Mankun Technology (301132) hit the 20.01% limit-up on October 30, and the market read the limit-up as meaning the high growth in the third quarter can continue. The 2026 annual report will list quarterly figures; as long as fourth-quarter 2026 single-quarter revenue is lower than the third quarter, this reading is invalidated. Can this be verified before April 30, 2027?
- For 2025, the second half was 6.66% and the first half 10.04%.
- Gross margin rose from 15.75% to 27.00%, up 11.25 percentage points.
- Multilayer boards: 9.34 yi earned 1.17 yi, 12.57%.
- The latter leg is doing more of the work in the 27%.
Mankun Technology (301132) hit the 20.01% limit-up on October 30, and the market read the limit-up as meaning the high growth in the third quarter can continue. The 2026 annual report will list quarterly figures; as long as fourth-quarter 2026 single-quarter revenue is lower than the third quarter, this reading is invalidated. Can this be verified before April 30, 2027?
满坤科技(301132)10月30日涨停20.01%，市场把涨停读成三季度的高增长还能续。2026年报会列分季度数字，只要2026年四季度单季营收低于三季度，这个读法就废了，2027年4月30日前能验证吗？
The desk answersverifiedWalnut, from public filings
Mankun Technology's 20.01% limit-up on October 30 was reading the 27.00% gross margin for the third quarter alone in the Q3 report — the highest single-quarter figure since 2023, 11.25 percentage points higher than the 15.75% in Q2. That day's path also showed how big the disagreement was: flat open, intraday low of -3.96%, close locked at the limit-up price of 56.08 yuan, amplitude 23.97% — the open was reading "below expectations," the close was reading "can continue."
Can this "can continue" be verified before April 30, 2027? Yes. But using "fourth-quarter single-quarter revenue lower than third-quarter" as the criterion verifies nothing.
This criterion does not hold up arithmetically
Fourth-quarter 2025 single-quarter revenue was 4.25 yi, third-quarter 2026 was 7.24 yi. Suppose the fourth quarter does 6.66 yi — 8.0% lower than the third quarter, exactly triggering "lower than third quarter" — its year-on-year is +56.9%, almost the same as the third quarter's +56.48%.
A quarter that satisfies "fourth quarter lower than third quarter" can at the same time show no deceleration at all. The quarter-on-quarter denominator is the high point it just set itself; the year-on-year denominator is last year's fourth quarter at only 4.25 yi.
What should be watched is three things: the step-up in year-on-year growth, whether the gross margin holds, and whether excluding non-recurring items and net profit attributable to parent are on the same line.
Where exactly did the third quarter explode
The single-quarter figures are all calculated by us: current-period cumulative minus prior-period cumulative; gross margin uses single-quarter operating cost divided by single-quarter revenue — cumulative gross margins cannot be subtracted directly.
Third-quarter revenue was 1.89 yi higher quarter-on-quarter, a magnitude of 35.2%; gross margin rose from 15.75% to 27.00%, up 11.25 percentage points.
Third-quarter single-quarter gross profit was 1.956 yi, versus 0.866 yi in the same period last year, 1.090 yi more earned. Of the 2.61 yi increase in revenue, at last year's same-period 18.71% gross margin it contributed 0.489 yi, accounting for 45%; the gross margin itself rose 8.29 percentage points, contributing 0.600 yi, accounting for 55%. Volume and price each took half.
Where the increment landed is also clear: not subsidies, not investment income. Third-quarter non-recurring gains and losses totaled only 328.38 wan yuan; excluding non-recurring items, 1.2523 yi accounted for 97.4% of the 1.2852 yi attributable to parent. This round of improvement is entirely in "operating cost" falling relative to revenue — it is operational.
Under the 27% sits a line item that allocates almost no cost
The 2025 annual report notes broke down cost by product; that year's structure was like this:
| Product | Revenue (yi) | Gross margin |
|---|---|---|
| Double-sided board | 5.33 | 0.59% |
| Multilayer board | 9.34 | 12.57% |
| Single-sided board | 0.03 | 16.61% |
| Other | 1.77 | 99.52% |
| Total | 16.47 | 18.02% |
Double-sided boards: 5.33 yi revenue earned only 314 wan gross profit, 0.59%. Multilayer boards: 9.34 yi earned 1.17 yi, 12.57%. And the "other" line: 1.77 yi revenue, operating cost only 84.19 wan yuan, gross margin 99.52%, contributing 1.76 yi gross profit — 59% of the company's total gross profit of 2.97 yi.
