China’s Ministry of Industry and Information Technology (MIIT) on March 10 issued a policy notice launching the “Industrial Data Foundation Action,” a national pilot program aimed at building high‑quality industry datasets for AI‑enabled manufacturing. The initiative targets key manufacturing sectors and sets end‑2026 goals for forming data‑cooperation consortia, building trusted data‑interconnection platforms, and producing standardized, tradable datasets that can support industry‑specific large models and industrial agents. The move signals Beijing’s intent to tackle industrial data bottlenecks that have limited AI deployment at scale, shifting the focus from isolated data collection to cross‑industry data supply infrastructure.
The notice (工信厅信发函〔2026〕64号) lays out a multi‑track roadmap. By the end of 2026, MIIT wants to cultivate sectoral data‑collaboration alliances, establish key industry data‑trust platforms, aggregate industry data resources, and advance core data technologies while producing industrial data standards. The policy’s explicit goal is to create high‑quality, circulation‑ready datasets that can be used to train and deploy industry large models and industrial agents, rather than ad‑hoc datasets locked inside individual factories.
MIIT and state media describe the action as a “pilot first” effort in priority manufacturing fields. Xinhua’s coverage notes that the program will explore mechanisms for efficient data collection and processing, trustworthy circulation, and deep integration of industrial data with AI applications. China’s industrial data is often fragmented across firms, supply chains, and regions; the policy therefore emphasizes the “collection–processing–use” chain and calls for common standards and platforms that enable data to move safely and compliantly across enterprises.
The policy push reflects a broader shift in China’s industrial digitalization strategy, alongside moves like China’s 2026 AI legislation push. MIIT has been promoting “data elements” as a production factor, and the new action treats industrial data as an infrastructure layer rather than a by‑product of operations. That framing matters for industrial AI adoption: without standardized, shareable datasets, sector‑specific AI models struggle to generalize beyond a single plant or vendor. The action’s explicit support for industry models and industrial agents puts the dataset supply chain on equal footing with compute and algorithm investments, such as Shanghai’s AI compute voucher program.
Industry scale is one reason this matters. According to C114’s reporting of MIIT data, China’s industrial internet core industry scale was expected to exceed 1.6 trillion yuan in 2025, driving roughly 2.5 trillion yuan in industrial value‑added growth. These figures highlight the economic stakes behind industrial digitalization. A national program that upgrades data quality and interoperability could therefore have a direct multiplier effect on industrial AI deployment, especially in sectors where data quality and ownership have slowed cross‑enterprise collaboration.
The action also codifies how MIIT expects data sharing to work. It calls for trusted interconnection platforms and industry data standards, which suggests a governance framework that balances data rights, security, and commercial value. For manufacturers, this may lower barriers to cross‑supplier analytics and AI training, because data can be exchanged under standardized rules rather than ad‑hoc bilateral agreements. For platform vendors, it creates a clearer market for data infrastructure products such as labeling, lineage, and compliance tooling.
For AI developers, the significance lies in the policy’s emphasis on “high‑quality, circulation‑ready” datasets. Industrial AI models and agents depend on structured, accurate, and context‑rich data, yet factories often generate data in incompatible formats. By pushing sector‑level standards and shared data infrastructure, MIIT is effectively attempting to increase the supply of usable training data for industrial AI, which could improve model performance in predictive maintenance, quality inspection, energy optimization, and supply‑chain coordination.
Near‑term outcomes will hinge on execution. The next steps likely include the designation of pilot industries, the formation of data‑cooperation consortia, and the build‑out of trusted platforms that can be audited for security and compliance. If these pilots succeed, they could establish replicable templates for other sectors, creating a path from policy guidance to operational datasets that AI developers can rely on.
What changed is that China’s industrial data strategy now has a national pilot mechanism, a 2026 timeline, and explicit deliverables tied to datasets, standards, and trusted platforms. What may happen next is a faster push toward cross‑enterprise data sharing and sector‑specific AI deployments, as MIIT’s framework turns industrial data from a siloed asset into a tradable, standardized foundation for manufacturing AI.
Sources
- https://www.miit.gov.cn/zwgk/zcwj/wjfb/tz/art/2026/art_f2b5d5d3dda44da5b8d3fa7d3198656a.html
- http://www.news.cn/tech/20260312/b604b7088eaa489c89c2a95e4910e9e0/c.html
- https://www.chinanews.com.cn/cj/2026/03-11/10585282.shtml
- https://finance.sina.com.cn/tech/roll/2026-03-12/doc-inhqtyyf8336851.shtml
- https://m.c114.com.cn/w241-1304644.html
