Key Takeaways:
Chinese AI developers are moving to charge cloud platforms for hosting open-weight models, a shift that could unlock billions in new revenue and narrow the gap with US rivals.
Chinese AI developers are preparing to charge cloud platforms for hosting their open-weight models, seeking to capture more revenue from surging global adoption, Goldman Sachs Asia internet research head Ronald Keung said.
"Chinese model makers could generate more revenue by requiring third-party vendors to purchase commercial licenses to provide inference services for these models on their own infrastructure," Keung said in an interview with the South China Morning Post.
The assessment comes as Chinese models — including Moonshot AI's Kimi K3 and Z.AI's GLM-5.2 — have reached performance levels only slightly behind leading US competitors. However, because Chinese developers typically release models under open-source terms, foreign platforms can freely download and host the core "weights" — the underlying parameters that encode the model's intelligence. Adoption has been rapid among small and midsize enterprises globally, and larger companies have started evaluating Chinese models, Keung said.
The licensing shift could significantly boost revenue for Chinese AI developers, potentially improving valuations in a sector where monetization has lagged behind US peers. Z.AI (02513.HK) shares fell 17.174% on July 28, with short selling reaching $457.93 million, or 8.213% of turnover, reflecting market uncertainty about competitive dynamics. Haitong International raised its price target on Z.AI to HKD2,300, while JPMorgan cut its target to HKD1,600, showing divergent views on the company's prospects.
The move toward paid licensing represents a strategic pivot for Chinese AI developers that have relied on open-source distribution to drive adoption. By charging cloud platforms for inference services — the computational process of running trained models on new data — developers could create a recurring revenue stream tied to usage volume.
Global cloud platforms including Amazon Web Services, Microsoft Azure and Google Cloud have hosted Chinese open-weight models at no cost, using them to attract developer traffic to their AI services. A licensing fee structure would force these platforms to either absorb the cost or pass it to end users, potentially reshaping the economics of AI model deployment. For US cloud providers, the added cost could accelerate their push toward proprietary models as an alternative to hosting Chinese open-weight alternatives.
For Z.AI, which trades on the Hong Kong Stock Exchange, the licensing debate comes at a critical juncture. The company's GLM-5.2 model has been benchmarked near US frontier models on key metrics, but the stock's 17% decline suggests investors remain uncertain about the path to profitability. The divergent analyst targets — HKD2,300 from Haitong versus HKD1,600 from JPMorgan — imply a potential upside of more than 40% from current levels if the licensing strategy succeeds. Z.AI trades at a discount to US AI peers, reflecting the market's skepticism about whether Chinese AI developers can translate technical parity into sustainable revenue.
This article is for informational purposes only and does not constitute investment advice.
