The measurement. Of sixteen profiled Chinese PCB plants, two name a laminate grade they actually use: Jingming publishes an exhibited board built on Shengyi S1000-2M at TG170, and Wanyexing names Wangling and Shengyi as base materials for high-frequency work. Two more name only a material category — Hong'an Xingsheng writes “FR-4”, Shengda Xinxing writes “single-sided aluminium-based”. The remaining twelve name nothing at all: no grade, no Tg, no mill.
Nobody names the plant or the lot. Zero of sixteen state which factory of which mill the material ships from, or the date code on the pallet. Grade-level traceability, the thing that would let you match a quote to a stack-up six months later, is not published by any plant here.
Why it matters now. The upstream concentrated in 2025. Prismark puts the rigid laminate market at US$18.7bn, up 24.4%, with the top three suppliers holding over 40% of it — and in the HDI-grade segment one supplier holds roughly 70%. Nine of the sixteen plants here claim HDI work. Every one of those claims sits downstream of that concentration.
The columns that matter are the third and fourth. A grade attached to a marketing sentence is not the same as a grade attached to an exhibited board, and neither is the same as a grade attached to a stack-up drawing — which no plant here publishes.
| Plant | What is named | What it is attached to | Source type |
|---|---|---|---|
| Jingming 景明 · Suining | Shengyi S1000-2M TG170 | One exhibited 8-layer UAV main board: 1+2+1 build-up, 0.3 mm minimum via, 11/8 mil line/space | Trade press report of the plant opening — not a document the company published |
| Wanyexing 万业兴 · Suining | Wangling, Shengyi brands, no grade | “Base materials” for high-frequency and microwave work; no grade, no Tg, no layer | Company B2B listing — a trade-platform profile, not its own site |
| Hong'an Xingsheng | “FR-4” a category, not a grade | Its process list, alongside a 10:1 plating aspect ratio and 3/3 mil line/space | Company site |
| Shengda Xinxing | “Single-sided aluminium-based” a category | Its core product family — aluminium-substrate boards are what the plant is known for | Registry and trade press; the plant has no website |
| Champion Asia, Yingchuangli, Suntak, Kaishengshi, Kinwong, Victory Giant, Shengyi Electronics, Xunjiexing, Glorysky, Puruisen, Haiying, Zhichao | Nothing | Twelve plants, no grade, no brand, no Tg on any page this directory could read | — |
The twelfth row includes the plant whose parent makes the material. Shengyi Electronics (688183) is 62.94% owned by 生益科技 (600183), the world's second-largest rigid laminate maker. It publishes no laminate grade, no supplier and no material data of any kind — while its parent's own annual report names the grades and the tonnage. The most privileged access in this directory is the least documented.
A naive scan finds the string “Shengyi” on six of the sixteen profiles and would report six plants naming their laminate supplier. That count is wrong, and the reason is worth knowing if you ever run this check yourself.
Four of those six mentions are not supply disclosures at all. Suntak, Kinwong and Xunjiexing each identify “Shengyi Electronics (生益电子, 688183), majority-owned by the largest laminate maker in China” as a competitor plant in a paragraph about where their rivals are located. The sixth is the Shengyi profile describing itself. In each case the string is a company name doing geographic work, not a material statement.
The real count is two, and both are quoted in the table above. This is the same discipline as reading a capacity figure: the number only means something once you know what it is attached to. A grade inside a marketing sentence, a grade attached to an exhibited board, and a grade on a stack-up drawing are three different strengths of evidence, and only the last one survives a change of purchasing manager.
Prismark’s June 2026 update puts the rigid laminate market (excluding flex and mass laminate, including prepreg) at US$18.7bn in 2025, up 24.4% from US$15.0bn. The ranking changed at the top for the first time in years, and the reason is the same reason nine plants here now advertise HDI: AI servers, high-performance computing and advanced networking pulled demand toward high-speed, low-loss material.
| 2025 rank | Supplier | Share | Sales | Note |
|---|---|---|---|---|
| 1 | EMC 台光电子 (Taiwan, China) | 16.2% | US$3.03bn | Took first place in 2025, displacing Kingboard |
| 2 | SYTECH 生益科技 (China) | 13.2% | US$2.47bn | The parent of one plant in this directory |
| 3 | Kingboard 建滔积层板 (Hong Kong, China) | 12.5% | US$2.31bn | Former leader, now third |
| 4 | Nan Ya Plastics 南亚塑胶 (Taiwan, China) | 7.4% | — | |
| 5 | Doosan 斗山 (Korea) | 6.1% | — | Second in HDI-grade material at ~16% |
| 6 | ITEQ 联茂 (Taiwan, China) | 5.7% | — | |
| 7 | Panasonic 松下 (Japan) | 5.6% | — | Megtron family |
| 8 | TUC 台耀 (Taiwan, China) | 5.2% | — |
Three things follow for a buyer, and none of them require you to forecast the market.
One: the material line in your quote has a short chain behind it. Three firms hold 42% of world supply, and the sub-segment that matters for HDI is tighter still — roughly 70% to EMC, about 16% to Doosan, about 5% to Shengyi, on Prismark data as summarised by CICC in July 2026. If your stack-up calls for an HDI-grade material and your factory quotes one grade without offering an alternative, you are quoting a single point of failure.
