China's 'open' AI is a terrible business, and nothing like open-source software - Business Insider

What the report says

Business Insider reported that China’s fast-rising “open-weight” AI labs are drawing developer attention but showing weak business fundamentals, arguing that their model differs sharply from the economics of open-source software. The article focuses on Z.ai, also known as Zhipu, whose executives recently attended a Hong Kong Stock Exchange listing ceremony. BI said Zhipu, the company behind the GLM 5.2 open-weight model, lost nearly $500 million last year on about $107 million in revenue, while its shares fell more than 40% over the past month.

BI also cited MiniMax, another publicly traded Chinese AI lab, which it said lost $250 million last year on $79 million in revenue and saw its shares drop more than 50% in the past month. The report contrasts these companies with traditional open-source software businesses such as Red Hat, where distributing code can be cheap and paid support or enterprise services can become profitable at scale.

The article’s central point is that AI inference remains costly: each user query requires chips, power and data-center capacity. BI noted that Moonshot AI’s Kimi K3 model drew industry interest but the company paused new customer sign-ups because of limited computing capacity. Open-weight developers may release trained model parameters, but revenue often flows to cloud providers or specialist inference firms that run the models for customers.

BI said analysts at William Blair and Barclays see difficult profit prospects, especially amid intense competition and low pricing. The report also framed China’s openness push as strategic, citing President Xi Jinping’s recent encouragement of open-source-style collaboration in Shanghai, which could pressure Chinese AI firms to prioritize broad adoption over near-term profitability.