Industrial AI vendors now face a cabinet yardstick built on deployment, with MIIT's October 15 catalog of over 1,000 manufacturing scenarios marking the use cases officials want replicated.
China’s State Council has moved its "AI Plus" initiative into a delivery phase. At an executive meeting on October 16 chaired by Premier Li Qiang, the cabinet heard a progress report on the campaign and called for application-led innovation, pilot testing and validation, and closer matching between AI suppliers and the real economy, according to Xinhua and the State Council’s English-language summary. The same session also reviewed a draft revision to the Teachers Law and a draft amendment to the Pricing Law.
Earlier the same day, a report carried by CNR on the morning of October 16 put numbers on the problem the cabinet now wants solved, drawing on the Ministry of Industry and Information Technology (MIIT) and the China Academy of Information and Communications Technology (CAICT). According to CAICT vice president Wei Liang, nearly half of China’s "AI+" application scenarios are still stuck at the pilot stage without being rolled out at scale. Wei cited statistics without naming their source.
Commentary published over the following three days reads the meeting as a turn from building capacity to proving returns. For companies selling AI software, computing and integration services to Chinese manufacturers, that turn shifts the emphasis from model benchmarks to deployments that can be copied from one plant to the next.
The cabinet’s second AI session this year moved from capacity to payoff
The October meeting was the State Council’s second this year centered on AI. A comparison by the 21st Century Business Herald sets it against a June 29 session that heard a general report on AI development and stressed key technologies, large computing clusters, data supply, talent and capital. The October agenda, by contrast, emphasized supply-demand matching, joint research, pilot validation and the cultivation of application scenarios. It also added a new line on coordinating computing capacity with electricity supply, as the gov.cn text renders it.
Both meetings described AI as a key factor in China’s future development, held safety as a fixed baseline and pointed to China’s complete industrial system and wide range of use cases as its advantages. The October text calls for differentiated, tiered oversight by sector and stronger risk assessment.
Zhu Keli, founding president of a new-economy research institute, told the Herald that the earlier approach of making a technical breakthrough and then searching for uses had been reversed, with real industrial demand now meant to drive algorithm and model upgrades. Wu Chaoming, chief economist at Caixin Financial Holdings, said the yardstick for AI policy was moving from model indicators to productivity indicators. That reading is the two analysts’ own; the meeting readout does not use those terms.
MIIT’s figures show scale, but the bottleneck is replication
The ministry’s own tally is large. MIIT vice minister Yu Xiaohui put China’s core AI industry above RMB1.2 trillion, with more than 6,600 AI companies and intelligent computing capacity of 2,592 EFLOPS, CNR reported. On October 15 the ministry published a catalog of high-value scenarios for AI-enabled new industrialization (人工智能赋能新型工业化高价值场景), covering more than 1,000 use cases across 20 sub-sectors including petrochemicals, steel, autos and consumer electronics.
MIIT has also held nearly 30 matchmaking events that connected more than 5,000 companies, backed 18 national AI innovation application pilot zones (国家人工智能创新应用先导区) including Beijing and Jinan–Qingdao, and helped industry draft close to 200 key AI technical standards, according to the same report. The RMB60 billion national AI fund it lists among its measures is older: the Securities Times reported that MIIT and the Ministry of Finance formally set it up on January 17, 2025.
Wei’s catalog sorts scenarios into three maturity grades: those ready to replicate, those that have finished a pilot but need wider use, and those still in technical development. Haibao News, in a report republished on NetEase, added that Wei pointed to companies following trends blindly, duplicated projects and successes that prove hard to copy. The same report cited Shanghai suit maker Baoniao, which it said cut custom-order delivery from 20 days to five after putting AI into design, scheduling and management, and Dongfeng Yipai, whose agent system reportedly cut urgent equipment-fault response from ten minutes to four.
Cost for smaller firms is the next constraint to watch
A commentary by Xie Ruolin of the Economic Observer, republished by Sina Finance on October 19, identifies the gap between adoption costs and what small and midsize companies can absorb as the main obstacle. Xie cites high upfront spending, scarce talent and patchy industry data and standards. She proposes verifiable scenario catalogs for manufacturing, agriculture, healthcare and city management, sector testing grounds, and a first-use mechanism that would help new technology into real operations through government procurement, evaluation and certification, and insurance compensation.
Several of those tools echo measures already in central guidance. EastFrontier reported on October 10 that a new policy opinion on AI and new productive forces called for national pilot-test bases and high-value scenarios while warning local authorities against blind investment.
The State Council readout sets no budgets or deadlines. The concrete next marker is the 2026 edition of the national guide to building a comprehensive AI industry standards system (《国家人工智能产业综合标准化体系建设指南（2026版）》), which Yu said should be issued quickly, according to CNR. A first-use mechanism with procurement or insurance backing, of the kind Xie proposes, would show cost relief for smaller manufacturers moving from commentary into policy.