688019.SHG

Anji is growing — and taking share of — an existing, secularly-expanding pie, not creating a new market. Its growth is domestic-substitution share capture inside well-established materials categories, which can be large but is bounded and contested; this is a bigger slice of a growing cake, not blue-sky category creation.

The pie is real and widening. The report cites a 2025 global semiconductor market of $795.6bn (up 26.2% on a WSTS basis) and SEMI's trillion-dollar-by-2026 expectation, with advanced nodes, new materials and advanced packaging raising CMP step-count and electronic-chemical demand per wafer. So Anji's addressable process wallet structurally expands as compute gets more performance-dense and packaging-heavy.

But the categories themselves are defined and modest. The global CMP-slurry market is only about USD 2.1bn in 2025 on a TECHCET basis for slurry alone (roughly $3.6bn including pads), growing high-single-digits. Anji's CMP-slurry revenue of CNY 2.040bn (about USD 285m) maps cleanly onto its disclosed roughly 13% global share — a useful cross-check. That share rose from about 7% in 2022 to 13% in 2025, while functional wet chemicals reached about 6% global share. The entire engine is displacement of incumbents (Entegris, Fujimi, Cabot) plus the "3+1" adjacency expansion into cleaning, plating and upstream raw materials.

The honest Baillie read: this is a strong share-gain runway, not an unbounded TAM. Headroom exists — 13% can plausibly travel toward 20-25% in CMP, and wet chemicals taps a larger adjacent pool — but Anji is filling a finite, contested category rather than inventing demand that did not previously exist. The ceiling is set by global CMP and wet-chemicals TAM and by how much a fast-follower can wrest from entrenched leaders, which is meaningful but well short of the "create a new market" archetype Baillie prizes most.

Yes — doubling revenue within five years is realistic and probably conservative, driven by volume and new business rather than price, and the structural growth survives stripping out semiconductor-cycle beta.

The base rate is reassuring. Revenue went CNY 1.238bn (2023) to CNY 1.835bn (2024) to CNY 2.504bn (2025), already roughly doubling in two years (about 42% CAGR). To merely double again by 2030, from CNY 2.504bn to about CNY 5bn, requires only about 15% CAGR — well under half the recent pace. Even with material deceleration, the doubling bar is clearable.

The drivers are the right ones. Growth is led by volume and share, not price: CMP-slurry revenue grew 32.06% on global share moving from about 7% (2022) to 13% (2025), while functional wet chemicals surged 63.73% to CNY 452.86m. Margins eased slightly (the IC-materials segment gross margin fell 1.66 percentage points), confirming this is not a price-led story. New business — wet chemicals plus electroplating chemistry now in mass production — adds incremental revenue lines beyond the CMP core.

Stripping the cycle is the key honesty test. Part of the 2024-25 surge rode an upcycle (global semis up 26.2%). But the domestic-substitution share-gain component is cycle-independent: even in a flat or down market, Anji can keep converting qualified-in positions into wallet share at mainland fabs, where the report notes remaining substitution room at foreign-background fabs plus overseas expansion in Taiwan and Singapore. A structural mid-teens-or-better growth rate is therefore defensible without assuming a perpetual boom.

The Baillie caveat: doubling is the floor, not the ceiling. For the stock — at about 88 times earnings — a mere doubling of revenue with roughly stable margins doubles profit over five years, which does not by itself justify the current multiple. The structural growth is genuine; the binding question is whether it is fast enough for the price, not whether it is real.

The second curve genuinely exists today and is funded — functional wet chemicals is already material and growing far faster than the CMP core, with electroplating chemistry as a real, if still small, third leg. This is one of Anji's strongest Baillie dimensions.

The second engine is already in the P&L, not on a slide. Functional wet electronic chemicals grew 63.73% to CNY 452.86m in 2025 at a 50.00% gross margin, reached about 6% global share, and now contribute roughly 18% of revenue — driven, management says, by post-etch and post-polish cleaning. This is the fastest-growing line in the business, not a theoretical option.

A third leg is emerging behind it. Electroplating liquids and additives are smaller in revenue but have already entered mass production in damascene and advanced-packaging use cases, and the "+1" of the "3+1" platform — upstream core raw-material localization — adds a vertical-integration curve. The pipeline therefore runs CMP to wet chemicals to plating to upstream, each an adjacency sold into the same qualified fabs.

It is demonstrably funded. R&D was CNY 444.70m in 2025 (up 33.64%, 17.76% of revenue); capex rose to CNY 355.38m with construction-in-progress of CNY 226.20m; a CNY 830.5m convertible bond was completed in April 2025; and the Ningbo 20,000-ton-per-year project plus the Shanghai Chemical Industry Park base (structurally topped out) are under way. Operating cash flow of CNY 439.77m nearly covers capex, so the build-out is financed from internal cash plus a dedicated growth instrument, not balance-sheet desperation.

The honest qualifier: these are adjacencies — more chemistry sold into the same customers — rather than a wholly orthogonal S-curve, and plating is still tiny in revenue terms. But on the literal Baillie test of whether the second curve exists today and is funded, Anji answers an unambiguous yes, with the early data already visible.

Anji has a real, qualification-based moat that is more likely to widen than narrow over three-to-five years through platform breadth — but it is a wide-enough, cumulative moat, not an impregnable one: Anji remains a fast-follower below the global frontier on leading nodes.

