ACM Research, Inc. (ACMR) - Deep Dive Research Report

Prepared 2026-06-26. Reporting calendar: ACM Research is a calendar-year filer. The six most recent quarterly earnings calls used throughout this report are Q4/FY2024 (Feb 26 2025), Q1 2025 (May 2025), Q2 2025 (Aug 6 2025), Q3 2025 (Nov 5 2025), Q4/FY2025 (Feb 26 2026), and Q1 2026 (May 7 2026). The most recent is within ~50 days of today.

1. What the company does

ACM Research builds the machines that clean, plate, and process silicon wafers inside semiconductor factories. When a chipmaker like SK Hynix or SMIC fabricates a chip, the wafer passes through hundreds of process steps, and after almost every one of them microscopic particles, residues, and contaminants have to be stripped off without damaging the increasingly fragile nanometre-scale structures etched into the surface. ACM makes the wet-process equipment that does this: single-wafer and batch cleaning tools, electroplating tools that deposit copper for wiring and packaging, furnaces, coater/developer "track" tools, and deposition equipment. In short, it sells capital equipment to fabs, then sells the spares and service that keep those tools running for a decade or more.

The company is unusual in its structure, and you cannot understand the business without it. ACM Research, Inc. is a Delaware corporation founded in 1998 by Dr. David Wang in the Bay Area (Fremont, California). It is listed on Nasdaq under ACMR. But the operating heart of the company is its subsidiary, ACM Research (Shanghai), formed in 2005, which designs and manufactures nearly all the tools, employs most of the people, and generates almost all the revenue. ACM Shanghai is itself separately listed on Shanghai's STAR Market (688082.SS), and the Nasdaq parent owns roughly four-fifths of it. So a buyer of ACMR is really buying a controlling stake in a Chinese semiconductor-equipment champion, wrapped in a US-listed holding company controlled by its founder through super-voting stock.

The founding story explains everything about the current product line. Wang's original idea was Stress-Free Polishing, an electrochemical alternative to chemical-mechanical planarisation (CMP). It did not win against the entrenched CMP incumbents. Rather than fold, ACM redeployed the same wet-chemistry and megasonic know-how into wafer cleaning, a less glamorous but enormous and recurring market, and chose to commercialise it in China, where a domestic alternative to Japanese and American tools was strategically wanted. The pivotal technical breakthrough was a megasonic cleaning method the company calls SAPS (Space Alternated Phase Shift), which agitates cleaning chemistry uniformly across the wafer, and a second-generation version called TEBO (Timely Energized Bubble Oscillation) that controls cavitation precisely enough to clean delicate patterned structures without breaking them. SK Hynix placed the first production SAPS order in 2011, and that single reference customer gave ACM the credibility to sell more broadly.

"We now have industry-leading offering across multiple product categories that enable our global customers to effectively solving their evolving production challenges." - David Wang, Q1 2026 call (May 7 2026)

The value proposition to a fab is concrete: ACM tools clean or plate as well as the global incumbents, often consume far less chemistry (one of its hybrid cleaning tools cuts sulfuric-acid and peroxide use by around 80%), and come from a supplier that is local, lower-priced, and politically aligned in a market where chipmakers are being pushed to "de-Americanise" their equipment lists. That combination of good-enough-to-better technology, lower cost of ownership, and domestic-supplier status is the whole game.

2. Business segments

ACM does not report classic divisions with separate P&Ls. It reports one operating segment but breaks revenue into three product groupings, and those groupings are the most useful way to understand the business. Each is built on a distinct technical capability, so I treat them as segments here.

2.1 Single-wafer cleaning, Tahoe and semi-critical cleaning (the core, roughly two-thirds of revenue)

This is the foundation of the company and the largest revenue contributor (management noted single-wafer cleaning, Tahoe and semi-critical cleaning tools were about 68% of revenue in Q3 2025). It covers the full menu of wafer-cleaning steps a fab needs. At the high-value "front-end critical" end sit the SAPS and TEBO megasonic platforms, which remove particles after etch, deposition, and lithography steps on advanced logic and memory wafers. Tahoe is a hybrid single-wafer/batch tool that performs sulfuric-acid-peroxide cleaning while slashing chemical consumption, which matters because those chemicals are expensive and their disposal is regulated. "Semi-critical" tools handle the less demanding cleans (scrubbers, auto-bench, batch) where competition and pricing pressure are heaviest.

