Intelligent agents become a new engine driving the transformation of the manufacturing industry

Economic Reference News Agency Reporter Li Baojin

At the start of 2026, the development of industrial intelligent agents enters a “golden period” of policy and industry resonance. The Ministry of Industry and Information Technology and seven other departments jointly issued the “Special Action Implementation Opinions on ‘Artificial Intelligence + Manufacturing’,” clearly aiming to launch 1,000 high-level industrial intelligent agents by 2027, injecting strong momentum into industry development. Against this backdrop, leading manufacturing companies like Midea and Haier, along with industry chain enterprises, are accelerating their deployment, focusing on a full-scenario intelligent agent matrix to promote intelligent manufacturing implementation.

Industry experts believe that the continuous release of policy dividends and deep practical application by enterprises are mutually empowering, driving industrial intelligent agents from pilot points to large-scale adoption. This shift is a core driver for transforming manufacturing from “scale expansion” to “quality leap” and cultivating new productive forces.

Policy Reinforcement

At the beginning of 2026, the industrial intelligent agent sector saw a triple policy boost: national top-level design, local detailed implementation rules, and guidance from the Two Sessions.

In January, the Ministry of Industry and Information Technology and seven other departments jointly issued the “Special Action Implementation Opinions on ‘Artificial Intelligence + Manufacturing’,” proposing that by 2027, 3-5 general large models will be deeply applied in manufacturing, forming industry-specific, comprehensive large models, launching 1,000 high-level industrial intelligent agents, creating 100 high-quality industrial data sets, and promoting 500 typical application scenarios.

Meanwhile, cities like Shenzhen and Chongqing quickly introduced supporting policies focusing on funding subsidies, computing power support, and scene opening to ensure precise implementation of industrial intelligent agents.

On March 5, the government work report emphasized: deepen and expand “Artificial Intelligence +”, promoting the rapid adoption of new-generation intelligent terminals and intelligent agents.

Du Yanze, senior research manager at IDC China, told the Economic Reference News Agency: “Industrial intelligent agents are becoming the core of ‘AI + manufacturing.’ The fundamental reason is that their role has shifted from ‘passive tools’ to ‘autonomous digital employees.’ Currently, more enterprises are paying attention to the ROI of AI investments, shifting from ‘whether to do AI’ to ‘how AI can truly help improve quality, reduce costs, and increase efficiency.’”

IDC’s survey of Chinese industrial enterprises in 2025 shows that the proportion of companies applying large models and intelligent agents has rapidly increased from 9.6% in 2024 to 47.5% in 2025; among them, companies applying across R&D, manufacturing, and supply chain stages simultaneously rose from 1.7% to 35%. This indicates that industrial intelligent agents are moving from single-point experiments to cross-linking applications.

Experts believe that policy empowerment will accelerate the penetration of intelligent agents across R&D, production, and supply chains, pushing manufacturing from “automation” to “autonomy,” fostering new production models like lights-out factories and flexible manufacturing, and reshaping industry competitive advantages.

Enterprise Accelerates Deployment

IDC predicts that by 2028, China’s industrial AI expenditure will approach 90 billion RMB, with a compound annual growth rate of 37.7%. This indicates that AI has moved from the “concept investment phase” to the “scalable expansion phase.” Given the continued policy support and gradually clearer ROI, maintaining about 35% annual growth in the next 3-5 years for the “AI + manufacturing” market is a reasonable and sustainable outlook.

As the “AI + manufacturing” industry chain continues to advance, leading manufacturing companies and industry chain enterprises are accelerating the implementation of intelligent manufacturing centered on full-scenario intelligent agent matrices.

In January 2026, Midea’s Me Yun Zhishu released the Meqi AIGC 3.1 platform and intelligent agent factory solutions, building an intelligent agent matrix covering R&D, manufacturing, supply chain, marketing, and management, with 158 core scenarios implemented. Midea states that on the manufacturing side, intelligent agents like TPM and molds enable equipment fault prediction and process optimization, helping improve Overall Equipment Effectiveness (OEE) by 30% and doubling inspection efficiency; on the supply chain side, intelligent agents shorten end-to-end delivery cycles by 39% and reduce inventory turnover days by 30%.

Haier Smart Home also announced that through the super intelligent agent “ZhiXiaoNeng,” they are achieving full staff AI integration, with parallel development of technology creation and application, resulting in a 90% increase in R&D efficiency, a 10% reduction in procurement costs, and an 80% boost in office efficiency, fostering a new work mode of human-AI collaboration.

Shenzhen tech company xTool, specializing in laser engraving machines, officially launched the world’s first AI creative intelligent agent—AImake—in January. As the industry’s first agent with “manufacturing contextual awareness,” AImake enables instant conversion from natural language ideas to processable design drawings, greatly lowering the technical barriers for non-professional users entering laser manufacturing.

In March, Huawei focused on solving the challenges of AI agent deployment by launching an AI data platform to strengthen the data foundation for enterprise digital transformation.

Additionally, listed company Wiston recently revealed on an interactive platform that Wiston Industrial AI intelligent agent software can meet industrial application scenarios, changing human-computer interaction modes, further promoting product technological upgrades, and enriching the company’s product line.

Han De Information states that the company has developed its own “DeLing” B-end AI application system, including the “LingShou” business intelligent agent series, covering manufacturing, marketing, finance, supply chain, HR, and comprehensive operations, which are gradually being deployed in actual scenarios for top clients.

Obstacles to Large-Scale Deployment Remain

Industry experts believe that with policy support, industrial intelligent agents are accelerating deployment and promoting the transformation and upgrading of intelligent manufacturing. However, during large-scale implementation, challenges such as technical adaptation, data infrastructure, cost-benefit analysis, and ecological collaboration remain intertwined, becoming key obstacles to moving from pilot projects to widespread adoption.

Experts point out that developing and deploying industrial intelligent agents requires significant investments in computing power, algorithms, and talent, with high technical barriers, large capital requirements, and data security risks—including costs for training industrial large models, customized development, and professional maintenance—posing heavy burdens especially for small and medium-sized enterprises.

Du Yanze notes, “In the short to medium term, the most urgent challenge for ‘AI + manufacturing’ industry upgrading mainly comes from the ‘market side’ rather than the ‘technology side.’” On one hand, Chinese manufacturing enterprises remain highly enthusiastic about AI, supported by policies, open-source ecosystems, and talent supply; on the other hand, profit margins are under pressure, and budgets are tight, making companies more cautious about AI investments.

Specifically, Du believes that current core constraints include: high ROI pressure requiring AI projects to quickly and clearly demonstrate business value; fierce market competition with homogeneous solutions and price wars squeezing sustainable investment capacity; and insufficient supply-demand matching, with a scarcity of truly “industry-savvy” AI products and services.

Additionally, some industry insiders note that many manufacturing SMEs face fragmented industrial data, low standardization, and poor security management, which hinder AI training and application. Moreover, the lack of unified standards and evaluation systems for industrial AI, incompatible system interfaces, and data formats across companies increase integration costs and implementation difficulties.