A newly released report projects that China’s active AI-agent count could reach 197 million in 2030, from a 2026 estimate of 4.95 million. The headline number is large, but it is not a current count, a government target, or a guaranteed result. It is a forecast attributed to the report titled the 2026 China AI Computing Power Development Assessment.
TechNode reports that the report was released September 21 and gave the China forecast. CCTV News reported the same report figures, including a projected 151.2% compound annual growth rate for China’s active agents over 2026 to 2030.
The report’s figures are useful because they identify the scale its authors expect. They are not evidence that China will definitely reach 197 million agents. The underlying sources also do not define active AI agents as users, applications, accounts, revenue, or completed tasks. The term should remain the report’s metric rather than being silently converted into a different measure.
The China and Global Projections
According to the report, China’s 2026 estimate is 4.95 million active AI agents. It places the 2030 figure at 197 million and assigns a 151.2% compound annual growth rate to that China forecast. The report’s scope is forward-looking, so each of those figures should be introduced with the report projects, the report estimates, or similar language.
The report also presents a global comparison. It places the global active-agent count at 79.4 million for 2026 and 2.216 billion for 2030. The China and global totals are distinct projections. The 151.2% CAGR in the source applies to China’s active-agent estimate and should not be recast as a growth rate for the global count.
TechNode summarized the report as a view that AI is moving from content generation toward task execution involving multi-step reasoning, tool use, and coordination across systems. CCTV’s summary similarly describes a shift from content creation to task execution. That is the report’s stated framing. It should not be expanded into unverified claims about particular industries, user behavior, or company deployments.
The current 4.95 million figure also requires restraint. It is the report’s 2026 estimate. The cited materials do not describe it as a real-time national census, and they do not say how many people use an agent or how many firms have deployed one. It is a model of the active-agent market as defined by the report.
Compute Forecasts Run Alongside Agent Forecasts
For China’s AI-computing market, the report gives a 2030 forecast of US$150.1 billion and describes it as four times the 2025 level. It projects the global AI-computing market will reach US$1.25 trillion by 2030, also four times its 2025 level.
These are also forecasts. The sourced summaries do not identify which companies will supply the capacity, how much comes from hardware versus cloud services, or what methodology produced each market total. It would therefore be wrong to turn the figures into a procurement forecast or a statement about a named vendor’s future revenue.
What the report does do is put agents and AI-computing in the same forward-looking assessment. It offers a set of linked estimates: China’s active-agent population, the worldwide active-agent population, China’s AI-computing market, and the global AI-computing market. The source does not establish a causal sequence beyond presenting these figures together.
For readers, the most useful way to read the numbers is as a scenario. The report expects far more systems it categorizes as active agents and a much larger AI-computing market by 2030. The size of the scenario makes the definitions, assumptions, and forecast horizon important, even though the brief source summaries do not provide the report’s complete methodology.
How to Read the 2030 Agent Forecast
The sources do not give a sector breakdown for the projected 197 million agents. They do not identify individual products, companies, regional adoption levels, revenue sources, investment totals, or employment effects. They do not specify whether an active agent is a consumer service, an enterprise system, an installed device feature, or another type of system.
The report also does not establish that the forecast represents government policy. It was released in current reporting, but a forecast is not a mandate. Nor does the cited material independently validate that the 2030 numbers will be reached.
Those omissions do not make the report unimportant. They define how the story should be read. A forecast of 197 million active agents is a marker of the scale analysts expect, not a count that can be treated as already achieved.
EastFrontier’s archive has previously examined China’s national guidelines for AI agents and agent connectivity standards. The present report adds a numerical outlook, but it does not replace the need to distinguish policy, technical standards, current activity, and projections.
The report’s core message is straightforward. It expects rapid expansion in the number of systems categorized as active agents and in spending on AI computing through 2030. The report’s stated figures are 4.95 million active agents in China for 2026 and 197 million for 2030; globally, 79.4 million and 2.216 billion; plus AI-computing-market forecasts of US$150.1 billion for China and US$1.25 trillion worldwide. Every one belongs to a forecast, and that label is the most important fact in the story.