Alibaba plans to require major commercial users of its upcoming Qwen3.8-Max model to share a portion of the revenue they generate from deploying it, according to two people familiar with the company’s plans cited by Reuters. The move follows an identical strategy by Moonshot AI, whose Kimi K3 licensing terms require companies generating more than $20 million in annual sales from the model to negotiate commercial agreements with revenue shares reaching up to 30%.
The model itself remains open-source and open-weight, meaning developers can download, run, and adapt the underlying learned settings without restriction. That distinction matters because “open source” has long been understood as synonymous with free. Alibaba and Moonshot are now proving it doesn’t have to be, and the implications for the global AI ecosystem are significant.
“You pay for collaboration with these open-weight model labs to make sure that you’re optimizing your deployment. You pay for getting early access for the next revision of the model,” said Paddy Srinivasan, CEO of cloud computing firm DigitalOcean Holdings, one of several U.S. firms offering Kimi K3 and other Chinese models.
Until now, Alibaba charged developers who used Qwen models through Alibaba Cloud’s inference API while generally allowing customers to deploy open-source models in their own data centers without payment. The new licensing terms change that equation for large-scale commercial deployments. Small developers and researchers remain unaffected, but any company building a significant revenue-generating product on Qwen3.8-Max will need to negotiate terms with Alibaba directly.
The strategy mirrors Silicon Valley’s proven playbook: offer software at zero cost to build adoption and ecosystem dependency, then monetize heavy commercial usage once the model becomes embedded in production systems. Alibaba’s Qwen models have been downloaded 700 million times on Hugging Face, making them the world’s most popular open-source AI system. That massive distribution creates leverage. Once thousands of companies have built products on Qwen, switching costs make revenue-sharing terms commercially rational even if competitors remain technically free.
Chinasoft International, a Chinese IT services provider, disclosed a revenue-sharing agreement with Moonshot in a regulatory filing last month without specifying the percentage, providing the first public confirmation that these arrangements are already being formalized at the enterprise level.
The competitive dynamics are shifting rapidly. Chinese AI developers are adopting freemium business models as they compete with US leaders including OpenAI, Anthropic, and Google, all of which operate closed-source models with direct API pricing. Meta’s Llama remains the primary Western open-source competitor, and its licensing currently does not include revenue-sharing provisions. Whether Meta follows Alibaba and Moonshot’s lead or maintains its current terms could determine which model family dominates enterprise AI deployments globally.
The revenue-sharing model also raises questions about what “open source” means in practice. The Open Source Initiative’s definition emphasizes freedom to use, modify, and distribute software without royalty obligations. Revenue-sharing requirements for commercial deployment arguably violate that principle, placing Qwen and Kimi K3 in a gray zone between truly open-source and proprietary licensing. Developers who chose these models specifically because they were free now face the prospect of renegotiating their economics as their products scale.
Alibaba has not publicly disclosed the planned revenue-sharing percentage, and discussions remain ongoing. The Qwen3.8-Max release is expected next week, meaning the licensing terms will become concrete almost immediately.
