Alibaba plans to ask major users of its upcoming Qwen 3.8-Max open-source AI model to share a portio...
The AMW Read
Alibaba's revenue-sharing plan for Qwen marks a strategic shift in open-weight monetization, updating its position in the foundation model landscape with potential industry-wide implications.
Alibaba plans to ask major users of its upcoming Qwen 3.8-Max open-source AI model to share a portion of the revenue they generate from products and services built using the technology, according to Reuters. The new model is open weight, meaning developers can access and modify its core settings, but unlike previous Qwen releases, it will not be entirely free for large commercial users. Alibaba intends to introduce licensing terms that require big businesses to negotiate commercial agreements and share revenue, though the exact percentage has not been decided. The measure could be announced as early as next week.
This move represents a notable shift for Alibaba, which previously charged only for models running on its own cloud platform, allowing most open-source releases to be used free in customer data centers. The strategy mirrors that of Chinese AI startup Moonshot, whose Kimi K3 model requires any reseller generating more than $20 million annually to strike a paid deal, with Moonshot able to seek up to 30% of revenue. The revenue-sharing approach could help Alibaba monetize its open-weight models while still attracting developers who favor flexibility over closed systems like those from OpenAI, Anthropic, and Google. It also signals a broader trend among Chinese AI labs to balance open-source adoption with sustainable commercialization.
For builders and investors, this development highlights the evolving economics of open-weight AI models. While open-weight models offer lower costs and greater control, they are not necessarily free for large-scale commercial use. Enterprises planning to build services on Qwen 3.8-Max should anticipate potential licensing fees and revenue-share obligations. Investors should watch whether Alibaba’s revenue-sharing model becomes a template for other Chinese labs, potentially reshaping the competitive landscape against U.S. closed models that often come with per-token pricing. The move could also intensify pricing pressure on American AI developers as Chinese models like Kimi K3 are already priced at roughly one-third of comparable U.S. offerings.