The core capability here is two decades of megasonic and wet-chemistry process knowledge, much of it patented (TEBO has US patents), plus the qualification track record at real fabs. A cleaning tool is not bought on a spec sheet; it has to be proven on the customer's specific device for months before it is trusted in production. ACM's installed base at SK Hynix, SMIC, Hua Hong, YMTC and CXMT is the moat. The newest extension is a high-temperature single-wafer SPM (sulfuric-peroxide-mixture) tool and a supercritical-CO2 dry tool, both aimed at advanced nodes where traditional cleaning damages high-aspect-ratio structures. Management frames these as a multi-hundred-million-dollar incremental opportunity. This segment is the cash engine and the reference base that lets ACM sell everything else.

2.2 ECP (electrochemical plating), furnace and other technologies (the fastest-growing group)

This grouping bundles the products ACM is using to expand beyond cleaning. ECP electroplates copper, both for front-end interconnect wiring and, increasingly, for advanced packaging where thick copper pillars and redistribution layers are built. In Q1 2026 management said ECP revenue tripled year over year, and the broader "ECP, furnace and other" line grew over 200%, making it the standout growth driver. The newest ECP product is "Ultra ECP ap," a horizontal panel-level plating tool aimed at the shift from round wafers to large rectangular panels in packaging.

"Furnace" covers vertical and horizontal thermal-processing tools (oxidation, diffusion, LPCVD, ALD) that compete in a market historically owned by Tokyo Electron and Kokusai. "Other" includes the genuinely early-stage platforms: Track (coater/developer tools that bookend the lithography step) and PECVD (plasma-enhanced deposition, including a silicon-carbide-nitride process under customer evaluation). The core capability ACM is leveraging is that once a fab trusts you on cleaning, you can sell adjacent wet and thermal tools into the same lines using the same sales relationships and the same local-service footprint. This is the growth bet of the company, the option value that turns ACM from a cleaning specialist into a multi-product equipment supplier.

2.3 Advanced packaging (excluding ECP), services and spares

Advanced packaging is where chips are assembled into the 2.5D/3D stacks that AI accelerators and high-bandwidth memory require, and it is one of the fastest-growing parts of the whole semiconductor capital-equipment market. ACM's packaging tools include cleaning, photoresist stripping, wetting, and both wafer-level and panel-level systems. Management said advanced-packaging revenue (excluding ECP) grew 45% in 2025 to about $76 million, roughly 8% of total revenue, and grew a further 62% in Q1 2026, with wafer-level systems shipped to OSAT customers in Singapore and panel-level vacuum-cleaning systems deployed to leading manufacturers. The "services and spares" piece is the recurring annuity: every tool ever shipped needs parts and maintenance, and that revenue compounds quietly with the installed base.

3. Products and business detail

The full catalogue, by family:

• Ultra C SAPS - single-wafer and batch megasonic cleaning using space-alternated phase-shift acoustics; the original flagship, used for front-end particle removal on advanced wafers.
• Ultra C TEBO - timely-energized-bubble-oscillation megasonic cleaning for patterned wafers; controls cavitation so it cleans fragile 2D/3D structures (FinFET, 3D NAND) without collapsing them. US-patented.
• Ultra C Tahoe - hybrid single-wafer/batch high-temperature sulfuric cleaning that cuts sulfuric-acid and hydrogen-peroxide use by roughly 80%, lowering both cost and environmental burden.
• High-temperature single-wafer SPM - a newer hot-SPM cleaning tool; management says SPM cleaning is about 30% of the cleaning market and targets meaningful share, citing best-in-class particle performance (sub-20 count at 50nm) and maintenance-free operation.
• Supercritical CO2 dry - a drying tool that avoids pattern collapse on high-aspect-ratio structures and reportedly cuts consumable cost ~40% versus competitors.
• Ultra ECP map / Ultra ECP ap - electrochemical copper plating for front-end (map) and advanced packaging (ap), including a new horizontal panel-level plating system for rectangular panels.
• Ultra Fn furnace - vertical/horizontal furnaces for oxidation, diffusion, anneal, LPCVD and ALD.
• Ultra C Track - coater/developer tools that prepare and develop photoresist around the lithography step.
• Ultra PECVD - plasma-enhanced CVD, including a silicon-carbide-nitride film process under customer evaluation.
• Advanced-packaging wet tools - cleaning, photoresist strip, wetting, and panel-level vacuum cleaning for 2.5D/3D and fan-out packaging.
• Services and spare parts - installed-base maintenance, the recurring revenue tail.

