- China's active AI agents will top 350 million by 2031, driving a more than 30-fold annual surge in token consumption, according to IDC.
- Alibaba's full rollout of its desktop agent QoderWork highlights an industry shift toward localized, secure execution and cost control.
Chinese companies are entering an unprecedented acceleration phase in artificial intelligence (AI) agent deployment, a trend underscored by Alibaba's full rollout of its desktop agent application QoderWork this week.
The number of active AI agents in China is projected to surpass 350 million by 2031, with a growth rate outpacing major global markets, according to the latest forecast by market research firm IDC.
This massive scale represents a compound annual growth rate of over 135%, driven fundamentally by leaps in domestic large model capabilities and the maturation of interconnected ecosystems, IDC said in a report on Wednesday.
Furthermore, the implementation of standardized protocols including MCP (Model Context Protocol) and skills has lowered the threshold for system integration, while the Chinese government's industrial policies have provided certainty for project budgets.
As the execution density and complexity of agent tasks increase, token consumption will experience an exponential leap.
Token consumption by AI agents is expected to grow at an average annual rate of more than 30 times, which will fundamentally alter corporate IT budgeting logic and introduce cost pressures, IDC forecast.
QoderWork, released by Alibaba on March 3, demonstrates how companies are addressing the challenges of data security and deployment costs.
Looking at the overall market structure in China, agents built on low-code or no-code platforms will account for the absolute majority in terms of volume in the future, according to IDC.
Although independent agents currently have a smaller deployment scale, they possess the greatest growth elasticity going forward as system openness and interoperability protocols develop, IDC said.
Acquiring AI agents is only the first step; the real challenge for enterprises lies in operating these AI tools stably and securely in production environments.
Full-stack observability, fine-grained access control, and auditable mechanisms will become indispensable foundational capabilities for enterprises deploying agents at scale, according to IDC.