Of the consolidated gross margin of 18.02%, nearly 60% of gross profit does not come from selling boards but from this line of revenue that allocates almost no cost. The material cost of the boards is already fully recorded in main business cost, so this money that drops down is pure gross profit. The annual report does not explain what this line contains, but from the cost composition it can only be external sales of edge scraps, waste material and the like that move with the copper price.
This point is critical to the judgment. Copper price rises, main business cost rises, and this line's revenue also rises: 2025 year-on-year +49.18%. So in the gross margin jump, how much is earned back by PCB processing itself and how much is lifted by this money — the Q3 report cannot break that out.
Peers leave Mankun no exclusivity
| Company | Q3 revenue (yi) | Revenue year-on-year | Gross margin | Gross margin year-on-year pp |
|---|---|---|---|---|
| 沪电股份 | 95.36 | 90.02% | 44.02% | 8.18 |
| 景旺电子 | 66.21 | 66.06% | 33.73% | 11.78 |
| 崇达技术 | 32.51 | 57.84% | 29.67% | 8.37 |
| 满坤科技 | 7.24 | 56.48% | 27.00% | 8.29 |
| 奥士康 | 21.27 | 44.95% | 33.37% | 13.68 |
| 世运电路 | 20.26 | 35.11% | 20.94% | -1.98 |
| 胜宏科技 | 68.31 | 34.30% | 34.17% | -1.03 |
Mankun ranks fourth in revenue growth, fifth in gross margin, fourth in year-on-year gross margin. Jingwang +11.78, Aoshikang +13.68, Chongda +8.37 — all larger than it. Five of the seven saw year-on-year gross margin rises, four had revenue growth above 55% — this third-quarter wave is the whole mid-to-low-end PCB price pass-through plus compute-power orders landing at the same time, not Mankun's story alone.
But it is not a broad rise across the board either. Shiyun Circuit's gross margin was -1.98 percentage points year-on-year, net profit attributable to parent -2.04%; Shenghong Technology -1.03 percentage points, net profit attributable to parent -52.17%. Shiyun also makes automotive electronics boards, and its gross margin did not rise — the automotive board line did not see broad price pass-through. Mankun stands on both legs at once, automotive electronics and AI server power; the latter leg is doing more of the work in the 27%. On the October 29 evening conference call, the company put it directly:
Benefiting from the high prosperity of the AI compute industry, market order demand in the AI compute hardware PCB industry is strong, the company's AI server power-related business is growing, and the customer base continues to expand.
In the same record there is another sentence explaining why the first half looked bad and why the third quarter looked good:
In the first half, affected by factors such as the capacity ramp-up at Ji'an Plant 3 and increased R&D investment, the company's profit came under phased pressure. Entering the third quarter... Plant 3's capacity utilization rose in steps, reaching a relatively good level by the end of September, economies of scale gradually emerged, phased pressures such as depreciation and R&D weakened at the margin, and gross margin recovered quarter-on-quarter.
When it can be seen, and what to see
| Time point | Document | Can verify | Cannot verify |
|---|---|---|---|
| End of December 2026 to early January 2027 | Shenzhen Stock Exchange annual report scheduled disclosure timetable | Exact disclosure date | Figures |
| Before January 31, 2027 | 2026 annual performance forecast | Full-year net profit range | Revenue, gross margin |
| Before April 30, 2027, most likely April 29 | 2026 annual report quarterly table | Q4 revenue, net profit attributable to parent, excluding non-recurring items, cash flow | Split between main business and other businesses |
| Same day | 2027 Q1 report | Whether Q1 gross margin declines for a second consecutive quarter | First-half business structure |
The premise is fine. The 2025 annual report does have a section "VI. Main Financial Indicators by Quarter," listing all four quarters' operating revenue, net profit attributable to the parent, net profit excluding non-recurring items, and operating cash flow, with a note that there is no material difference from the quarterly reports already disclosed. The 2026 annual report will be presented in the same format—no need to bet on it.
The cadence is also steady: both years' third-quarter reports came on 10-30, and the 2025 annual report and the 2026 first-quarter report were released on the same day, 2026-04-29. At this cadence, the fourth-quarter standalone figures for 2027 will most likely arrive on April 29, at the latest April 30.
