Alibaba to Charge Big Users for Next Open-Source AI Model

Chinese artificial intelligence laboratories are implementing commercial revenue-sharing models for open-source AI releases, led by Moonshot and Alibaba, as they push to capture market share from U.S. competitors like OpenAI, Anthropic, and Alphabet’s Google according to recent industry reports.

Moonshot and Alibaba Push for Open-Source Revenue Sharing

Chinese technology giant Alibaba plans to ask major users of its upcoming Qwen open-source AI model for a share of revenue generated from the offering, according to two people familiar with the plans. This follows a similar blockbuster move by Chinese AI startup Moonshot, whose Kimi K3 license requires developers offering the model as a service and generating more than $20 million in annual sales to negotiate a commercial agreement.

While open-source software is traditionally regarded as free, these updated licensing provisions mark a strategic shift. According to sources familiar with the strategy, Moonshot’s licensing terms require partners to share up to 30% of revenue. Alibaba plans to implement a comparable measure for its open-source model, though exact rates remain under discussion.

Pricing and Technical Economics of Chinese AI Models

The financial structures emerge alongside aggressive pricing strategies that undercut U.S. competitors. Kimi K3 costs roughly one-third of Anthropic’s Fable model based on listed input and output token prices, according to market data. Dan Fu, vice president of kernels at Together AI, noted that firms specializing in AI infrastructure generate revenue by optimizing service layers and token utilization rather than charging solely for raw model access.

Cloud computing providers are actively adapting to these commercial terms. Paddy Srinivasan, CEO of DigitalOcean Holdings—one of several U.S. firms offering Kimi K3—confirmed his company maintains a commercial agreement with Moonshot. Srinivasan described the arrangement as a tried-and-tested freemium model where clients pay for deployment optimization and early access to model revisions.

Did You Know?
Open-source models distribute underlying learned settings for download, allowing developers to adapt the systems locally. In contrast, U.S. developers like OpenAI, Anthropic, and Google primarily maintain closed-source architectures.

Global Expansion and Regulatory Pressures

The monetization pivot unfolds amid geopolitical friction. The White House has previously accused Moonshot of misappropriating technology from Anthropic, an allegation that Chinese officials have firmly rejected as unfounded. Despite these tensions, domestic IT services providers are openly disclosing partnerships.

Chinasoft International disclosed a revenue-sharing agreement with Moonshot in a regulatory filing last month, omitting the specific percentage. Meanwhile, U.S. startups are entering the open-source space. Thinking Machines Lab, founded by former OpenAI Chief Technology Officer Mira Murati, released its first open-source model, signaling broader industry movement toward open-weight architectures.

Frequently Asked Questions

What is an open-source AI model?

An open-source or open-weight AI model provides the underlying learned settings for download, enabling developers to run, modify, and adapt the system independently.

How much do Chinese AI revenue-sharing agreements cost?

According to sources familiar with the plans, Moonshot requires up to a 30% revenue share for partners generating over $20 million in annual sales using Kimi K3.

Are U.S. companies offering these Chinese models?

Yes. U.S. cloud providers and infrastructure firms, including DigitalOcean and Together AI, offer access to models like Kimi K3 under established commercial agreements.

Stay Updated on AI Industry Trends

Subscribe to our newsletter for the latest verified reports on artificial intelligence infrastructure, global market competition, and enterprise technology shifts.

Leave a Comment