Manufacturing is concentrated in China. The principal production base is the Lingang campus near Shanghai, where the first building is in volume production and a second building is planned to open later in 2026; combined, management says the campus can support up to roughly $3 billion of annual output. A key competitive asset embedded at Lingang is a proprietary R&D "mini line," a small in-house fab line on which ACM can run a customer's specific process before shipping, compressing tool qualification from "more than a year to a few quarters." That shortens the single most important bottleneck in selling new equipment.

Geographically the company is overwhelmingly China-centric (mainland China is well over 80% of revenue), but a deliberate diversification is under way. ACM is standing up an Oregon facility in the United States to enable in-house demos and US-made production by year-end 2026, partly to serve non-China customers and partly to insulate that business from tariffs and export rules. It has shipped tools to OSATs and a foundry in Singapore, and is engaging customers in Korea, Taiwan, Europe and North America. Management's stated goal is more than 20 tools installed outside mainland China by year-end 2026 across roughly 10 customers in five countries.

Milestones that changed the business: the 2011 SK Hynix SAPS production order (first credibility); the 2017 Nasdaq IPO of the parent; the 2021 STAR Market IPO of ACM Shanghai (capital access and a China-listed currency); the move into ECP and advanced packaging from ~2020 onward; and the December 2024 addition of ACM Shanghai to the US Entity List, which crystallised the company's de-Americanisation strategy.

4. Customers

ACM sells to chipmakers and packaging houses. The named and reported customer base is concentrated in China: SMIC, Hua Hong / HLMC, YMTC (3D NAND), CXMT (DRAM), and Nexchip, among others, with SK Hynix as the historically pivotal first international reference and Intel cited (in independent coverage) as having qualified ACM tools at its Oregon fabs. More recently ACM has added a Singapore foundry customer and OSAT packaging customers in Singapore.

Inside the customer, the buyer is the fab's process-integration and equipment-engineering organisation, not a procurement clerk. The decision criteria are, in order: does the tool meet the particle/defect and yield spec on our specific device; how reliable is it across thousands of wafers; what is the cost of ownership (chemical consumption, uptime, footprint); and can the vendor service it locally and fast. Price matters, but it is rarely the first gate. The sales cycle is long: a new tool typically goes through evaluation, then qualification on the customer's process (historically a year or more, now compressed by ACM's mini line), before it earns repeat production orders.

Switching costs are high and they cut both ways. Once a cleaning or plating recipe is qualified on an ACM tool and tied into a fab's yield, swapping vendors means re-qualifying, which risks yield and costs months. That installed-base lock-in is why the SK Hynix reference mattered so much and why each new qualified account compounds. The flip side is concentration: a heavy reliance on a handful of large Chinese fabs means ACM's fortunes track those customers' capex cycles and their access to capital and to the wafers and materials they need.

Customer concentration is real and is best read as a function of where ACM is allowed and wanted to sell. The Chinese fabs are buying domestic equipment partly by national policy, which is both the opportunity (guaranteed demand) and the risk (the demand is policy-dependent and politically exposed). Contracts are largely purchase-order and milestone based rather than long-term take-or-pay, so revenue is driven by the order book and shipment timing, which is why management talks constantly about backlog and shipments as leading indicators. The services-and-spares tail adds a layer of recurring, more predictable revenue on top of the lumpy tool orders.