The third-quarter report does not break out main business versus other business, so right now we cannot calculate how much each of the two accounts for within the 27%; the annual report's quarterly table also gives only four lines of totals. What can be broken out is "second half minus first half": the annual report's product-segment note gives full-year main business revenue and cost, the semi-annual report gives the first half's, and subtracting the two gives the second half.
This method was run once for 2025: full-year main business gross margin 8.23%, first half 10.04%, back-calculating the second half to 6.66%. The numbers work because the revenue and cost of the four lines in the annual report's product-segment note sum exactly to the income statement's 16.4711 yi revenue and 13.5031 yi cost. So the 2026 second-half main business gross margin can be calculated, and that is harder than any single-quarter revenue ranking.
The strongest opposing point
The company itself wrote two sentences in the semi-annual report that, taken together, are exactly what I am worried about:
As of the end of the reporting period, the company's orders on hand grew relatively fast year-on-year, mainly due to the expectation of rising raw material prices; downstream customers' willingness to stock up strengthened, and the phenomenon of orders being placed early was relatively obvious.
Customers place orders early for fear of price increases. In the third-quarter report, inventory was 3.66 yi, up 47.91% from 2.48 yi at the beginning of the year, with the reason given as "raw material prices rising, raw material stocking increased, and at the same time orders undertaken increased, work in progress increased."
With customers placing orders early on one side and the company stocking up materials early on the other, both ends are moving early, so one end must be borrowing demand from later. A quarter-on-quarter decline in the fourth quarter is very likely not demand deteriorating but early buying being finished; conversely, if it still moves up quarter-on-quarter, the scale of early pulling was not as large as imagined. Neither case is the same thing as "whether high growth can continue."
Contract liabilities were 2366.77 wan yuan, versus 36.18 wan yuan at the beginning of the year, up 6440.85%, with the reason given as "increase in advance payments received from customers." The direction is right, but the absolute amount is only equivalent to 3.3% of third-quarter revenue—there is evidence of order locking, but the magnitude cannot support a new level.
Another point is harsher: the 27% may be point-in-time arbitrage of "low-priced inventory cost + already-raised selling price." What was stockpiled was material from before the price increase, and what is sold is board from after the price increase; this spread will disappear as the low-priced inventory is used up. Once this holds, "can continue" is false, and it has nothing to do with whether fourth-quarter revenue is lower than third-quarter revenue.
Four falsification tests
One: before January 31, 2027, the company will very likely issue its 2026 annual performance forecast. Attributable net profit for the first three quarters is already 1.725 yi; for the full year, as long as it exceeds 1.792 yi (1.5 times 2025's 1.195 yi), the mandatory forecast threshold is triggered, and fourth-quarter standalone needs only 672 wan. This company follows the rules accurately: when 2025 semi-annual net profit was +62.32% year-on-year, it issued a forecast on July 21; when the full year was +12.19%, it did not. If the forecast full-year attributable net profit is below 2.60 yi (corresponding to fourth-quarter standalone of less than 0.87 yi, down more than one-third from third quarter's 1.285 yi), "can continue" is void on the spot.
Two: in the "Main Financial Indicators by Quarter" section of the April 30, 2027 annual report, if fourth-quarter standalone revenue is below 5.36 yi—not below the third quarter, but falling back to the second quarter's platform—it means the release from Plant No. 3 was a one-off order pulse, not a capacity step-up.
Three: in the same table, if fourth-quarter standalone net profit excluding non-recurring items is below 0.80 yi (64% of third quarter's 1.252 yi), it means the two dividends of price pass-through and low-priced inventory have been used up, and gross margin is falling faster than revenue.
Four: use the annual report's product-segment note minus the semi-annual report to calculate the 2026 second-half main business gross margin. For 2025, the second half was 6.66% and the first half 10.04%. If 2026 second half is below 12%, then the portion of the third quarter's 27% supported by the "other" line is the main cause, and PCB processing itself has not truly recovered.
Therefore
It can be verified, by April 30, 2027 at the latest, and at the two-year cadence most likely April 29. The late-January performance forecast will give an answer on profit direction, but it will not show revenue, nor gross margin.
