From the 2024 Central Economic Work Conference, which first proposed comprehensive rectification of “involution” competition, to June 17, 2026, when the Ministry of Industry and Information Technology deployed in-depth rectification and simultaneous promotion of the “artificial intelligence + manufacturing” special action, the policy signal has been clear: the manufacturing industry can no longer rely on the endless cycle of “low price – low profit – low R&D”.
However, the same anti-involution policy applies to the three major tracks of industrial AI software and hardware, high-end equipment, and precision manufacturing, but its benefit logic is completely different.
For the industrial AI software and hardware track, the core benefit logic is directional policy support to guide the flow of resources from “hype outlets” to hard-core research and development.
In the past two years, the track has been severely involuted and focused on low-level redundant construction.
A large number of small and medium-sized manufacturers have gathered together in common industrial AI scenarios, and launched basic question answering and data analysis only through prompt word optimization, without any technical barriers; the supply of digital intelligence services in traditional manufacturing clusters is also highly homogeneous, data is not interoperable, and a large amount of investment has been wasted.
However, the “Implementation Opinions on the “Artificial Intelligence + Manufacturing” Special Action” jointly issued by eight departments changed the situation.
Policy requirements guide the misaligned development of enterprises and prevent industry involution competition. At the same time, they coordinate national scientific research funds and provide targeted support for key software and hardware collaboration technologies such as high-end training chips, end-side reasoning chips, and small models of industrial segmentation scenarios.
This means that companies that relied on following the trend to receive subsidies in the past will be eliminated. Companies such as Shenzhen Technology, which has mastered the “Material × AI” software and hardware integration technology and obtained hundreds of millions of yuan in financing, are the targets of resource tilt.
Morgan Stanley estimates that the industrial AI ecosystem will create an incremental market value of US$3 trillion to US$4 trillion.
The industry structure is moving from a “thousand-team battle” to a two-level structure of “full-stack leader + segmented scene specialization and innovation”, and the competitive logic has shifted from fighting for low prices to fighting for core technologies.
The benefit logic for the high-end equipment track is to accelerate the clearance of low-end homogeneous production capacity through capacity regulation, raising entry barriers and targeted procurement of first sets.
The involution distribution of this track is uneven. Competition in low-end mass production segments such as general magnetic components, ordinary industrial lasers, and conventional pharmaceutical equipment is fierce. The focus is on cost-effectiveness without differentiation barriers. In hard-core segments such as precision reducers, high-end servo systems, deep-sea offshore equipment, and high-power ship engines, the technical threshold is extremely high. Domestic leading manufacturers have ordered orders until 2028, and supply exceeds demand.
The anti-involution policy has been implemented accurately. The 2026 version of the “Implementation Measures for Production Capacity Replacement in the Steel Industry” has raised the replacement threshold and guided the concentration of production capacity towards high-end equipment and supporting materials; the Shanghai Stock Exchange has revised the listing guidelines for the Science and Technology Innovation Board to list robots as an independent second-level track under high-end equipment, and only supports the application of enterprises with core hard technologies; the Ministry of Industry and Information Technology and the State-owned Assets Supervision and Administration Commission have launched a special action for real-life training of humanoid robots and clearly prioritized the purchase of the first set of domestically produced equipment.
These measures have achieved remarkable results. In 2026, orders from domestic head laser manufacturers will exceed production capacity by 70%, and the average product price will increase by more than 15% year-on-year.
Industry profits have been restored, pricing power has recovered after the clearance of low-end production capacity, leading companies have seized the global market with the “China R&D + overseas delivery” model, and the proportion of CR5 has continued to increase.
The precision manufacturing track does not rely on direct subsidies from policies or production capacity regulation, but to repair “process premiums” through standard system upgrades and vicious low-price competition governance.
Its involution is a typical example of “bad money driving out good money”. A company that has accumulated micron-level processing technology for decades may lose to an opponent who simply lowers the price when bidding. The reason is that the industry lacks unified core evaluation standards for accuracy, energy consumption, and intelligence.
Zeng Yuqun, chairman of CATL, pointed out that the core root cause of the involution of precision manufacturing is insufficient protection of intellectual property rights. Copycats invest less in R&D but sell at low prices. Original creators can only follow the price reduction, and the innovation return chain is broken.
The anti-involution policy is highly targeted. Since 2025, China has released a total of 1,327 upgraded national standards in the field of machinery and equipment manufacturing to unify accuracy, safety, and energy efficiency standards; Hebei’s shared intelligent manufacturing policy covers 107 key precision manufacturing industry clusters, diluting the investment cost of high-end equipment through shared factories. In 2025, the revenue cost of enterprises above the cluster’s 100 yuan will decrease by 1.3 yuan, and total profits will increase by 16% year-on-year.
With the implementation of standards, the scarcity of technology accumulation has been transformed into product premiums, and the industry has shifted from “volume price” to “volume product power”.
In general, no one of the three benefits more, but they benefit in different ways.
Industrial AI software and hardware follows the fast track of “policy transfusion + technology iteration”, and the industry structure is drastically restructured; high-end equipment follows the “capacity clearing + order-driven” repair path, with strong certainty of profit recovery, and is in an upward cycle of capital expenditures; precision manufacturing follows the “standard dimension upgrade + process premium” quality route, and long-term accumulated know-how gains pricing power.
Their common goal is to promote the manufacturing industry to completely shift from the involution model of scale expansion to technology-driven, high-quality and healthy competition.
