all of Qwen's fame, this format does not yet bring Alibaba significant revenue, so the company needs a tougher commercial framework.

Team reshuffle

Two prominent figures who largely defined the earlier course have already left the Qwen team: Lin Junyang, the former Qwen technical director, and Hui Binyuan, a researcher who specialized in code generation and writing. These were the people associated with Qwen's engineering direction and its reputation among developers.

The changes look less like routine staff rotation and more like a signal that Alibaba is rebuilding its priorities around a more applied, profit-oriented strategy.

The new chain of command is built for the same purpose. Eddie Wu personally chairs the AI strategy committee, while day-to-day leadership passes to Zhou Jingren, formerly technical director of Alibaba Cloud. His main task is to tie model development to cloud business metrics. It is a notable shift: the team used to spend much of its attention on model speed and results in industry tests, but now it is expected to deliver not only strong science but a clear commercial result.

What is changing in Qwen

Alibaba is not abandoning AI altogether and is not burying Qwen as a brand, but it is changing the packaging and the business model. The company has already begun releasing closed models, which customers access through Alibaba Cloud infrastructure. It does not intend to abandon open source entirely: some models will stay open for specific use cases and ecosystem needs.

In other words, instead of its earlier bet almost exclusively on developer popularity, Alibaba is building a hybrid scheme: open models for reach, closed ones for direct monetization.

This is a difficult turn for Alibaba precisely because Qwen gained weight through open distribution. Open models helped the company build a broad developer community and strengthen its reputation in the global AI scene, but popularity alone does not guarantee cash flow. The more the ecosystem got used to seeing Qwen as an accessible tool, the harder it becomes to explain what exactly a customer should pay for in the cloud. Hence the bet on closed versions, service scenarios and tighter ties to Alibaba Cloud infrastructure.

• the Token Hub unit has been launched to speed up AI commercialization
• Eddie Wu personally oversees the new AI strategy committee
• Zhou Jingren is responsible for linking models to cloud revenue
• Alibaba has already introduced several closed-source models
• the open-source lineup will remain for some use cases

What it means

For the market, this is one more sign that the "growth first, money later" era of generative AI is ending. Even companies with strong open-source models now have to prove that their work can bring stable revenue through the cloud, APIs and agent services.

The turn matters especially for Alibaba: it already has a recognizable AI brand and a large developer base, but the next stage will depend not on the number of Qwen downloads but on how well the company turns interest in the models into a paying business.