When verifying, do not use "fourth-quarter revenue lower than third-quarter revenue" as the yardstick. The third quarter's 7.24 yi is itself a step-up; after standing on it, a quarter-on-quarter pullback is the normal shape of a step-up, not a loss of speed. What to watch is: whether fourth-quarter year-on-year growth falls from 56% to below 40%, and whether second-half main business gross margin recovers above 12%.
Source: Ji'an Mankun Technology Co., Ltd. "2026 Third Quarter Report", announcement no. 2026-1054, 2026-10-30; "2026 Semi-Annual Report", 2026-08-26; "2025 Annual Report", 2026-04-29; "2025 Semi-Annual Earnings Forecast", announcement no. 2025-1013, 2025-07-21; "October 29, 2026 Investor Relations Activity Record", no. 2026-005, 2026-10-30. Peer single-quarter data are taken from each company's 2026 and 2025 third quarter reports, self-restated as "current period cumulative minus prior period cumulative".
满坤科技10月30日那个 20.01% 的涨停，读的是三季报里第三季度单季 27.00% 的毛利率——2023年以来单季最高，比二季度的 15.75% 高了 11.25 个百分点。当天的路径也说明分歧有多大：平开、盘中最低 -3.96%、收盘封在涨停 56.08 元，振幅 23.97%——开盘读的是不及预期，收盘读的是能续。
这个"能续"能不能在2027年4月30日前验掉？能。但拿"四季度单季营收低于三季度"当判据，验不出任何结论。
这条判据在算术上就不成立
2025年四季度单季营收 4.25 亿，2026年三季度 7.24 亿。假设四季度做到 6.66 亿——比三季度低 8.0%，正好触发"低于三季度"——它的同比是 +56.9%，跟三季度的 +56.48% 几乎一样。
一个满足"四季度低于三季度"的季度，可以同时完全没减速。环比的分母是自己刚创的高点，同比的分母是去年那个只有 4.25 亿的四季度。
该看的是三样：同比增速的台阶、毛利率守不守得住、扣非跟归母是不是一条线。
三季度到底爆在哪
单季数字都是自己算的：本期累计减上期累计；毛利率用单季营业成本除单季营收，累计毛利率不能直接相减。
三季度营收环比多出 1.89 亿，幅度 35.2%；毛利率从 15.75% 抬到 27.00%，多了 11.25 个百分点。
三季度单季毛利额 1.956 亿，去年同期 0.866 亿，多赚 1.090 亿。收入多出来的 2.61 亿，按去年同期的 18.71% 毛利率算，贡献 0.489 亿，占 45%；毛利率自己抬了 8.29 个百分点，贡献 0.600 亿，占 55%。量和价各占一半。
增量落在哪一科也清楚：不是补贴，也不是投资收益。三季度非经常性损益合计只有 328.38 万元，扣非 1.2523 亿占归母 1.2852 亿的 97.4%。这波改善全在"营业成本"相对收入下降上，是经营性的。
27% 下面压着一行不摊成本的东西
2025年报附注按产品拆过成本，那一年的结构是这样的：
| 产品 | 收入(亿) | 毛利率 |
|---|---|---|
| 双面板 | 5.33 | 0.59% |
| 多层板 | 9.34 | 12.57% |
| 单面板 | 0.03 | 16.61% |
| 其他 | 1.77 | 99.52% |
