Alibaba is changing its AI monetization strategy by introducing a revenue-share model for its upcoming open-weight models, moving beyond the industry standard of using such releases as loss leaders for cloud infrastructure. With the release of Qwen3.8 expected around August 10, 2026, the company aims to formalize a monetization layer targeting deployment rather than relying solely on API access. Previously, Alibaba charged for cloud platform access while offering self-hosted deployments without additional licensing fees. The Qwen3.8-Max API recently reached general availability with pricing at $2 and $6 per million tokens, maintaining parity with OpenAI’s GPT-5.6. The new revenue-share requirement for major commercial users of the open-weight version treats model weights as a royalty-bearing asset class. This strategy follows a precedent set by Moonshot AI, whose Kimi K3 model requires companies exceeding $20 million in annual sales to enter a commercial agreement, with revenue-share rates reaching as high as 30%. Alibaba has not finalized its specific rate, with negotiations ongoing. The competitive landscape includes alternatives without such financial burdens: DeepSeek uses a custom, royalty-free license that is perpetual and irrevocable, while Meta’s Llama Community License allows free commercial use for entities with fewer than 700 million monthly active users, requiring a separate license only for the largest tech conglomerates. This creates a three-tier licensing environment: royalty-free options, conditional-free models, and the emerging