A list, 34 names, and a PDF attachment.
On November 19, at the 2026 World Intelligent Manufacturing Conference, China's industry ministry published the 2026 cohort of "leading-tier" smart factories. The document number is Gongxin Ting Lian Tong Zhuang Han [2026] No. 562, dated November 17.
Two days between signature and announcement, which is ordinary in ministry documents.
Only 49 seats
Leading tier is the top rung of a four-level system for smart manufacturing.
The framework was launched in 2024 by six ministries and departments, rising from basic to advanced, excellent and leading. This is the second cohort, with 34 factories. Adding the first cohort of 15 in 2025, the national total is 49.
Spread across 31 provincial-level regions, that is fewer than two per province.
The distribution is uneven. Beijing has one: Xiaomi Auto's super factory.
Familiar names appear: Seres and its industrial-brain plant, ZTE's Nanjing base, Goertek, Hisense Visual, Tasly, Luzhou Laojiao, Hailan Home, Shougang Jingtang, Jiangxi Copper.
The industry spread is surprising — liquor, traditional medicine, apparel, steel, copper, cars, telecom equipment, all on one list.
The spread is the signal: showcase lists have long favoured electronics and autos, sectors born digital. Adding liquor, medicine and apparel suggests the criteria now weight difficulty of conversion.
These are not filler. One criterion requires major techno-economic indicators to lead globally. Another is harder: AI-enabled application scenarios must reach at least 60%.
What a 60% threshold means
Sixty percent of scenarios sounds soft.
On a shop floor it means this: how many steps in a line are judged, tuned and decided by a model.
Not how many cameras were installed, or whether a dashboard exists. It means the inputs and outputs of those steps are actually handed to an algorithm.
According to Shanghai Securities News, the 2026 criteria also require leading-tier plants to export advanced manufacturing models and solutions up and down the supply chain through a "mother factory" arrangement.
That clause is the real dividing line.
An ordinary demonstration plant upgrades itself: my line runs faster and costs less, so it succeeded.
Leading tier asks for something else: my method can be copied by someone else.
The trouble sits here. How to count scenarios is not specified. A line can be sliced into twenty workstations or grouped into five processes; change the denominator and the 60% means something different. Review reads the decomposition in the application; acceptance reads whether it reproduces on site. Between them sit a site visit and an on-stage defence.
Where 117 suppliers come from
The ministry supplied a figure: the 34 plants have already pulled nearly 4,000 companies into digital and intelligent upgrading.
4,000 divided by 34 is about 117.6.
On average, each leading-tier plant drags 117 suppliers along behind it.
That is my division, and the scopes may not match — "cumulative" in the numerator, 34 in the denominator — but the scale is informative. Leading tier is not a certificate for one factory. It is an entrance ticket for a supply chain.
Another figure is more direct: the second cohort raised labour productivity by an average of 34.4% and production efficiency by 19.3%.
Productivity grew at nearly twice the rate of efficiency. That gap usually comes from two things: fewer people, or handing off low-efficiency steps.
The word "cumulative" also leaves room. An automaker pulling 300 parts suppliers onto one data platform counts. A supplier buying a domestic robot because it won an order also counts. The first changes how work is coordinated; the second changes one machine. Both sit inside the same 4,000, so read it with a discount.
Dushanzi's 90%
One entry stands out: PetroChina's Dushanzi Petrochemical.
It is the only refining and petrochemical project on the list, and PetroChina's first leading-tier site. The plant has run for 90 years.
A 90-year-old plant is not a new smart factory; its path is iteration on mature units. Retrofitting is harder than building new.
According to China News Service, the company built an AI-driven platform for predicting and designing molecular structures and synthesis routes for SSBR, working backwards from target performance to molecular design and process route.
Developing solution-polymerized styrene-butadiene rubber used to mean testing hundreds of formulations one at a time. Now simulation screens out 90% of invalid options, leaving only high-potential formulations for the bench.
Development cycles for new SSBR products fell by 50%.
Put those two numbers together and the claim becomes concrete. The model does not run experiments. It excludes: it crosses out nine of ten formulations destined to fail so the bench only handles the tenth. The bottleneck moves from "can we think of a formulation" to "how many can we validate in a day" — and the pilot plant's schedule is what sets that pace.
Beyond the list
Wang Ruihua, a member of the national expert committee on intelligent manufacturing, framed the list this way: leading tier should move Chinese manufacturing from imitation to accumulating its own paradigms.
Accumulation rests on three things: industry-specific industrial models, proprietary equipment systems, and domestic industrial software.
Those three are exactly what has been easiest to "integrate" away. Buy foreign software, install a foreign line, and it can still be called a smart factory.
Leading tier moves the standard: not what you bought, but whether you can write down your method, hand it over, and have someone else reproduce it.
So 49 will not rise quickly.
The thresholds are the 60% scenario share, the mother factory, and 117 suppliers. None of those come together in a year.
The list is a list. What gets tested is how many of those 4,000 suppliers have actually rebuilt a line three years from now.
*Disclaimer: This article draws on the joint notice by six ministries and departments, and reporting by Shanghai Securities News and China News Service. Calculations are the author's own. It is for reference only and does not constitute investment advice.*