Two: the grade name is the price tier. Standard FR-4, high-Tg FR-4 and low-loss hydrocarbon are three different materials with three different cost bases, and the difference is not visible in the phrase “FR-4”. Of the sixteen plants here, the two that name a grade name a mid-tier one — S1000-2M at TG170 is high-Tg, not low-loss. Nobody in the sixteen names a Megtron-class or Rogers-class material on a stack-up; the only plant that names those is outside the sixteen and under judicial auction (see below).
Three: the cost share is large enough that the grade decides the deal. Industry summaries derived from Prismark data put laminate at 30–40% of PCB production cost — the largest single material input. That is an industry estimate, not a figure this directory measured; treat it as a reason to ask, not as a number to quote in a negotiation.
The plant that names the most grades is the one you should not buy from. Guangtian Electronics (Suining) names three: Rogers RO4350B, Taconic MW-3350 and Shengyi SJ7036N. It is the most materially specific disclosure anywhere in this directory. The same profile records active enforcement, a judicial auction, and a second entity at the same address. Disclosure quality and factory quality are different variables, and this is the cleanest example of the two diverging.
Two share figures for one company are both right. Shengyi’s own 2025 annual report says rigid laminate sales have ranked second globally for several years, at 13.7% in 2024. Prismark’s 2025 figure is 13.2%. Neither number is wrong; the years are different, and a page that mixes them without labelling the year is the kind of page you should stop reading.
Zero of sixteen publish a date code. The grade on a pallet is only checkable on the material itself: the mill’s label, the date code, and an incoming-inspection record. No plant here publishes any of them. Everything above is about what plants say; the only way to close the gap is to ask for a photograph of the label on the next shipment.
生益科技 (600183) is the only laminate maker whose numbers a buyer can read in the same filing as a PCB plant’s. Its 2025 annual report, published 24 April 2026, is unusually specific, and four lines in it are useful even if you never buy a share.
| 2025 filing line | Figure | Change | Why a PCB buyer cares |
|---|---|---|---|
| Revenue | ¥28.431bn | +39.45% | The material layer grew faster than most of its customers |
| Net profit attributable | ¥3.334bn | +91.75% | Margin expansion upstream, at a time when PCB gross margins here run 14–31% |
| Laminate produced | 158.13m m² | +10.03% | Volume, so the capacity story is not just price |
| Laminate sold | 160.63m m² | +11.95% | Sold 2.50m m² more than it made — inventory drawn down 1.6% in a year |
The fourth row is the one worth pausing on. A producer selling more than it produces is supplying the difference from stock, which is what a tight market looks like from the inside. It is arithmetic on two numbers in the same filing, not an interpretation: 160.63 − 158.13 = 2.50 million m².
Two more lines are checkable against each other. Laminate and prepreg revenue was ¥17.774bn on 160.63m m² of laminate sold — about ¥110 per square metre, as an upper bound on realised laminate price, because the numerator includes prepreg that the denominator does not. And the Prismark figure of US$2.47bn converts to ¥17.8bn at roughly ¥7.2 to the dollar, which is how you know the ranking and the filing are counting the same thing rather than two different definitions of “laminate”.
The parent is also spending: a ¥5.2bn high-performance laminate project at Songshan Lake, announced with the same annual report, split across two phases — ¥2.37bn of production equipment, ¥1.57bn of buildings, ¥0.66bn of supporting equipment. Capacity added at the top of the grade range is the one upstream change that eventually reaches a buyer’s price, and it is the reason to re-ask the laminate question every year rather than once.
Everything else in this directory can be checked from outside with a browser: the legal entity in a credit registry, a certificate number in the issuing body’s database, a bill of lading, a robots.txt. Laminate cannot, and it is worth being explicit about why.
The grade is a property of the material, not of the company. Two plants with identical equipment, identical certifications and identical websites will produce different boards if one buys standard FR-4 and the other buys high-Tg, and nothing on either website will tell you which. The three artifacts that would settle it are all physical or internal: the mill’s label on the pallet, the factory’s incoming-inspection record for that lot, and a glass-transition measurement (DSC) or decomposition-temperature measurement (TGA) on a coupon cut from your board. A buyer can request all three. None of the sixteen plants publishes any of them.
This is the same finding as the layer-ceiling page, one level up the chain: the number that decides whether a plant can build your board is the number plants do not publish. A layer count is at least attached to a factory. A laminate grade is attached to a purchase order.
The same six items, worded for pasting into an RFQ email and with the replies to distrust, are on the laminate request sheet.
Not on the quotation summary. If the drawing says “FR-4” and nothing else, the widest cost range in the whole quote is still open.
Brand is not plant. Ask for the producing site, then ask whether the material is on allocation and when that allocation was last confirmed.
One price for one grade is a single point of failure. Two prices tell you what the upgrade costs and whether the first quote was mid-tier or bottom-tier.
The clause, not the promise. A validity window with a stated pass-through rule is the difference between a quote and an option on the material market.
Mill name, grade, date code, on the pallet, before it is cut. It costs the factory ten seconds and it is the only externally checkable evidence of grade in the entire transaction.
Given roughly 70% segment concentration in one supplier, ask which family the line is qualified on and what the second source is. A factory that answers “we use whatever is available” has answered the question.
Upstream of this page is the process layer: high-layer-count PCB (5 of 16 state a ceiling) · HDI and any-layer (9 of 16 claim it, 0 state a build-up limit) · IC substrate and SLP · metal-core, ceramic and heavy copper · FPC and rigid-flex · quick-turn prototype. Downstream, the commercial layer: what a Chinese PCB quote actually contains, where the laminate line item is one of nine. The material question also decides what a buyer can verify in twenty minutes, and what a factory visit should photograph at goods-in.