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

China’s ‘open-weight’ AI models, while technically impressive, are proving to be a poor business model, leading to significant financial losses for companies like Zhipu and MiniMax. Unlike open-source software, AI requires substantial ongoing costs for computation and infrastructure, meaning profits are captured by cloud providers rather than the model creators. This strategy may be a deliberate move by China to commoditize AI and pressure US competitors, even at the expense of its own companies’ short-term profitability.

• Open-weight AI models are distinct from open-source software and are proving to be a weak business model.
• Chinese AI labs like Zhipu and MiniMax are experiencing significant financial losses and stock plunges despite releasing advanced open-weight models.
• The core economic difference lies in AI’s high operational costs (chips, electricity, data centers) for each use, unlike software’s near-zero marginal cost.
• When AI models are ‘open-weight,’ their trained parameters are released, and inference (running the model) is often handled by third-party cloud providers, who capture the revenue.
• Companies that release open-weight models may receive little to no ongoing revenue, despite investing heavily in training.
• This strategy might be a deliberate attempt by China to challenge US AI leaders like OpenAI by making advanced AI a commodity and forcing price reductions.
• The Chinese government, under President Xi Jinping, is encouraging ‘openness’ in AI strategy, potentially prioritizing strategic advantages over immediate financial gains for domestic companies.
  Continue reading https://www.businessinsider.com/china-ai-boom-terrible-business-open-weight-models-2026-7