Goldman Sachs says China's AI firms are finding ways to monetize even as their models stay open

12:00 AM · July 20, 2026

Goldman Sachs analyst Ronald Keung told CNBC that Chinese large language models are reaching a critical stage of global adoption, with rising API usage creating a feedback loop in which growing developer adoption improves the underlying models while an expanding developer ecosystem in turn drives further usage. Keung's central argument is that releasing a model's weights openly does not mean a company has no path to revenue, since Chinese AI firms are finding other ways to monetize their technology, including commercial deployment contracts with enterprises, licensing arrangements, and revenue sharing structures built around how the models are actually used in production rather than charging directly for access to the model itself. The analysis pushes back against a common assumption that open weight models are inherently a weaker commercial proposition than closed, subscription based ones, arguing instead that the business model is simply different, oriented around services and deployment rather than access itself. Coming from one of Wall Street's most closely watched voices on the AI sector, the framing is notable at a moment when investors are actively trying to determine whether Chinese labs like Moonshot and DeepSeek represent a genuine long term commercial threat to companies like OpenAI and Anthropic or whether their open weight strategy ultimately limits how much revenue they can realistically capture from their own technical advances.