| 合计 | 16.47 | 18.02% |
双面板 5.33 亿收入只赚 314 万毛利，0.59%。多层板 9.34 亿赚 1.17 亿，12.57%。而"其他"这一行 1.77 亿收入，营业成本只有 84.19 万元，毛利率 99.52%，贡献 1.76 亿毛利——占公司全部毛利 2.97 亿的 59%。
合并毛利率 18.02% 里，将近六成毛利不来自卖板子，来自这一行几乎不摊成本的收入。板子的材料成本已全额记在主营成本里，所以这块钱落下来就是纯毛利。年报没解释这一行的内容，但从成本构成看，只能是边角料、废料这类随铜价走的对外销售。
这条对判断很关键。铜价涨，主营成本涨，这一行的收入也涨：2025年同比 +49.18%。所以毛利率跳升里，多少是PCB加工自己赚回来的、多少是这块钱涨上来的，三季报拆不出来。
同行没给满坤留独家
| 公司 | Q3收入(亿) | 收入同比 | 毛利率 | 毛利率同比pp |
|---|---|---|---|---|
| 沪电股份 | 95.36 | 90.02% | 44.02% | 8.18 |
| 景旺电子 | 66.21 | 66.06% | 33.73% | 11.78 |
| 崇达技术 | 32.51 | 57.84% | 29.67% | 8.37 |
| 满坤科技 | 7.24 | 56.48% | 27.00% | 8.29 |
| 奥士康 | 21.27 | 44.95% | 33.37% | 13.68 |
| 世运电路 | 20.26 | 35.11% | 20.94% | -1.98 |
| 胜宏科技 | 68.31 | 34.30% | 34.17% | -1.03 |
满坤收入增速第四、毛利率第五、毛利率同比第四。景旺 +11.78、奥士康 +13.68、崇达 +8.37，都比它大。七家里五家毛利率同比上升，四家收入增速过 55%——三季度这波是整个中低阶PCB顺价加算力订单同时落地，不是满坤一家的故事。
但也不是全线普涨。世运电路毛利率同比 -1.98 个百分点、归母 -2.04%；胜宏科技 -1.03 个百分点、归母 -52.17%。世运同样做汽车电子板，毛利率没涨——汽车板这条线并没有普遍顺价。满坤同时踩了汽车电子和AI服务器电源两条腿，27% 里后面那条使的劲更大。10月29日晚的电话会上，公司说得直接：
受益于AI算力产业高景气发展，AI算力硬件PCB行业市场订单需求旺盛，公司AI服务器电源相关业务增长，客户群体持续拓展。
同一份记录表里还有一句，解释了上半年为什么难看、三季度为什么好看：
上半年受吉安三厂产能爬坡、研发投入增加等因素影响，公司利润阶段性承压。进入第三季度……三厂产能利用率阶梯式提升，截至9月末已达到较好水平，规模效应逐步显现，折旧、研发等阶段性压力边际弱化，毛利率环比修复。
什么时候能看到，看到什么
| 时点 | 文件 | 能验证 | 不能验证 |
|---|---|---|---|
| 2026年12月底至2027年1月初 | 深交所年报预约披露时间表 | 确切披露日 | 数字 |
| 2027年1月31日前 | 2026年度业绩预告 | 全年净利区间 | 营收、毛利率 |
| 2027年4月30日前，大概率4月29 | 2026年报分季度表 | Q4营收、归母、扣非、现金流 | 主营与其他业务拆分 |
| 同日 | 2027年一季报 | Q1毛利率是否连续第二季回落 | 上半年业务结构 |
前提没问题。2025年报里确实有"六、分季度主要财务指标"一节，把第一到第四季度的营业收入、归母净利、扣非净利、经营现金流四行逐季列全，并注明与已披露季度报告不存在重大差异。2026年报照同一格式列，不用赌。
节奏也稳：两年三季报都是 10-30，2025年报和2026年一季报同一天 2026-04-29 发的。按这个节奏，2027年拿到四季度单季数大概率4月29日，最晚4月30日。
三季报不拆主营和其他业务，现在算不出 27% 里两块各占多少；年报的分季度表也只给四行总数。能拆的是"下半年减上半年"：年报分产品附注给全年主营收入与成本，半年报给上半年的，两者相减就是下半年。
这个方法2025年走过一遍：全年主营毛利率 8.23%，上半年 10.04%，倒算下半年 6.66%。数字走得通，是因为年报分产品附注里四条线的收入成本加总，正好等于利润表的 16.4711 亿收入和 13.5031 亿成本。所以2026年下半年主营毛利率算得出来，这比任何单季营收排名都硬。
