Industrial AI adoption is accelerating: how can industrial internet platforms meet demand?
Citing IDC research, the material says the proportion of Chinese industrial enterprises using large models and AI agents rose from 9.6% in 2024 to 47.5% in 2025. China Post Securities sees promise in industrial internet platforms’ accumulated data and industry knowledge, but expanding applications into factory deployments also requires connecting business processes and resolving compatibility across platforms and equipment.
Original source：财联社First published on Unhidea：Original source date：
Key takeaway
Industrial AI adoption is rising rapidly. Platforms’ accumulated data, models and ecosystems form the basis of competition, while connected business processes and hardware and software compatibility determine how those assets translate into factory applications.
How much has industrial AI adoption accelerated?
The material cites IDC research: the proportion of Chinese industrial enterprises using large models and AI agents rose from 9.6% in 2024 to 47.5% in 2025, an increase of 37.9 percentage points in one year. This shift turns the research focus to which platforms can meet the demand arising from the expansion of enterprise applications.
The China Academy of Information and Communications Technology expects penetration of leading industrial AI to reach 60% over the next three years. This forecast uses a different measure from IDC’s “proportion using large models and AI agents” and is also qualified by “leading”; the two figures should not be directly combined into an industry-wide penetration curve.
Why are institutions optimistic about established industrial internet platforms?
Chen Hanbo of China Post Securities believes that the data gathered, models developed, industry knowledge and ecosystem built by platforms at an earlier stage create a first-mover advantage. The business logic is that industrial AI must fit specific production scenarios, and platforms’ existing assets can provide a foundation for developing high-quality datasets, industrial large models and industrial AI agents.
Policy is also encouraging platforms to shift from expanding connectivity to improving capabilities. The Action Plan for Promoting High-Quality Development of Industrial Internet Platforms (2026–2028) calls for more than 450 influential platforms by 2028, over 120 million connected units of industrial equipment, and a platform adoption rate above 55%, while strengthening intelligent capabilities and the accumulation of data and models.
Application pathways are differentiated by scenario: promote discriminative AI in clearly defined, rule-based processes such as production control and risk identification; explore generative AI in complex processes such as process optimization and solution design. Platforms then support multi-agent orchestration through agent interconnection protocols, making the connection among data, models and production tasks central to meeting demand.
How can platforms’ accumulated assets become practical factory applications?
The Industrial Internet Operating System Industry Alliance was established in Shenzhen, bringing cross-domain collaboration to the fore. Li Qingxin, a researcher at the Chinese Academy of Sciences’ Shenyang Institute of Automation, said that deploying embodied intelligence in factories requires connecting both the design-to-manufacturing and management-to-control business processes, integrating people, machines, materials, processes and the environment.
This means deployment barriers also include cross-domain and cross-platform design, trusted task execution, consistency when transferring models, and compatibility with multiple types of hardware. Accumulated models and data are only a starting point; whether systems can connect business processes and work with equipment directly constrains application expansion.
Yonyou Jingzhi, cited in the material, has been selected for the Ministry of Industry and Information Technology’s list of cross-industry, cross-domain industrial internet platforms for four consecutive years. It serves 2.44 million enterprises, connects 1.39 million units of industrial equipment, and brings together 22 thousand industrial apps and 2476 industrial mechanism models. These existing resources reflect the platform foundation valued by China Post Securities; turning them into competitive strength still depends on building high-quality datasets, large models and AI agents.
