Alibaba plans revenue sharing for major users of Qwen AI model

### Alibaba Plans Revenue-Sharing Terms for Commercial Users of New Qwen AI Model

Alibaba is preparing to introduce a new commercial licensing approach for its upcoming Qwen 3.8-Max artificial intelligence model, potentially requiring major businesses to share a portion of the revenue generated from products and services built using the technology. The planned policy represents a significant shift in how Alibaba approaches its open-source and open-weight AI models, which have traditionally been available for companies to deploy in their own environments with relatively few commercial restrictions.

Qwen 3.8-Max is expected to be released with its core model settings available for developers to access, modify and deploy, giving businesses greater flexibility than they typically receive from closed AI systems operated by companies such as OpenAI, Anthropic and Google. However, Alibaba’s approach highlights an important distinction between open-weight technology and completely unrestricted commercial use. Under the proposed licensing terms, large companies that build commercial products around the model could be required to negotiate separate agreements with Alibaba and potentially provide the company with a share of the revenue they generate.

The exact percentage Alibaba intends to request has not yet been determined, according to people familiar with the plans. The company is expected to announce the new licensing arrangements soon, although the details had not been publicly confirmed at the time of the reports. The move follows a similar strategy adopted by Chinese AI startup Moonshot AI for its Kimi K3 model. Moonshot’s licensing terms reportedly require companies that resell the model as a service and generate more than $20 million in annual revenue to enter into a paid commercial agreement.

Under certain arrangements, Moonshot can reportedly seek a revenue share of up to 30%. Alibaba’s planned strategy could provide a new way for AI developers to monetize open models while continuing to make the underlying technology broadly accessible to developers and smaller users. In the past, Alibaba generally charged customers when its AI models were operated through its cloud platform, while many of its open-source releases could be deployed directly within customer-owned data centers without additional licensing fees.

The emerging licensing model reflects the growing pressure on AI companies to generate sustainable revenue as the cost of developing and training advanced models continues to rise. Open-weight AI developers are increasingly exploring approaches that combine free or low-cost access with commercial fees for large-scale deployments, specialized support, early access and other premium services. DigitalOcean CEO Paddy Srinivasan has described this approach as a form of open-source “freemium” business strategy, in which basic access remains widely available while companies pay for additional commercial benefits.

Alibaba’s plans also come amid increasing competition between Chinese and U.S. AI developers. Chinese companies have been aggressively developing lower-cost alternatives to leading American models, creating additional pressure on established U.S. AI laboratories. Moonshot’s Kimi K3, for example, has been positioned as a significantly cheaper alternative to comparable American AI models, potentially making it attractive to businesses seeking to reduce AI infrastructure and model usage costs. The competition has also generated controversy over how Chinese AI companies develop their models.

U.S. officials have accused Moonshot of using techniques associated with model distillation to learn from Anthropic’s advanced AI technology during the development of Kimi K3. American officials have described such alleged activity as a threat to proprietary AI research, while Chinese authorities have rejected the accusations as unfounded and defended the country’s AI development efforts. The dispute illustrates the broader geopolitical competition surrounding artificial intelligence, where model performance, pricing, computing resources and intellectual property have become increasingly important.

Alibaba’s new licensing strategy could therefore have implications beyond its own Qwen ecosystem. If major businesses are required to share revenue from commercial products built using open-weight models, other AI developers could consider similar approaches to monetize widely distributed technologies. At the same time, businesses may need to weigh the flexibility and lower costs of open models against potentially more complicated licensing obligations.

Alibaba’s decision ultimately reflects the changing economics of artificial intelligence, where making a model openly available does not necessarily mean giving up the opportunity to generate commercial returns from its widespread adoption.