5. Competitive landscape

The wafer-cleaning and wet-process equipment market is an oligopoly globally and a land-grab domestically in China. At the global level the cleaning market is led by a small number of incumbents: SCREEN Holdings (the historical share leader in single-wafer cleaning), Tokyo Electron, Lam Research, and Applied Materials, which together with a few others hold the large majority of global revenue. ACM competes against all of them on technology, and against them and a rising set of Chinese peers on price and locality.

Where ACM wins: inside China, it wins on domestic-supplier status, lower price, faster local service, and technology that is now genuinely competitive on key cleaning and plating steps (and, on cost of ownership through reduced chemical consumption, sometimes better). Where ACM loses or is exposed: at the leading edge outside China, the incumbents still own the qualified installed base and the most advanced nodes, and ACM has limited presence in many of those fabs. Within China itself, ACM no longer has the domestic field to itself: Naura, Kingsemi, PNC Process and others are building competing cleaning and adjacent tools, so part of ACM's future is defending domestic share against domestic challengers, not just displacing foreign incumbents.

Barriers to entry are high but not impregnable. The barriers are process know-how (megasonic acoustics, plating chemistry), the multi-year qualification track record, patents, and the service footprint. They are being partially offset in China by state funding that lets new domestic entrants buy their way up the learning curve, and by the same de-Americanisation tailwind that helps ACM also helping its Chinese rivals. The structural shift to watch is the AI-driven capex wave (advanced packaging, HBM) opening new wet-process demand that is less locked-in than legacy cleaning, which is exactly where ACM is pushing.

6. Industry

Demand for ACM's equipment is driven by one thing above all: how much money chipmakers spend building and upgrading fabs (wafer-fab equipment, or WFE, spending). That, in turn, is driven by end-demand cycles - currently the AI compute and high-bandwidth-memory build-out - plus the secular rise in process steps per chip (smaller nodes and 3D structures need more cleaning and more plating steps) and, in China specifically, a government-backed drive for semiconductor self-sufficiency.

The wafer-cleaning equipment market is sized by various research firms in the low single-digit billions of dollars annually and growing at a mid-to-high single-digit to low-double-digit CAGR, with the broader WFE market an order of magnitude larger. Cleaning is one of the more resilient WFE categories because every process step generates contamination, so cleaning intensity rises as chips get more complex even when unit volumes are flat. Advanced packaging is the faster-growing adjacency, pulled by AI accelerators that stack logic and memory and therefore need far more wet processing and plating.

ACM sits in the wet-process and thermal layer of the WFE supply chain, downstream of the lithography and etch giants but essential to yield. In China, the structural story is import substitution: domestic fabs historically bought the vast majority of their cleaning and plating tools from Japan and the US, and policy now pushes them to localise; estimates put Chinese self-sufficiency in cleaning at roughly 50-60% and rising, which is the wave ACM is riding. Regulation is the dominant industry force here and it is double-edged: US export controls restrict what the most advanced Chinese fabs can buy from American suppliers (helping domestic vendors win share) while also restricting what those fabs can build and what US-linked suppliers like ACM can ship. The industry is cyclical - WFE spending swings with memory prices and macro capex - but the China-localisation and AI-packaging tailwinds are currently offsetting the normal cycle. The headwinds are equally specific: tighter export rules, the risk of ACM Shanghai's own Entity-List status, and the possibility that Chinese fab capex is being pulled forward by subsidy rather than sustained end-demand.

7. Growth triggers

All points below are drawn from the six earnings calls.

• Single-wafer SPM ramp. Management expects more than 15 to 20 SPM units delivered by year-end 2026, citing best-in-class particle results and maintenance-free operation as differentiators against incumbents. (Q1 2026 call, May 7 2026; SPM qualification first flagged Q4 2024, Feb 26 2025 - repeated.)

