From Chain-Based Global Value Chains to Intelligent Service Ecosystems: New Forms and Governance Challenges in China's Digital Services Trade Driven by Large AI Models

Abstract

The large-scale adoption of artificial intelligence (AI) models is changing the logic of value creation in digital services trade. Traditional global value chains are mainly organized around sequential stages such as research and development, production, processing, delivery, and sales. Early software and digitally delivered services reduced the importance of physical transportation, but still relied largely on software products, outsourcing projects, platform functions, and subscription-based delivery. Generative AI, agentic applications, application programming interfaces, and token-based metering are producing a different organizational form. This paper develops the concept of an intelligent service ecosystem to describe a networked system in which models, computing power, data, cloud platforms, developers, enterprises, and users jointly create and continuously deliver intelligent capabilities. The paper argues that the object of trade is shifting from software products to callable capabilities, delivery is shifting from project completion to continuous invocation, and pricing is becoming increasingly granular through tokens and usage. China already has a substantial digital-services and software-industry base for this transition, but important challenges remain in statistical classification, token standards, cross-border data governance, computing infrastructure, service-quality disclosure, and international rule alignment. Policy should therefore move beyond model exports toward ecosystem-level competitiveness and governance.