Alibaba Considers Revenue Sharing for Qwen3.8-Max, asking major commercial users to share earnings generated from the open-weight AI model.

Alibaba considers revenue sharing for Qwen3.8-Max to offset AI development costs

Alibaba revenue sharing for Qwen3.8-Max could introduce a significant change to the business model surrounding open-weight artificial intelligence. The Chinese technology company is reportedly considering asking major commercial users to share a portion of the revenue they generate from applications built with its next Qwen model. The proposal reflects a growing challenge for AI companies: developing frontier models is becoming increasingly expensive, while open access can make it difficult to recover those costs directly.

According to Reuters, Alibaba is discussing the plan with large users, but the exact percentage has not been finalized. The company’s strategy could apply to commercial organizations that download the model and operate it on their own infrastructure rather than accessing it through Alibaba Cloud.

A shift from Alibaba’s current model

Alibaba has generally followed a dual-track approach with Qwen. Developers and businesses can pay to use its models through Alibaba Cloud, while many open-source or open-weight releases can be downloaded and run inside a customer’s own data center without a direct licensing fee.

A revenue-sharing requirement would change that arrangement for some large commercial users. Instead of charging only for cloud-hosted access, Alibaba could seek a percentage of the income earned from products and services powered by Qwen3.8-Max.

The proposal remains under discussion, and Alibaba has not publicly confirmed a final licensing rate. That uncertainty will be closely watched by developers, cloud providers, and companies building AI products around open models.

Qwen3.8-Max raises the stakes

Alibaba unveiled Qwen3.8-Max on August 3 as its most capable Qwen model so far. Reports describe it as a mixture-of-experts system with 2.4 trillion total parameters, approximately 95 billion active parameters, and a context window reaching up to 1 million tokens.

The model is available through Alibaba Cloud’s Model Studio and QwenCloud. Its listed API pricing is approximately $2 per million input tokens and $6 per million output tokens, while cached input can cost considerably less. Alibaba has also announced plans to release open weights for Qwen3.8-Max, alongside a smaller Qwen3.8-27B model, through platforms including Hugging Face and ModelScope.

That combination — frontier-level capability, competitive API pricing, and downloadable weights — could make Qwen3.8-Max attractive to companies that want more control over data, deployment, and operating costs.

Why Alibaba is considering revenue sharing

Training and maintaining advanced AI models requires enormous investment in computing hardware, data centers, energy, engineering talent, and ongoing model improvements. Open-weight distribution can expand adoption quickly, but it may also allow successful businesses to monetize the technology without contributing directly to the model developer’s costs.

Alibaba’s reported proposal appears designed to address that gap. If a company turns Qwen3.8-Max into a profitable software product, an enterprise platform, or a consumer service, Alibaba could receive compensation linked to the commercial success of that application.

The approach resembles a broader industry debate over how open AI should be licensed. Moonshot AI’s Kimi K3 reportedly uses a commercial arrangement that can require partners to share as much as 30% of revenue, although Alibaba’s potential rate remains unclear.

The licensing challenge

Revenue sharing could give Alibaba a new way to fund AI research, but it may also make Qwen3.8-Max less attractive to some businesses. Companies often choose open-weight models because they want predictable costs, flexible deployment, and fewer restrictions. A percentage-based fee could complicate financial planning and encourage some developers to choose models with more permissive licenses.

Alibaba will therefore need to balance monetization with adoption. A strict commercial license might generate more money from major customers, but a complicated agreement could slow the growth of the Qwen ecosystem.

For now, the proposal is not a completed policy. Still, it highlights an important shift in the AI market: open models may be freely downloadable, but the companies behind them are searching for new ways to participate in the value created by commercial deployments.

Summary: Alibaba is reportedly considering revenue sharing for businesses that use Qwen3.8-Max to build profitable products. The plan could help offset the rising cost of AI development, but its success will depend on whether Alibaba can generate revenue without weakening the openness and adoption that have helped Qwen gain attention.