Alibaba plans to introduce a revenue-sharing mechanism next week for major commercial users of its open-source AI model Qwen3.8-Max. Companies generating more than US$20 million annually from the model will be required to share a portion of their revenue with Alibaba [1, 2, 3, 4].
Qwen3.8-Max is an open-weight model that developers can freely download and modify, similar to the Kimi K3 model launched by Chinese startup Moonshot in July. Moonshot demands commercial users exceeding the same $20 million revenue threshold to share up to 30% of income [1, 2, 3].
Alibaba has historically charged fees only when AI models were hosted on its cloud platform. Customers running the models on-premise in their own data centers generally paid no fees [1, 3, 4]. The company’s forthcoming revenue-share policy marks a shift toward charging based on commercial scale rather than hosting alone.
Chinese AI firms are adopting a business model closer to that of Silicon Valley labs, providing open-source models free or at low cost initially, then monetizing large-scale usage and premium services. US firms like DigitalOcean have commercial deals with Chinese open-source AI developers such as Moonshot, reflecting this trend. DigitalOcean’s CEO said this model "has been proven in practice" as a way to add value to free open-source AI [1, 2, 3, 4].
The exact revenue-sharing ratio Alibaba will impose has not been finalized and remains under discussion [1, 2, 3]. DigitalOcean’s CEO Srinivasan added that companies pay these AI labs to optimize deployment and gain priority access to next-generation models.
Alibaba aims to roll out the revenue-sharing policy for Qwen3.8-Max around August 14, one week from today [1, 2, 3].