反方最强的一条
公司自己在半年报里写了两句话，合起来就是我担心的东西：
截至报告期末，公司在手订单同比增长较快，主要受原材料价格上涨预期影响，下游客户备货意愿增强、订单前置现象较为明显
客户怕涨价提前下单。三季报里存货 3.66 亿，比年初 2.48 亿增 47.91%，理由写的是"原材料价格上涨，原材料备货增加，同时承接订单增加，在产品增加"。
一边客户前置下单，一边公司前置囤料，两头都在提前，就有一头在透支后面的需求。四季度环比掉下来很可能不是需求变差，是提前买完了；反过来若环比还往上走，说明前置规模没想象中大。这两种都跟"高增长能不能续"不是一回事。
合同负债 2366.77 万元，年初 36.18 万元，增 6440.85%，理由是"收到客户预付货款增加"。方向对，但绝对额只相当于三季度营收的 3.3%——订单锁定的证据有，量级撑不起新台阶。
另一条更狠：27% 可能是"低价库存成本 + 已涨价售价"的时点套利。囤的是涨价前的料，卖的是涨价后的板，这段差价会随低价库存耗尽而消失。这条一旦成立，"能续"就是假的，而且跟四季度营收是否低于三季度完全无关。
四条证伪
一、2027年1月31日前，公司大概率要发2026年度业绩预告。前三季度归母已 1.725 亿，全年只要超过 1.792 亿（2025年 1.195 亿的 1.5 倍）就触发强制预告线，四季度单季做到 672 万就够。这家公司跟规则跟得准：2025年半年报净利同比 +62.32% 时它在 7 月 21 日发了预告，全年 +12.19% 时没发。若预告的全年归母低于 2.60 亿（对应四季度单季不足 0.87 亿，比三季度 1.285 亿掉三分之一以上），"能续"当场作废。
二、2027年4月30日年报"分季度主要财务指标"里，四季度单季营收若低于 5.36 亿——不是低于三季度，是掉回二季度那个平台——说明三厂释放的是一次性订单脉冲，不是产能台阶。
三、同一张表里四季度单季扣非若低于 0.80 亿（三季度 1.252 亿的六四折），说明顺价和低价库存两个红利已用完，毛利率回落快过收入。
四、用年报分产品附注减半年报，算出2026年下半年主营毛利率。2025年下半年是 6.66%，上半年 10.04%。若2026年下半年低于 12%，三季度那 27% 里靠"其他"那一行撑起来的成分就是主因，PCB加工本身没真正修好。
所以
能验，最晚2027年4月30日，按两年节奏大概率4月29日。1月底的业绩预告会给一个利润方向的答案，但看不到营收，也看不到毛利率。
验的时候别用"四季度营收低于三季度"这把尺子。三季度那 7.24 亿本身就是一个台阶，站上去以后环比回落是台阶的正常形态，不是失速。要盯的是：四季度同比有没有从 56% 掉到 40% 以下，以及下半年主营毛利率有没有修到 12% 以上。
- 01吉安满坤科技股份有限公司《2026年第三季度报告》，公告编号 2026-1054，2026-10-30
- 02《2026年半年度报告》，2026-08-26
- 03《2025年年度报告》，2026-04-29
- 04《2025年半年度业绩预告》，公告编号 2025-1013，2025-07-21
- 05《2026年10月29日投资者关系活动记录表》，编号 2026-005，2026-10-30。同行单季数据取自各公司2026年、2025年三季度报告，按"本期累计减上期累计"自行还原
## 27% 是两个数叠出来的
三季报不拆主营和其他业务，半年报拆。2026 年半年报第 95 页那张"营业收入和营业成本"：主营收入 8.3372 亿、成本 7.8535 亿，毛利 0.4837 亿，毛利率 5.80%；其他业务收入 1.0866 亿，成本一栏空着，毛利就是 1.0866 亿。合并毛利 1.5703 亿，那一行占 69.2%。
上半年将近七成毛利不来自卖板子。2025 年报第 139 页同一个格式：主营 14.7052 亿 / 13.4947 亿，毛利率 8.23%；其他 1.7664 亿 / 84.19 万，毛利率 99.52%，占全年毛利 2.9684 亿的 59.5%。
这一行是什么，公司没写过一句。从成本接近零倒推，只能是生产时已全额计进主营成本、再单独卖出去的东西，边角料、废料一类。它随产量走：2025 年生产量 +35.09%，这一行收入 +49.18%。