The moat sources are genuine. First, qualification lock-in: the report describes a product cycle of roughly one-to-two years of R&D, more than a year of testing, and more than a year of rollout before scaled sales, after which switching a qualified supplier is slow, expensive and operationally risky for the fab. Multiple copper-barrier products became first-choice suppliers at new clients in 2025. Second, cumulative IP: 308 granted invention patents and 383 applications, with 21 grants added in 2025 — technical, not merely relational. Third, platform breadth: the "3+1" coverage across CMP, cleaning, etch, plating and upstream deepens account control, the process-embedded flywheel the report scores highly.

Why it should widen: incumbency leads to line-item expansion, which leads to broader platform coverage, which deepens customer dependence — and each new qualification enlarges the option set. National enterprise-technology-center status and a Shanghai single-champion designation reinforce the technical base.

Now the honest narrowing risks. Anji is a domestic-substitution share-gainer, a fast-follower against the global benchmark Entegris (after the CMC Materials acquisition) and Japan's Fujimi, rather than a frontier monopolist; outside analysis has flagged that its business has been heavily bound to SMIC and that a technology-generation gap to the leaders exists. Domestic peers such as 300236.SHE and 603078.SHG are chasing the same substitution wallet, and the 1.66-percentage-point segment gross-margin slip is a small competitive tell. So the moat is wide within the domestic-substitution lane and widening across product lines, but it is defended by qualification and breadth more than by an unassailable technology lead.

Anji shows solid reinvention DNA for a materials company — its repeated expansion from a single CMP product into wet chemicals, plating and upstream materials is itself proof of repeatable renewal — but evidence on how it handles mistakes and bad news is thin, so this dimension is inferred from strategy and R&D rather than stress-tested by a public crisis.

The reinvention track record is concrete. Anji did not stay a one-product specialist: it moved CMP across copper, copper-barrier, dielectric, tungsten, ceria-based, metal-gate, substrate and advanced-packaging use cases, then built functional wet chemicals to about 6% global share and pushed electroplating into mass production. Those are several successful S-curve extensions inside its domain, which is the relevant form of reinvention for a process-chemistry platform.

The engine behind it is well-funded and continuous. R&D ran at CNY 444.70m in 2025 (17.76% of revenue), with 106 new invention applications and 21 grants added in the year. Management's stated operating philosophy — "small steps, fast run, rolling planning," building physical capacity ahead of demand and then flexing line investment — is an adaptive, optionality-preserving posture rather than a single irreversible bet.

On mistakes and bad news, the evidence gap deserves a plain statement. The core CMP franchise faces low disruption risk because the underlying physics is entrenched, so reinvention here is mostly about adjacency expansion, which Anji demonstrably does well. But I found no major public misstep — a recall, a failed node transition, a governance scandal — against which to judge how the company actually metabolizes failure. The report's 100% complaint-closure and satisfaction figures are directional at best and should not be over-read. So the verdict is constructive but partly inferential: strong demonstrated ability to extend into new products and nodes; unproven, simply because untested, on absorbing a genuine crisis.

Strong on founder leadership and long-term operational commitment, but only medium on deep owner-binding by Baillie's skin-in-the-game standard. Founder-CEO Wang Shumin has run Anji since inception and invests aggressively for the long term, yet the listed company officially has no actual controller, her look-through economic stake is single-digit-percent, and her direct on-market holding is only about 0.05%. This is a returnee-scientist, ownership-dispersed company, not a deeply founder-family-bound one.

The leadership is genuinely long-term. Dr. Shumin Wang (王淑敏) is a returnee scientist who founded Anji and has served as its chairman and general manager from the early days (director and CEO since 2004; company established 2006; STAR-Market IPO in 2019). Operational long-termism is evident: R&D at 17.76% of revenue, deliberately building raw-material inventory (CNY 492.21m to CNY 865.94m) and capacity (the Ningbo 20,000-ton project, the Shanghai park) ahead of demand, and 313 staff — about 39.87% of headcount — under stock incentives that align the technical team. She is clearly willing to spend today for years three-to-ten.

But the ownership binding is loose, and the web-verified structure is the load-bearing point. The largest holder is Anji Cayman, an investment-holding vehicle that held about 30.9% at IPO and was trimmed to 30.00% by November 2025; within it Wang controls RUYI (24.02%) plus the Anjoin employee platform (2.73%), and because no party exceeds 50%, the company discloses no actual controller. Her look-through economic interest is therefore single-digit-percent, her direct registered stake is only about 89,700 shares, and in January 2026 she disclosed her first small reduction, of up to 22,420 shares or 0.0133%. The National IC Fund participates in the structure and state funds have been trimming semiconductor stakes broadly — a financial rather than permanent-strategic holder base.

The Baillie-honest synthesis: leadership longevity and willingness to invest score high; but a deeply bound owner who will sacrifice profit for a decade is only partly evidenced here. Ownership is dispersed, there is no actual controller, the founder's economic stake is modest, and broad employee incentives align talent without amounting to deep owner-binding. Medium-strong on this dimension, not top-tier.

Customers would miss Anji acutely in the short run — its slurries and cleaning chemicals are mission-critical, qualified-in consumables whose sudden removal would disrupt qualified production lines — and its growth is pro-social and sustainable. But as a firm it is ultimately replaceable: the function is essential, the specific supplier is substitutable over a painful, multi-quarter re-qualification cycle.

On the "how much would it be missed" test, the short-run answer is "a great deal." CMP slurry and post-etch and post-polish cleaning chemicals are consumed on every wafer and are qualified in over one-to-two years; the report stresses that switching a qualified supplier is slow, expensive and operationally risky. If Anji disappeared overnight, fabs that had standardized on its products would face genuine disruption until alternates were re-qualified — that is real, mission-critical embeddedness.

