Alibaba is preparing to introduce a revenue-sharing licensing model for its upcoming open-weight AI model, Qwen3.8-Max, marking a significant shift in how the Chinese technology giant monetizes its open-source artificial intelligence ecosystem. Under the proposed framework, developers will continue to be able to download and modify the model’s weights, but companies generating substantial revenue by offering Qwen3.8-Max as a commercial service will be required to enter revenue-sharing agreements with Alibaba. The move is aimed at helping offset the rapidly rising costs of developing frontier AI models while preserving the benefits of open-weight distribution.
The strategy mirrors a growing trend among Chinese AI developers to adopt a “freemium” open-source model, where basic access remains free while large-scale commercial deployments become a source of recurring revenue. Alibaba’s decision follows a similar licensing approach introduced by AI startup Moonshot for its Kimi K3 model, highlighting an industry-wide effort to balance open innovation with sustainable business models amid soaring AI infrastructure costs.
Alibaba Plans Revenue Sharing for Qwen3.8-Max
According to people familiar with the company’s plans:
- Qwen3.8-Max will remain an open-weight AI model.
- Developers will continue to download, modify, and deploy the model.
- Large commercial users generating significant revenue from hosted AI services will be required to negotiate commercial licensing terms with Alibaba.
- The policy is expected to accompany the official release of Qwen3.8-Max.
Policy Snapshot
| Item | Details |
|---|---|
| Company | Alibaba |
| Model | Qwen3.8-Max |
| Distribution | Open-weight |
| New Monetization | Revenue-sharing for large commercial providers |
| Objective | Recover AI development and infrastructure costs |
Why Alibaba Is Changing Its Licensing Model
Until now, Alibaba primarily generated revenue from Qwen models when customers used them through Alibaba Cloud. Developers who downloaded open-weight models and deployed them in their own data centers generally did not pay licensing fees.
The proposed change expands Alibaba’s monetization strategy by allowing it to earn revenue even when Qwen is hosted outside Alibaba Cloud.
The move reflects several industry trends:
- Rapidly increasing AI model training costs.
- Growing demand for open-weight foundation models.
- Rising commercial adoption by cloud providers and AI startups.
- Pressure to build sustainable revenue streams beyond cloud hosting.
Similar to Moonshot’s Licensing Strategy
Alibaba’s approach closely resembles the licensing framework introduced by Chinese AI startup Moonshot for its Kimi K3 model.
According to reports:
- Moonshot requires commercial providers generating more than $20 million in annual revenue from hosted Kimi services to negotiate licensing agreements.
- Some agreements reportedly include revenue-sharing of up to 30%.
This signals a broader shift in China’s open-source AI ecosystem, where companies are increasingly monetizing commercial usage while keeping model weights publicly accessible.
Open-Weight vs Traditional Licensing
| Traditional Open Source | Alibaba’s Proposed Model |
|---|---|
| Free commercial deployment | Free for most users; revenue-sharing for major commercial providers |
| Revenue mainly from cloud APIs | Revenue from cloud services and commercial licensing |
| Few restrictions | Commercial terms for large-scale hosted services |
Open-Weight Does Not Mean Completely Free
The proposed licensing model illustrates an important distinction between open-weight AI and unrestricted commercial use.
Under an open-weight approach:
- Model parameters remain publicly available.
- Developers can inspect and fine-tune the model.
- Research and experimentation remain largely unrestricted.
- Large commercial hosting providers may face additional licensing obligations.
This differs from closed-source models offered by companies such as OpenAI and Anthropic, where model weights are not publicly released.
Growing Focus on Sustainable AI Economics
Training frontier AI models now requires enormous investments in:
- High-performance GPUs.
- Data center infrastructure.
- Large-scale datasets.
- Research talent.
- Ongoing inference capacity.
As these costs continue to rise, AI companies are exploring new monetization strategies that allow them to maintain open ecosystems without sacrificing long-term financial sustainability.
What It Means for Developers
For most developers, researchers, and startups, the proposed licensing framework is unlikely to change day-to-day usage.
However, companies that:
- Host Qwen as a commercial API,
- Sell AI services built directly on Qwen,
- Or generate significant recurring revenue from the model,
could be required to negotiate commercial licensing agreements with Alibaba.
Looking Ahead
Alibaba’s proposed revenue-sharing model for Qwen3.8-Max represents an important evolution in the economics of open-weight artificial intelligence. Rather than abandoning open distribution, the company is seeking to preserve developer access while ensuring that businesses generating substantial commercial value from its models contribute to the significant costs of building and maintaining frontier AI systems. If implemented, the approach could offer a middle ground between fully open-source software and closed, API-only AI services.
Looking ahead, Alibaba’s strategy could influence how other AI developers monetize open-weight models as development costs continue to climb. If successful, revenue-sharing agreements may become a common feature of the AI industry, allowing companies to expand open innovation while creating sustainable funding for next-generation model research and infrastructure.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
