MIIT Drafts New Smart-Manufacturing Policy to Push AI Agents Into China’s Factories

The ministry will validate factory AI agents for scheduling, defect detection and maintenance one by one; China has over 13,000 advanced-level smart factories.

WHY IT MATTERS

Suppliers of industrial AI agents face a ministry that will promote agents only after validation, against an existing target of 1,000 high-level industrial agents by 2027.

China’s Ministry of Industry and Information Technology (MIIT) is drafting policy measures for what it calls new-generation intelligent manufacturing (新一代智能制造), aimed at tying artificial intelligence more closely to factory production. MIIT chief engineer Wang Weiming announced the work at a State Council Information Office press conference on October 20, The Paper reported. He gave no timeline, budget or numerical targets for the new measures.

The most concrete element is a program for industrial AI agents, software that can plan and carry out tasks in a factory within set limits. Wang said MIIT will run a model-data resonance action (模数共振行动) around intelligent production scheduling, parameter tuning, defect detection and equipment maintenance, developing and validating agents prudently and promoting each one only once it matures.

The ministry is building on a large installed base. China has built more than 13,000 advanced-level smart factories, up from about 8,200 reported in June. Wang gave the figure at the briefing, according to Xinhua, and the 21st Century Business Herald reported that he also cited more than 500 excellence-level factories, the second-highest tier.

MIIT set four priorities for the next phase

According to Beijing News, which carried CCTV’s report, MIIT will focus on four areas: cultivating smart factories tier by tier, developing smart equipment, accelerating research on intelligent technologies and their adoption, and deepening international cooperation on smart manufacturing.

Each priority has a defined scope. Factories at the basic and advanced levels are to strengthen their digital and networked foundations, while those at the excellence and leading levels explore new system designs. Smart equipment covers machine tools, robots and inspection and control gear, as well as construction, agricultural and medical machinery. The technology track includes industrial datasets, models built for specific industries and an industrial operating system that connects data, models, software and equipment. International work will run through BRICS and other mechanisms.

The policy follows the National Advanced Manufacturing Conference in Beijing on September 16 and 17, which named new-generation intelligent manufacturing the main direction for the sector, the 21st Century Business Herald reported. MIIT spokesperson Tao Qing said the ministry would build a new batch of leading-level and excellence-level smart factories.

The agent push extends an earlier 2027 target

Wang said AI applications now cover more than 1,000 scenarios in 20 industries, including steel, autos and pharmaceuticals, and that more than 100 industrial agents capturing the know-how of experienced workers have been selected and promoted, according to the 21st Century Business Herald.

Those numbers sit well short of an existing goal. In January, MIIT and seven other departments issued implementation opinions on AI Plus Manufacturing (《“人工智能+制造”专项行动实施意见》) that set targets for 2027: 1,000 high-level industrial agents, 100 high-quality industrial datasets and 500 typical application scenarios, according to an Economic Information Daily report on Xinhua’s website. The same report cited IDC data showing the share of Chinese industrial companies using large models or agents rising from 9.6% in 2024 to 47.5% in 2025.

The smart-factory count has risen quickly since mid-year. In June, MIIT reported 35,000 basic-level, more than 8,200 advanced-level and more than 500 excellence-level factories, with 15 companies on a candidate list for the top, leading level, China News Service reported. MIIT said then that the tiered program had cut defect rates by an average of 47% and shortened development cycles by 38%. EastFrontier covered a city-level version of the push when Shanghai set 2028 targets for smart factories, AI and robots.

Compute and devices are part of the same plan

Xie Cun, an MIIT spokesperson and director of its Information and Communications Administration, said the ministry will speed up the expansion of computing facilities to support advanced manufacturing. Xinhua reported that China’s intelligent computing capacity reached 2,592 EFLOPS by the end of September, up 146% from a year earlier.

Xie also said more than 100 AI devices now meet the L3 assisted level of China’s national standards for grading AI terminals, covering phones, computers, televisions, glasses, car cockpits and earphones, according to People’s Posts and Telecommunications News, an MIIT-affiliated paper, as republished by Sina.

The broader industrial backdrop is steady. Industrial value added at large companies rose 5.3% in the first three quarters, and industry contributed about 35% of economic growth, China News Service reported.

The text of the new-generation intelligent manufacturing measures is the next document to watch. It will show whether MIIT attaches agent targets, funding or certification rules to the validation process announced on October 20.