But the firm, as distinct from the function, is replaceable. Global alternatives exist — Entegris (after the CMC Materials deal), Fujimi and Cabot — alongside domestic peers such as 300236.SHE and 603078.SHG. The cost of Anji vanishing is the re-qualification lag and a stretch of supply tightness, not a permanent loss of capability. That is the honest difference between a critical consumable and an irreplaceable monopoly: customers would be hurt, then they would recover.

On sustainability and social value, Anji scores well. It enables domestic-substitution supply security for China's semiconductor base, sells a real input consumed in real production, and carries none of the addiction, predation or regulatory-harm profile that would make growth socially fragile. It is, if anything, policy-favored, as a national enterprise technology center. The main sensitivity is sitting inside an export-control-exposed supply chain — but that is a risk to Anji, not harm done by it.

So Q7 splits cleanly: high on mission-critical stickiness and on pro-social, sustainable growth; only medium on irreplaceability, because the need outlives any single supplier.

Anji's unit economics are excellent and sit above typical semiconductor-materials peers — about 58% CMP-slurry gross margin, 50% wet-chemicals margin and 25.18% ROE — but the trajectory at scale is mixed (segment margin slipped 1.66 points on mix), and incremental cash is currently being plowed into capacity and strategic inventory rather than returned, so free cash is thin during this investment phase.

The margin and return profile is high-quality. In 2025 CMP slurry earned a 58.28% gross margin, functional wet chemicals 50.00%, the blended IC-materials segment 56.79%, and weighted-average ROE was 25.18%, with core profit excluding non-recurring items still up 32.36%. Those are franchise-grade economics for a materials company.

Against a benchmark they look strong, not ordinary. Entegris — the global CMP-and-materials leader — ran a full-year 2025 adjusted gross margin of about 44.8% on roughly $3.2bn of sales, with net income near $236m, implying a high-single-digit-to-low-double-digit ROE. Anji's roughly 57% blended gross margin and 25% ROE are comfortably above that, so on current economics it is a premium-margin operator rather than a laggard.

Better or worse at scale? Mixed. The 1.66-point segment gross-margin slip reflects mix — faster-growing wet chemicals at 50% diluting the 58% slurry, plus lifecycle pricing — which suggests blended margins may drift modestly lower as the platform broadens, even while absolute economics stay high; the steady 25% ROE is the more reassuring scale signal.

Where the cash goes matters for honesty. Capex rose to CNY 355.38m, construction-in-progress to CNY 226.20m, and raw-material inventory swelled from CNY 492.21m to CNY 865.94m; an CNY 830.5m convertible funds the build-out. Operating cash flow of CNY 439.77m roughly covers capex, so reported free cash is slim — incremental returns are real on an ROE basis but are being reinvested into the second curve and supply resilience, not handed back to owners. Reasonable for the growth stage, yet not a cash-harvest story today.

A 10-year 5x from CNY 304.80 is possible but demanding, and it is not the base case: today's roughly 88x multiple already capitalizes years of flawless execution, so the easy mispricing is gone and almost everything must go right at once.

Start with the arithmetic. A 5x over ten years is about 17.5% annual price growth, taking market cap from about CNY 69.34bn toward roughly CNY 347bn. Even on a still-premium but de-rated exit multiple of about 35-40x, that implies roughly CNY 8.5-10bn of net profit — some 11-13 times the 2025 figure of CNY 783.65m, i.e. about 26-29% profit CAGR sustained for a decade. That is far above the report's own near-term frame (low-20s 2026 growth) and well above the structural double-revenue-in-five-years case (about 15%).

So the conditions that must hold simultaneously: sustained mid-to-high-20s profit growth for ten years (global CMP share climbing well beyond 13%, wet chemicals and plating scaling into major lines, upstream localization paying off); margins holding high despite mix and competition; the multiple compressing only modestly rather than normalizing toward a quality-industrials rating; no qualification or quality failure, no severe cycle, no geopolitical supply shock; and overseas plus the H-share plan genuinely monetizing. That is a long chain of "ands."

What the price implies today is acceleration, not just durability — the stock is up about 161% over twelve months and trades near a record high. The report's own scenarios make the asymmetry explicit: the optimistic 12-month case is only about +12% to +31%, while neutral is about -20% to -7% and pessimistic about -44% to -31%.

Strip the cycle and the structural growth is genuine at mid-teens-to-20%, but the gap to the 26-29% a 5x demands must be filled by multiple persistence and flawless second-curve scaling. That makes a clean 5x realistic only in the blue-sky tail; the odds improve sharply from the report's CNY 190-230 re-underwriting zone rather than from CNY 304.80.

The market has noticed — emphatically — so this fails the classic Baillie hidden-gem test. A 161% one-year run and roughly 88x earnings mean the mispricing is not undiscovered; any remaining edge is a mild "can't-see-far" on the duration of the share-gain runway, not a "can't-understand" or "looks-down-on" blind spot.

The premise of the question is largely false here, and honesty requires saying so. Anji closed at CNY 304.80 on 2026-06-29, near a record high, after rising about 161% over twelve months, at roughly 88 times 2025 earnings, on unusually heavy volume. The report itself calls the name crowded rather than ignored. The marginal buyer is not discovering a leader; he is underwriting acceleration.