主营用"全年减上半年"倒算：2025 上半年 10.04%，下半年 6.67%，2026 上半年 5.80%。三季报不拆，按这一行占主营收入 13%（上半年 13.03%）推，三季度那块约 0.83 亿，剩下主营 6.41 亿、毛利 1.12 亿，主营毛利率约 17.5%；占比取 11.5% 到 15%，区间 16% 到 19%。
```chart
type: line
title: 主营毛利率：两个半年，加一个单季推算
unit: %
labels: 2025H1, 2025H2, 2026H1, 2026Q3
values: 10.04, 6.67, 5.80, 17.5
```
合并 27% 像新台阶，可比的是主营 17.5%——2025 全年才 8.23%。一家双面板毛利率只有 0.59% 的公司，主营做到 17.5%，历史上没参照。
## 主营那 17% 是怎么弹回来的
2025 年报第 13 页：
> 原材料成本上行压力向终端下游传导存在合理时间滞后性；另一方面三厂募投项目产能爬坡阶段固定成本较高
一句是料涨了售价没跟上，一句是三厂刚投产摊不开。三季度是这两句的反面。
利用率这条腿有硬证据，但它到顶了。三季度单季研发费用 1968.43 万，占当季营收 2.72%；去年同期 1511.95 万，占 3.27%。绝对额 +30.2%，费用率反降 0.55 个百分点。
绝对额涨、费用率降，这是收入摊薄，不是结构升级——高阶 HDI 真大幅放量，研发费用率该往上走。而三厂"截至 9 月末已达到较好水平"，摊薄这件事三季度基本做完了。
顺价这条腿方向是坏的，还有保质期。直接材料占营业成本 69.21%（2025 年报），材料成本相当于主营收入的 63.5%，材料涨 1%，主营毛利率掉 0.64 个百分点。2026 上半年主营只剩 5.80%，料价再涨 9% 就吃干净。
三季报里营业成本 +32.74%，公司写的理由是"原材料价格上涨"——料涨的年份，成本增速还比收入慢 3.58 个百分点，这 3.58 个点就是全部改善。不是需求给的。
两条腿之外，还有一笔往下压的：折旧。在建工程 1.8213 亿，比年初增 43.04%，理由是"已到厂未验收设备增加"。2025 年折旧 1.0031 亿、今年三季度末固定资产 10.36 亿，比例 9.7%；这 1.82 亿一年折 1763 万，一季约 440 万，对 7.24 亿营收是 0.6 个百分点。不大，方向朝下。
## 同行给的不是"行业好"，是一条门槛
| 公司 | Q3收入同比 | Q3毛利率 | 毛利率同比pp |
|---|---|---|---|
| 沪电股份 | 90.02 | 44.02 | 8.18 |
| 景旺电子 | 66.06 | 33.73 | 11.78 |
| 崇达技术 | 57.84 | 29.67 | 8.37 |
| 满坤科技 | 56.48 | 27.00 | 8.29 |
| 奥士康 | 44.95 | 33.37 | 13.68 |
| 世运电路 | 35.11 | 20.94 | (1.98) |
| 胜宏科技 | 34.30 | 34.17 | (1.03) |
按收入增速排，规律很干净：44.95% 及以上的五家毛利率同比全升，掉到 35% 一线的两家全降。
七家面对的是同一根铜价。所以毛利率扩不扩张，分野不在"是不是 PCB"，在产能利用率有没有跨过某个点。收入增速只是利用率的替身：厂子满了固定成本摊薄，毛利率就上；厂子没满，料涨价是纯压力。满坤 +56.48% 站在门槛上面。
## 反方最强的一条
世运电路做汽车板，收入 +35.11%，毛利率同比 -1.98 个百分点，归母 -2.04%；胜宏科技做高阶 HDI 和 AI 板，收入 +34.30%，毛利率 -1.03 个百分点，归母 -52.17%。同样铜价同样料，这两家没弹。
如果 27% 是行业顺价窗口给的，就该一起弹。只有满坤弹，说明弹的是它自己的东西：AI 服务器电源订单、光模块送样转小批量、高阶 HDI 定点。定点一旦导入，2027 年继续放量是合同约束，不是一个价格窗口能关掉的。再加一条，10 月 13 日公司把回购价上限从 48.48 元调到 86.91 元，一次调高 79%。
回应三句。第一，世运和胜宏的问题不是产品是利用率，收入增速掉在 35% 一线正卡在门槛下面，拿它俩证明"这不是行业窗口"，证明不了。
第二，回购不构成证据只构成表态：三季报写"截至 9 月 30 日"回购 575,124 股、2,637.45 万元，10 月 29 日记录表写"截至 10 月 13 日"还是同样的股数和金额，两个时点一模一样，中间两周一股没买，而这两周股价正在往上；之后买没买要等 11 月初的公告，现在查不到。