  "SPM cleaning process tool has occupied 30% of the cleaning market. We believe our innovative hot SPM tool will take a significant market share." (Q1 2026, May 7 2026)

• ECP / panel-level plating expansion. Electroplating revenue tripled year over year, with horizontal panel-level plating (Ultra ECP ap) gaining traction in Asia and with global customers; first Ultra ECP horizontal panel tool delivered with multiple OSAT/packaging orders scheduled for Q1 2026 delivery. (Q1 2026, May 7 2026; orders first announced Q4 2025, Feb 26 2026 - repeated.)
• Advanced packaging deployments. Panel-level vacuum-cleaning systems deployed to leading manufacturers and wafer-level systems shipped to OSAT customers in Singapore; orders announced from three global customers across wafer- and panel-level systems. (Q1 2026, May 7 2026; Q4 2025, Feb 26 2026.)
• International installed base outside China. Target of more than 20 tools installed outside mainland China by year-end 2026 across ~10 customers in five countries. (Q1 2026, May 7 2026.)

  "More than 20 tools installed outside Mainland China by year-end 2026." (Q1 2026, May 7 2026)

• US production capability (Oregon). In-house demo lab with multiple tools and US-made production capability targeted by year-end 2026, to serve non-China customers and mitigate tariffs. (Q4 2025, Feb 26 2026; reiterated Q1 2026, May 7 2026 - repeated.)
• Lingang second building. First building in volume production; second building opening later in 2026, supporting combined annual output capacity up to roughly $3 billion. (Q4 2025, Feb 26 2026; Q1 2026, May 7 2026 - repeated.)
• New platform commercialisation (furnace, Track, PECVD). Vertical furnace, high-throughput Track, and PECVD silicon-carbide-nitride systems under customer evaluation, expected to contribute from 2026 onward. (Q4 2025, Feb 26 2026; Q1 2026, May 7 2026.)
• Supercritical-CO2 dry and next-gen cleaning. Management identified over $1 billion of incremental next-generation cleaning opportunity in mainland China alone, with the CO2 dry tool cutting consumable cost ~40%. (Q4 2025, Feb 26 2026.)
• ACM Shanghai Hong Kong H-share listing. ACM Shanghai completed a ~$110 million minority share sale and announced a secondary H-share listing in Hong Kong, expanding capital access. (Q1 2026, May 7 2026.)

8. Key risks

• Export-control / Entity-List escalation. ACM Shanghai and its China and Korea subsidiaries were added to the US Entity List in December 2024. The mechanism that hurts: tighter rules could restrict ACM Shanghai's access to US-origin components, software, or to non-China customers, or restrict what its fab customers can build. The parent and non-China subsidiaries were not added, and management has said the supply-chain impact is "minimized and manageable" via alternative suppliers, but a further tightening is a high-probability, moderate-to-severe drag because the entire business sits at the centre of the US-China tech conflict. This is the single most important risk.
• Customer and geographic concentration. Over 80% of revenue comes from mainland China, and a handful of large fabs (SMIC, Hua Hong, YMTC, CXMT) dominate. If Chinese fab capex slows - whether from a memory downturn, subsidy fatigue, or restricted access to wafers/materials - ACM's order book contracts quickly because orders are PO-based, not long-term contracted.
• China capex is partly policy-driven. Much of the demand is underwritten by national self-sufficiency policy and subsidy. If that fab build-out is being pulled forward rather than reflecting sustained end-demand, a digestion period could follow a boom. Management's own bullishness on a multi-year domestic cleaning opportunity assumes the localisation drive persists.
• Margin pressure from product mix and competition. Q4 2025 gross margin fell to 41%, below the 42-48% target, on competitive pricing in semi-critical products (~5 points of headwind) and higher inventory provisions (~4 points). The mechanism is structural: as ACM grows volume in lower-margin semi-critical tools and faces domestic price competition, mix can drag margins even as revenue grows.

  Management characterised the lower margins as "temporary" and said the company is "willing to spend more R&D ... worth to spend money now" given the AI opportunity. (Q4 2025, Feb 26 2026) That is a credible explanation, but it also means margin recovery is a forward promise, not a delivered fact.

• Rising domestic competition. Naura, Kingsemi, PNC and others, funded by the same state push, are building competing cleaning and adjacent tools. ACM's domestic-supplier advantage is shared with these rivals, so part of its growth depends on out-executing them, not just displacing foreign incumbents.
• Holding-company / governance structure. US ACMR holders own the business only through a controlling stake in STAR-listed ACM Shanghai, and David Wang controls voting through super-voting stock. Capital returns happen at the Chinese subsidiary level (see Section 10). A minority position in a China-listed, founder-controlled entity carries governance and value-capture risk that a simple operating company would not.
• New-product execution. Furnace, Track, PECVD and panel-level ECP are still in evaluation or early ramp. If qualification slips, the growth-diversification story (and the margin-recovery story that depends on higher-margin new products) is delayed.