Run Baillie's three failure modes and none fits cleanly. Can't understand? No — the domestic-substitution CMP champion turning into a process-chemistry platform is consensus and simple enough for institutions, policy believers and momentum capital alike. Looks down on? No — Anji commands a scarcity premium, not a discount; the market pays up precisely because it respects the moat. Can't see far? This is the only place a thin edge survives: the market may under-credit the duration of the share-gain runway and the speed at which wet chemicals, electroplating and upstream localization compound into a genuinely diversified platform.

So the narrative inflection, if there is one, is proof rather than discovery: a re-rating from premium CMP specialist with options to proven multi-product process-chemistry platform, earned by wet-chemical and plating revenue visibly scaling and overseas validation landing. Until that proof arrives, the stock is fully-to-richly priced for what is already known.

The Baillie-honest conclusion: this is a weak hidden-gem setup. The opportunity is not a mispriced compounder waiting to be found; it is a fully-discovered quality leader whose attractive entry depends on a market-wide de-rating handing back the CNY 190-230 zone, or on the platform narrative proving bigger and faster than today's price assumes. Discovery is over; only execution can re-rate it from here.

天花板足够大、但属于「做大并吃掉别人手里的既有蛋糕」，不是凭空创造一个全新市场——这一点必须诚实分清，否则会把成长故事讲过头。

安集卖的是 CMP 抛光液与功能性湿电子化学品，这两块本身是已经存在几十年、由海外巨头长期占据的成熟半导体材料市场，而非新物种。它的天花板由两层叠加而成：一层是「赛道总量随半导体扩张」的被动增长，另一层才是「从海外厂手里抢份额」的主动增长，后者才是安集真正的价值来源。研报援引 WSTS/SIA 数据 给出赛道总量的量级：2025 年全球半导体销售额约 7049 亿美元、同比增长 11.2%，其中逻辑、存储增速分别约 38.8%、39.0%，先进逻辑、DRAM、HBM、3D NAND、先进封装与混合键合都在增加更精细的平坦化与化学步骤。富士胶片在其半导体材料业务简报中预计铜互连用 CMP 浆料 2024-2030 年 CAGR 可达 10%，这是「单片晶圆用量上升」对天花板的乘数。

更关键的是份额这条主动线，安集已经走出了实质进展，而非停在叙事。研报披露其全球半导体抛光液市场份额近三年由约 8% 升至约 13%（新浪科技援引公司口径），功能性湿电子化学品全球份额约 6%。这两个数字说明它的天花板不是「国产替代到顶就结束」，而是「全球细分份额还有一倍以上的提升空间」——从 13% 往 Entegris、富士胶片这类龙头的位置走，本身就是一块很大的、属于别人的蛋糕。

诚实的边界在于两点。其一，它不是创造新市场的颠覆者，抛光液、电镀液、清洗液都是客户产线上早已存在的工步，安集做的是替换供应商、再随制程升级吃增量，而非定义一个原本不存在的需求。其二，所谓「AI 天花板」是间接传导，研报明确把 AI 限定在先进逻辑、HBM、先进封装对材料用量的拉动，并自我警告「把安集当成 AI 直接弹性股会把故事讲过头」。所以它的天花板是「一个结构性扩张的成熟大市场 × 一个仍有翻倍空间的份额提升」，量级可观、确定性也不低，但叙事属性上是抢蛋糕而非造蛋糕。这决定了它的成长更像高质量复利、而非指数级爆发。

五年内收入翻倍的概率不低，但不应被当成「躺赢」；驱动主要来自「量 + 新业务」两条腿，而非提价，这恰恰是它质量较高的一面。

先看基数与近期斜率。研报与公开数据一致：2023 年营收 12.38 亿元、2024 年 18.35 亿元、2025 年 25.04 亿元（新浪科技披露 2025 年营收 25.04 亿、同比增长 36.47%），两年内营收已经翻倍。要在 2025 年的 25.04 亿元基础上再于五年内翻倍到约 50 亿元，对应复合增速只需约 15%。这在「过去两年已经翻倍」的趋势面前并不激进，反而是相对克制的门槛。2026 年一季报营收 7.24 亿元、同比增长 25.90%（东方财富披露 2026Q1 净利 2.08 亿、增长 23.01%），增速虽较 2025 全年放缓，但仍站在高位，给翻倍提供了起跑惯性。

再拆驱动结构，这是判断质量的关键。第一驱动是「量」：随本土晶圆厂扩产与制程升级，CMP 抛光液在同一客户中的应用层数增加、新客户线体导入持续，2025 年 CMP 抛光液收入 20.40 亿元、同比增长 32.06%（新浪科技口径），这是渗透与用量驱动、不是涨价。第二驱动是「新业务」：功能性湿电子化学品 2025 年收入 4.53 亿元、同比增长 63.73%，集成电路大马士革电镀液及添加剂已实现量产突破，正从里程碑事件进入收入验证期。第三才是「价」，而安集恰恰几乎不靠提价——研报反复强调它「放量不必降价」，但也没有提价叙事，价格更多是稳态。

诚实的风险点有二。其一，增速正在自然回落，从 2025 年的 36% 降到 2026Q1 的 26%，线性外推未来三年都维持 30%+ 是危险的，研报自己也警示这一点。其二，电镀液这条增量目前更像「期权」而非已兑现收入，若它迟迟做不大、湿电子化学品增速从 60%+ 回落，翻倍节奏会被拉长。综合看：五年翻倍（约 15% CAGR）是合理中性假设，由量与新业务共同驱动、提价贡献极小；但若想复制「两年再翻倍」的乐观路径，必须电镀液与先进封装相关材料同步放量，这一条尚未被证明。

