Goldman Sachs Sees Chinese AI Labs Shifting to Paid Model Weights for Revenue

Goldman Sachs analysts, led by Ronald Keung, anticipate that Chinese AI developers such as Moonshot AI and Zhipu AI will transition from freely distributing model weights to imposing commercial licensing fees on cloud providers that host these models, a shift termed "paid weights." This move aims to capture more economic value from the widespread adoption of their high‑performing, low‑cost models, which have driven significant usage by companies seeking cheaper alternatives to U.S. offerings. While openness helped labs build global brands and offload inference costs, it generated little direct revenue, and rising training expenses and investor pressure now motivate monetization. The report forecasts China’s AI model API and subscription revenue growing from ~35 billion RMB in 2026 to 879 billion RMB by 2035, envisioning hybrid arrangements where community editions remain free but enterprise deployments require revenue‑sharing. Risks include developers forking the last fully open versions to avoid fees, potentially undermining the labs’ influence, and geopolitical scrutiny over technology transfer. Ultimately, Chinese AI firms must balance openness with sustainable business models as global markets weigh performance against cost.

(Source：Webpronews)