第三，有一条我认：应收周转天数从去年三季度的 113.3 天降到今年三季度的 100.5 天，客户在加快付钱，不是公司放宽账期塞货。
它证明的是订单真、回款真。证明不了 17.5% 能留住。
## 2027 年 4 月 30 日能验什么
| 时点 | 文件里哪张表 | 算出来的东西 |
|---|---|---|
| 2027-01-31 前 | 2026 年度业绩预告 | 全年归母；低于 2.60 亿即 Q4 单季不足 0.87 亿 |
| 2027-04-30 前 | 2026 年报"六、分季度主要财务指标" | Q4 单季营收、归母、扣非 |
| 同日 | 年报附注主营与其他业务两行、分产品表 | 下半年主营毛利率、双面板毛利率 |
| 同日 | 年报存货附注四行 | 原材料还囤不囤，成品积不积压 |
第二行最关键。算法是：2026 年报的主营收入与成本，减掉半年报的主营收入与成本，就是下半年主营毛利率。这个算法在 2025 年报上走过一遍——全年 8.23%，上半年 10.04%，倒算下半年 6.67%；而且年报分产品附注四条线加总，正好等于利润表的 16.4715 亿和 13.5031 亿，加得上。
拿 2026 下半年主营毛利率比 2025 下半年的 6.67%，同半年同算法，这才叫比。
验不了的直说：三季报不拆主营，四季度单季主营毛利率拿不出来，只能拿"下半年"这个半年数，里面混了 Q3 和 Q4。分辨率是半年不是单季。要单季得等 2027 年一季报和半年报，那就到 8 月，越过 4 月 30 日了。
## 四条证伪
一、2027 年 1 月 31 日前全年归母低于 2.60 亿（四季度单季不足 0.87 亿，比三季度 1.2852 亿掉三分之一），"能续"作废。触发线算得出来：前三季归母 1.7250 亿，全年超过 1.792 亿（2025 年 1.1948 亿的 1.5 倍）就要强制预告，四季度做到 672 万就够。
二、2026 下半年主营毛利率落在哪一档。高于 15%，意味着四季度主营还守在 13% 以上，那就不是时点差是台阶，我错；低于 10%，意味着四季度主营掉回 2% 上下，三季度整段是脉冲。我押它落在 10% 到 15% 之间。
三、2026 年报分产品表里，双面板毛利率还趴在 1% 附近、而多层板和高阶板大幅抬升，那是结构升级，我错了；反过来，双面板从 0.59% 抬到 8% 以上，就是全行业成本推动的普涨。双面板最没技术含量，它涨价只能是供需和料价推的，这条区分得干净。
四、反向的：2026 年末原材料余额高于 2025 年末的 7,058.71 万，说明囤料还在继续，低价库存的红利没走完，四季度毛利率可能不降反升；如果原材料回落而在产品、库存商品堆起来，那才是订单没跟上投料——真减速，而且比毛利率下滑更晚被看见。
## 回到那两个数
四季度营收低于三季度，4 月 30 日前一定验得到。但它跟"能不能续"之间没有一条线连着。
2025 年四季度营收 4.2455 亿。2026 年四季度做到 6.66 亿，同比 +56.9%，跟三季度 +56.48% 几乎一样，而它低于三季度的 7.2423 亿。第一版算过。真正的门槛不在 7.24 亿，在 5.94 亿——那是四季度同比 +40% 的位置，也是同行里毛利率开始转负的那条线。掉到 5.94 亿以下，才叫掉档。
即便营收守住 7 亿，主营毛利率也可能从 17.5% 掉回 10%。弹它的两条腿，一条已到顶，一条有保质期。2027 年营收和毛利率会分开走，这才是"能不能续"。
一句话：4 月 30 日能给三个数——Q4 单季营收、下半年主营毛利率、双面板毛利率。第一个回答有没有掉档，后两个回答掉的是周期还是结构。涨停那天市场买的是后两个的答案，而它们要等五个月。
来源：吉安满坤科技股份有限公司《2026年第三季度报告》2026年10月30日，公告编号 2026-1054；《2026年10月29日投资者关系活动记录表》2026年10月30日，编号 2026-005；《2025年年度报告》2026年4月29日；《2026年半年度报告》2026年8月26日。同行数据取自沪电股份、景旺电子、崇达技术、奥士康、世运电路、胜宏科技各自2025、2026年三季报利润表，单季数用前三季累计减上半年累计。回购与在建工程数字出自上述定期报告。