第二曲线今天确实已经存在、而且已经在赚钱，不是 PPT；但它仍偏小、且最被市场寄予厚望的那一段（电镀液）还停在「刚量产」的早期，结构尚未坐实。

安集的第二曲线是清晰的两层结构。第一层是功能性湿电子化学品，这一层已经从「新品故事」变成「实打实的高毛利增长业务」：2025 年收入 4.53 亿元、同比增长 63.73%、毛利率仍有 50.05%（新浪科技援引公司口径披露湿电子化学品收入 4.53 亿、增长 63.73%）。一项新业务能在高增长的同时维持 50% 毛利，说明它不是靠亏损换规模，盈利基础已经具备，这就是「第二曲线今天存在」最硬的证据。第二层是集成电路大马士革电镀液及添加剂，2025 年实现量产突破，开始从里程碑事件进入实际收入验证期——它代表第三层潜在曲线，但目前体量小、可见收入有限。

把它放进公司战略叙事里看会更立体。研报指出，安集自 2023 年起在年报中把自己表述为「3+1」技术平台：化学机械抛光液、功能性湿电子化学品、核心原材料，再加电镀液及添加剂。这意味着第二曲线不是临时起意，而是从单一 CMP 龙头向半导体材料平台演化的既定路径，且公司用 17.76% 的研发投入占比 在持续供养它。

诚实的边界有三条。其一，规模差距仍大：湿电子化学品 18.08% 的收入占比相对 CMP 抛光液 81.45% 仍是小弟，它能改善增速、但还不足以分散对主业的依赖。其二，电镀液这条最性感的线，目前仍是「量产突破」而非「可见收入」，研报明确把它列为「期权」，并把「电镀液在未来四个季度内仍无可见规模收入」设为重新评估信号——也就是说，第三曲线能否成立今天还没有答案。其三，每个新品类的客户验证与技术难点并不等价，抛光液的成功不能简单平移。结论是：第二曲线（湿电子化学品）已经存在、已盈利、可验证；但它从「叙事」彻底变成「结构」的最后一公里，要看湿电子化学品占比能否持续上台阶、电镀液能否兑现成真实收入——这正是未来一两年最该盯的事。

核心竞争优势是「客户验证壁垒」——材料一旦写进晶圆厂工艺 recipe 就极难被替换；这条护城河未来三到五年总体会变宽，但宽的方向正在从「单点材料」转向「系统方案」，安集必须证明自己能跟上这次形态升级，否则护城河会在新维度上被相对削弱。

先说护城河为什么真实成立。CMP 材料不是买来即用的标准品，它与机台、垫材、上下游化学液、清洗步骤、材料结构全部相关；客户一旦完成验证并把材料写进 recipe，切换供应商就要承担良率、缺陷与再验证成本。这种黏性靠「进入客户工艺窗口」构成，而非靠销售关系，决定了收入的耗材属性与重复采购。研报给出三条可量化的支撑：其一是技术积累，公司 2025 年报披露已形成「3+1」技术平台、拥有 308 项已授予境内外发明专利、当年又新增申请 106 项；其二是份额验证，全球抛光液份额近三年由约 8% 升至约 13%（新浪科技援引公司口径），靠政策红利无法做到这一点；其三是盈利质量，2025 年 CMP 抛光液毛利率 57.92%、ROE 达 25.18%，高毛利与高回报本身就是壁垒的财务投影。

未来三到五年「变宽」的逻辑有两层。一层是制程升级顺风：先进逻辑线路层数增加、DRAM 层数增加、HBM 与混合键合等后段新技术都会提升 CMP slurry 用量与验证价值，头部供应商越往先进节点越容易吃份额——研报援引富士胶片简报给出铜互连 CMP 浆料 2024-2030 年 CAGR 约 10% 的行业锚。另一层是本地化服务：对中国本土晶圆厂，安集团队能更快到厂做问题定位、配方微调与失效分析，这种响应速度与共研深度是海外巨头的现实短板。

但必须诚实指出护城河面临的相对侵蚀，而非假装它只会单向变宽。研报点名鼎龙股份是最值得警惕的正面竞争者：它从 CMP 抛光垫起家做到国内龙头（2025 年 CMP 抛光垫收入 10.91 亿元、同比增长 52.34%），再向抛光液、清洗液延伸，正把竞争从「单点材料」升级为「pad+slurry+cleaner 一站式 CMP 方案」。如果客户越来越倾向一站式采购，安集「单品做到最深」的优势就需要靠平台化扩品来对冲。研报因此把「平台化扩品」列为「尚在形成中、不能按成熟状态估值」的第四条护城河。综合判断：核心护城河（验证壁垒）真实且会随制程升级加深；但竞争维度正在从配方优化升级为材料-设备-工艺协同，安集的护城河三五年内大概率变宽，前提是它在系统方案这一新战场上不掉队。

安集具备一定的自我重塑基因，但更多体现为「在相邻品类里横向复制验证能力」，而非「核心被颠覆后整体掉头重生」的强韧性；对待坏消息的态度，从年报披露看是偏坦诚务实的，但缺乏一次被真正逆境检验过的记录。

