On 28 October 2024 six ministries — MIIT, the National Development and Reform Commission, the Ministry of Finance, SASAC, SAMR and the National Data Administration — jointly launched the 2024 graded smart-factory cultivation action, issued as Gong Xin Ting Lian Tong Zhuang Han [2024] No. 399. The document establishes levels of factory intelligence and a cultivation path from basic digitalisation to advanced intelligence.
Why grading matters more than subsidy
A grading scheme is a market-shaping instrument. Once factories are classified, the level becomes a reference point in procurement, credit, land allocation and subsidy decisions, and suppliers of industrial AI start being asked whether their product moves a customer up a level. That converts a technology question into a measurable compliance and performance question, which is good news for vendors whose claims can be evidenced and uncomfortable for those whose claims cannot.
The six-ministry signature also matters. A joint document from the industry, planning, finance, state-asset, market-regulation and data authorities signals that the levels are intended to be used across those systems, not only within MIIT programmes. In practice that means a level reached in one channel can be cited in another, and the assessment records become reusable assets rather than one-off application materials.
Where it plugs into the wider system
The action is one segment of a longer chain. The near-term deployment guidance for manufacturing appears in the MIIT AI Plus manufacturing deployment notice, and the sector-level targets are set in the AI Plus manufacturing action plan. Compute supply depends on the compute interconnection action plan and on the national computing network service platform. Data inputs are handled by the industrial high-quality datasets action plan. The policy direction was reinforced at the AI industry industrialisation conference.
What AI companies should do with it
First, read the grading criteria as a product requirements document. If a customer’s funding depends on reaching a level, the features that satisfy the criteria are the features that will be bought.
Second, prepare the evidence trail. Graded cultivation implies assessment against indicators, and assessment implies records: deployment logs, measured efficiency gains, data governance documents and security assessments.
Third, watch the data boundary. Industrial AI deployments frequently involve plant data leaving the site, and the industrial dataset and data circulation policies increasingly expect that flow to be documented rather than informal. Contracting for that early is cheaper than remediating later.
This article is based on the 2024 graded smart-factory cultivation action (Gong Xin Ting Lian Tong Zhuang Han [2024] No. 399) and public information and is for reference only; it does not constitute legal advice.