CITIC Securities: Global AI may be diverging into two parallel ecosystems.

The report from CITIC Securities states that the global AI may be diverging into two parallel ecosystems. American AI is primarily driven by private capital, with closed-source flagship models and enterprise software subscriptions, and the core of the business model is to replace high-cost human labor. Chinese AI, on the other hand, is driven by policies and industries, with open-source lightweight models and low-cost inference services, focusing on upgrading the manufacturing industry and increasing digital consumption. It is believed that this divergence is not just a temporary technological gap, but a path divergence determined by economic structure, policy governance, capital preferences, and factor endowments, which may be difficult to converge in the short term. In terms of different stages, the gap between Chinese and American AI decreases from top to bottom along the industrial chain, with the largest gap in upstream chips and the smallest gap in application layers. It is believed that the certainty in the US lies in the cash flow of enterprise-end revenue and computing power supply chain already validated by top model companies, with variables in the race between commercial realization and financing patience; while the certainty in China lies in the domestic substitution of inference computing power, global penetration of open-source ecosystems, and the landing of manufacturing scenarios, with variables in whether computing power utilization and commercial pricing can be smoothly implemented. Looking ahead from a macro perspective, the United States may be the first to experience a rise in service industry productivity driven by AI, while China will evolve along two paths: the deepening of intelligence in the manufacturing industry, and filling the productivity service industry gap with cost-effective intelligence.