先谈自我重塑基因。这道题的隐含前提是：如果 CMP 抛光液这条核心业务被某种新材料路线或新平坦化技术颠覆，公司有没有能力换一条腿走路。安集给出的现实答案是「能力外溢」——它把抛光液上积累的颗粒分散、缺陷控制、客户验证 know-how，复制到功能性湿电子化学品（2025 年收入 4.53 亿元、同比增长 63.73%，新浪科技口径）和大马士革电镀液上，形成「3+1」平台。这种横向迁移本身就是一种重塑基因：它证明公司不是只会做一款产品的单点厂，而能把底层工艺理解迁到相邻战场。但要诚实说清边界——这几条相邻品类与 CMP 共享客户、共享验证逻辑、共享研发体系，本质是「同心圆扩张」，不等于「核心被彻底颠覆后另起炉灶」的极端韧性。半导体材料的颠覆通常是渐进的节点演进、而非一夜替代，这降低了被瞬间掀翻的概率，也让安集的渐进式扩品策略足以应对大多数情形，却未必能扛住一次范式级技术断层。

支撑「重塑能力」的底层资源是高研发强度与高回报的良性循环：2025 年研发投入占收入 17.76%、ROE 达 25.18%（新浪科技披露 ROE 25.18%、研发占比 17.76%），收入扩大十倍后研发强度仍接近 18%，说明公司愿意持续投入去培育下一条腿，而不是吃老本。

再谈如何对待错误与坏消息，这点研报给的证据偏正面但有限。一方面，公司在年报里主动写出对自己不利的事实：把全球关税政策变化、贸易摩擦、地缘政治、供应链成本上升与供应不稳定列为风险，没有回避客户集中度（前五大客户占 75.65%、第一大客户占 26.40%）和现金流走弱（2025 年经营现金流仅 4.40 亿元、显著低于 7.84 亿元净利润）。对 2025 年现金流偏弱，公司给的解释是为增长加大原材料备货，并与存货从 4.72 亿元升至 8.26 亿元相互印证——这是一个可被资产负债表交叉验证的解释，而非含糊搪塞。治理上过去三年不存在受证券监管机构处罚的情形，2025 年非独立董事 Chris Chang Yu 因个人原因辞任，公司也如实披露。另一方面也要承认局限：安集上市以来基本处于顺风期，尚未经历一次真正的行业下行或核心客户流失，它「在逆境中如何承认并纠错」缺乏实战样本。结论是：自我重塑基因存在但属于「同心圆扩张型」而非「绝境重生型」；对坏消息的披露态度坦诚务实、可交叉验证，但这种韧性还没被真正的逆境压力测试过。

管理层展现了清晰的长期视野，也确实在用「持续高研发、为增长提前备货与扩产」的方式牺牲短期现金以换长期成长；但在「创始人利益与公司深度绑定」这一柏基最看重的维度上，安集是个特例——它无实际控制人、不存在强创始人神话，绑定主要靠机构化治理而非个人股权信仰，这既消除了实控人折价，也意味着没有一个「赌上身家做十年」的灵魂人物。

先看长期视野与「为未来牺牲当下利润」的实证，这一点安集给得相当扎实。其一是研发强度：收入从 2018 年 2.48 亿元扩大到 2025 年 25.04 亿元、放大约十倍后，研发投入占收入仍接近 18%（2025 年 17.76%，新浪科技披露研发占比 17.76%、ROE 25.18%），这是典型的「宁可牺牲当期利润率也要养下一条腿」。其二是为增长前置投入营运资本：2025 年公司为保障后续供应加大原材料备货，导致存货从 4.72 亿元升至 8.26 亿元、经营现金流仅 4.40 亿元、显著低于 7.84 亿元净利润——管理层主动选择「现金回收效率」让位于「增长准备度」，这正是柏基欣赏的、不为短期报表好看而牺牲长期布局的取舍。其三是资本运作服务于长期扩张：研报披露公司 2025 年发行 8.305 亿元可转债、2026 年一季度完成转股与赎回摘牌，把资金投向产能与研发平台化，说明它仍处在成长型资本开支阶段、而非现金牛收割期。

再看治理结构与利益绑定，这里需要诚实分清利弊。研报披露：控股股东 Anji Microelectronics Co. Ltd. 2025 年末持股 30.70%，公司无实际控制人，也没有特别表决权安排；过去三年无证券监管处罚。这种结构的好处是没有「一人说了算」的强实控人治理折价、对中小股东相对友好；代价是缺少一个利益与公司深度捆绑、愿意为五到十年后押上个人声誉与身家的创始人灵魂，市场只能主要依靠经营兑现来评估管理层，而非依赖创始人信仰。研报对管理层可信度给出「中高」而非「高」的画像评分，正是这个权衡的体现。2025 年非独立董事因个人原因辞任，是一个需要跟踪、但目前影响有限的变化。

综合判断：在「长期视野」和「愿为未来牺牲当下利润」两点上，安集用高研发、为增长备货、持续扩产给出了明确的肯定证据；但在柏基最看重的「创始人式深度利益绑定与长期主义灵魂」上，它是机构化治理的特例——稳健、无强实控人风险，却也少了那种「我要做一家伟大公司」的个人意志，这让它更像一台可信但相对中性的成长机器。

如果安集明天消失，它的核心客户会相当难受、但不至于停产——这是一种「高切换成本型不可或缺」，而非「无可替代型不可或缺」；而它的增长方式高度健康，不仅不损害社会与监管，反而踩在「半导体供应链安全」这条受政策鼓励的主线上，可持续性是它最干净的一面。

先看不可或缺性这重前提。CMP 材料一旦通过验证、写进晶圆厂工艺 recipe，客户更换供应商就要重新承担良率、缺陷与再验证成本，短期内换不掉——这正是研报反复强调的「客户验证壁垒」。所以若安集突然消失，已经把安集材料写进产线的客户会立刻面临良率波动与重新验证的阵痛，尤其在它做到国内龙头、全球抛光液份额约 13% 的高端节点上（新浪科技援引公司口径），替代选项有限、迁移周期以季度计。但要诚实划清边界：抛光液并非世界上只有安集能造，Entegris、富士胶片等海外龙头以及国内鼎龙股份都在同一战场，客户最终能找到替代，只是要付出时间与良率代价。因此「想念」的程度是「很疼、要疼一阵」，而不是「无可替代、一旦消失就崩」。反过来看,这也意味着它的不可或缺性是相对的、需要靠持续进新工艺层来不断续期。

再看「增长是否可持续、是否损害社会与监管」这重前提,这恰恰是安集最经得起推敲的地方。其一,商业模式本身是清白的:它卖的是让芯片制造更高良率、更先进制程的工艺材料,客户是中芯国际、华虹、台积电等晶圆厂(研报披露的长期客户脉络),不存在损害消费者、透支监管套利或制造外部性的成分,收入来自真实工艺价值而非伤害性手段。其二,它的增长方向与监管和社会利益同向:本土供应链安全考量强化了晶圆厂导入国产材料的意愿,国产替代是受政策鼓励、而非被政策压制的方向。其三,治理合规干净,研报披露公司过去三年不存在受证券监管机构处罚的情形。

不过仍要把唯一的可持续性风险讲清楚,而不是粉饰成零风险:它来自地缘政治的另一面。研报援引公司年报提示,全球关税政策变化、双边贸易摩擦与地缘政治风险,可能带来供应链成本上升、供应不稳定;安集仍有部分原材料与海外供应链关联,意味着它既受益于国产替代、也暴露于上游国际贸易摩擦。这不是「它损害社会」的风险,而是「外部环境损害它」的风险——性质相反,但对增长可持续性同样关键。综合判断:不可或缺性为「高切换成本型、相对而非绝对」,客户会很想念但能找到替代;增长方式健康、与社会与监管利益同向,可持续性是其最干净的维度,唯一外生变量是上游地缘供应链。

单位经济非常优秀,且具备「规模越大、利润率越有机会抬升」的正向经营杠杆,这是安集商业质量最硬的部分;赚来的钱主要投回研发、产能与营运资本备货以服务未来增长,而非分红收割——它仍是成长型再投资公司,不是现金牛。

先看毛利与盈利结构,这是单位经济的底色。2025 年综合毛利率约 56.7%,其中 CMP 抛光液毛利率 57.92%、功能性湿电子化学品 50.05%(新浪科技援引公司口径披露湿电子化学品毛利率约 50%)。一家材料公司能把主业做到近 58% 毛利,且第二曲线一上来就是 50% 的高毛利区间,说明它卖的是嵌入客户制程的配方与验证价值、而非大宗化学品的吨位价差。增量回报上,2025 年 ROE 达 25.18%(同口径),收入扩大约十倍后这一回报率反而继续上升,这是「放量不必降价、扩张不必稀释盈利」的有力证据,安集没有掉进典型制造业「越做越难赚」的陷阱。

再看规模变大后会更好的机制,即经营杠杆。研报给出可量化的拐点信号:2025 年期间费用合计增幅约 20.52%、明显低于收入增幅 36.47%,于是利润增速(归母净利同比增长约 46.85%,新浪披露净利 7.84 亿、增长 46.85%)高于收入增速。原理是安集的成本结构里,真正的大头是前端研发、工艺支持、验证服务、分析检测这些高固定成本;一旦产品进入量产,固定成本被收入摊薄,利润就会加速释放。所以它的单位经济是「规模越大、利润率越有机会抬升」,而非走差。反过来也要诚实指出代价:一旦收入放缓,最难压缩的恰恰也是研发与质量体系开支,公司不能在下行时立刻砍研发,否则下一轮节点导入会出问题——高固定成本是双刃剑。

最后看「赚来的钱花在哪」,这是判断它是否值得长期持有的关键,答案是几乎全部投回成长。其一投研发,2025 年研发投入占收入 17.76%(同口径),持续供养「3+1」平台与电镀液新品。其二投产能,2025 年固定资产增加约 1.45 亿元、在建工程增加约 0.10 亿元,叠加可转债融资 8.305 亿元用于扩产。其三投营运资本,为保障后续供应加大原材料备货,使存货从 4.72 亿元升至 8.26 亿元——这也是 2025 年经营现金流仅 4.40 亿元、CFO/净利润比值降到约 0.56 的直接原因。这里必须诚实标注一个隐忧:高质量的单位经济目前没有完全转化为自由现金流,研报穿透测算后认为,扣除维持性 capex 后的所有者收益约 3.7-3.9 亿元、对应当前市值的所有者收益率不足 1%。换句话说,生意本身赚钱效率很高,但公司选择把钱继续投回增长、而非吐出现金,因此「单位经济优秀」与「自由现金流充沛」在当下不是一回事——投资者买的是再投资复利,不是当期分红。

要让安集十年涨五倍,需要「盈利持续高增 + 第二曲线兑现 + 估值不大幅收缩」三个条件同时成立,而今天约 60 倍 PE 的股价已经把前两个利好的相当一部分提前买走,留给投资者的安全边际很薄——这是一家好公司处在不友好买点的典型样本。

先把「十年五倍」拆成可检验的条件。当前股价约 225 CNY、市值约 511.9 亿元、市盈率(TTM)约 60 倍(据 cn.investing.com 截至 2026-06-15 收盘数据;研报口径为 2026-06-12 收盘 230.45 元、静态 PE 约 66.9 倍)。十年五倍意味着市值要到约 2560 亿元、年化回报约 17.5%。要实现它,以下几条必须同时成立:其一,盈利十年增长至少四到五倍,对应净利润从 2025 年的 7.84 亿元升到约 35-40 亿元,要求营收与利润长期复合增速维持在 17%-18% 以上;其二,第二曲线真正接棒,功能性湿电子化学品占比从 18.08% 持续上台阶、电镀液及添加剂从「量产突破」变成可见规模收入,否则单靠 CMP 主业无法支撑这个体量;其三,十年后市场仍愿意给它一个不太低的估值倍数(比如 25-30 倍而非 15 倍),即估值中枢不发生剧烈去溢价;其四,全球份额继续提升、海外收入占比显著抬升,证明它不只是中国本土替代商。

这些条件现实吗?要分层诚实回答。盈利端的 17%-18% 复合增速并不离谱,毕竟它过去两年营收已经翻倍、2026Q1 仍增长 25.90%(东方财富口径),CMP 主业 2025 年增长 32.06%、湿电子化学品增长 63.73%,基本面支持中高速增长。但难点在于「同时成立」:第二曲线兑现、海外突破、估值不收缩这三条彼此独立、并非自动连成一串,任何一条掉链子都会让五倍打折。尤其估值这一条是最不可控的——今天 60 倍的起点本身就偏高,十年里只要倍数从 60 倍回落到 30 倍,就等于把盈利翻五倍的成果硬生生砍掉一半。

今天股价隐含了什么预期?这是最该讲透的部分。研报穿透得很清楚:静态 PE 约 66.9 倍、市销率约 20.9 倍、所有者收益率不足 1%,意味着市场已经隐含三条乐观假设——CMP 主业至少维持约 20% 增长、湿电子化学品持续 40%+ 增长并提升占比、电镀液从量产突破迅速走向规模收入。前两条有现实基础,第三条更像「期权」却已被计入价格。一个更残酷的参照:若未来三年盈利零增长而市场仍给当前约 66.9 倍 PE,隐含盈利收益率仅约 1.5%,而研报披露 2026-06-12 中国 10 年期国债收益率约 1.74%——也就是说,当前买入价对应的静态收益率甚至没跑赢无风险利率。结论是:十年五倍在「条件全部兑现」的乐观路径下并非不可能,但今天的股价已经替这些好消息付了大半订金;安全边际很薄,这正是研报给「持有」而非「买入」、并把理想买入区定在 135-145 元的根本原因。

这道题对安集要诚实地反着答:市场并没有「还没意识到」它的好,恰恰相反,市场早已充分认识、甚至已经按平台成长股的高估值把它的好提前定价——它不是一只「被忽视的便宜龙头」,而是一只「被看得很清楚、且被给足溢价」的股票。柏基常问的「看不懂、看不起、看不远」三种错杀,在安集身上基本都不成立,真正的预期差只剩两件尚未被完全计入的事。

先逐一排除三种「市场没意识到」的可能,这是诚实的关键。其一,市场没有「看不懂」:安集 2019 年科创板上市起就被当作稀缺国产替代材料股,卖方覆盖充分,研报披露其商业模式、客户验证壁垒、份额提升(全球抛光液份额由约 8% 升至约 13%,新浪科技援引公司口径)都被反复研究,这不是一个藏在角落、无人理解的冷门标的。其二,市场没有「看不起」:截至 2026-06-15 收盘股价约 225 CNY、市值约 511.9 亿元、市盈率(TTM)约 60 倍(据 cn.investing.com 数据),研报口径静态 PE 约 66.9 倍、市销率约 20.9 倍——这是「看得很起」的价格,而非被低估。其三,市场也没有「看不远」:它愿意为电镀液这条尚未兑现的第三曲线、为 AI/HBM/先进封装带来的长期材料用量提前付费,看得相当远、甚至有点过远。所以三种经典错杀都不适用,这正是研报反复克制的根源:好公司不等于好价格。

那么真正还没被彻底写进价格的是什么?研报给出两件、且只有两件。第一件,是电镀液及添加剂能否从「量产突破」变成「实质性可见收入」——这条第三曲线目前被市场当作大概率事件计入,但尚未被收入证实,一旦兑现会强化平台溢价、一旦落空会触发估值身份下修。第二件,是经营现金流能否重新追上利润——2025 年经营现金流仅 4.40 亿元、显著低于 7.84 亿元净利润(新浪披露净利 7.84 亿),CFO/净利润比值降到约 0.56,存货从 4.72 亿元升至 8.26 亿元,市场目前接受了「为增长备货」的解释,但还没有把「现金流到底能不能修复」定论进价格。这两件事既是上行期权、也是下行风险,构成当下唯一真实的预期差。

什么会成为「叙事拐点」?分两个方向诚实列出。向上的拐点:2026 年中报若显示湿电子化学品收入占比进一步上台阶、CMP 主业仍保持约 25% 增长、同时经营现金流修复且库存被收入消化,市场会确认「第二曲线开始接棒 + 利润质量回升」,愿意维持甚至上修高估值。向下的拐点更容易突然出现:任何一份财报里若收入还在增、但现金流与库存显著恶化(CFO/净利润持续低于 0.7、存货/营收升破 35%),或前五大客户占比逼近 80%、第一大客户逼近 30%,或电镀液在未来四个季度仍无可见规模收入,市场就会把它从「高质量平台成长股」改价为「高成长但重投入材料股」,在基本面仍不错时先杀估值。换句话说,安集的拐点不取决于「市场何时意识到它好」(早已意识到),而取决于「它能否持续兑现已被高价计入的好」——这是高估值优质成长股最典型的处境。