9. Walk the talk

The six calls used: Q4/FY2024 (Feb 26 2025), Q1 2025 (May 2025), Q2 2025 (Aug 6 2025), Q3 2025 (Nov 5 2025), Q4/FY2025 (Feb 26 2026), Q1 2026 (May 7 2026).

Start with the FY2024 call in February 2025. Management reported 2024 revenue of $782 million, up about 40%, and guided 2025 to $850-950 million, roughly 15% growth, deliberately a step down in growth rate after a very strong 2024 shipment year. They flagged that 2024 shipments had jumped over 60% and that 2025 would grow off that high base but more modestly. This was a notably more conservative posture than the company's prior hyper-growth, and it set a beatable bar.

Through 2025 the company tracked that guide with reasonable fidelity. Q1 2025 (May) delivered $172 million, up ~13%, with gross margin at 48.2%, above the long-term target, and management held the $850-950 million full-year range. Q2 2025 (August) came in at $215 million with a strong 48.7% gross margin, again above target, though shipments were lumpy and the stock fell on a revenue miss versus the Street; management kept the full-year range. So far, so consistent: margins were running at or above the top of the model, and the annual guide was intact.

The Q3 2025 call (November) is where the texture changed. Revenue was a record $269 million, up 32%, but gross margin dropped to 42.1%, the low end of the model, and management narrowed full-year guidance to $875-925 million. The narrowing was honest and landed inside the original range, but the margin slide foreshadowed trouble. That trouble arrived at the FY2025 call (February 2026): full-year revenue came in around $901 million (within the guided range, so the top-line promise was kept), but Q4 gross margin fell to 41%, below the 42-48% target for the first time. Management did not bury it; they attributed roughly 5 points to semi-critical pricing pressure and 4 points to inventory provisions, called it temporary, and explicitly chose to keep spending on R&D.

"We expect lower gross margins to be temporary." - CFO Mark McKechnie, Q4 2025 (Feb 26 2026)

The Q1 2026 call (May 2026) is the early test of that "temporary" promise, and so far it supports management. Gross margin recovered to 46.5%, back above the midpoint, revenue grew 34% to $231 million, and the company reaffirmed the 2026 guide of $1.08-1.175 billion (25% growth at the midpoint) that it had first set at the FY2025 call. The margin bounce-back one quarter after the "temporary" framing is a tick in management's favour.

The pattern across six calls: management is consistent on the top line - it has hit or landed inside its annual revenue ranges repeatedly, and it tends to set beatable guides. On margins it is more volatile and more optimistic; the 42-48% "long-term model" has been breached on the downside, and the recovery is a promise that has been kept once (Q1 2026) but is only one quarter old. On product roadmap, the SPM, ECP, advanced-packaging, Oregon and Lingang commitments have been repeated consistently across calls and are progressing, though most of the headline new-product payoffs are still ahead and therefore unproven. The fair assessment: this is broadly credible management that delivers on revenue and roadmap milestones and is candid about problems, but that runs hot on margin expectations and asks investors to trust a "temporary" framing on the most important profitability metric. They mostly do what they say; watch the margin line, not the revenue line, to catch them out.

10. Shareholder friendliness index

Dividends: The US-listed parent, ACM Research, Inc. (ACMR), does not pay a dividend, and has not over the last three years; it retains and reinvests. The nuance that matters is that capital return happens one level down, at the STAR-listed subsidiary ACM Shanghai (688082.SS). For fiscal 2025, ACM Shanghai declared a cash dividend of RMB 6.233 per 10 shares, distributing roughly RMB 299 million, about 21.4% of its 2025 net profit (ACM Shanghai 2025 profit-distribution filing). A US ACMR holder receives no direct dividend; they benefit only indirectly through the parent's ~80% economic interest in that subsidiary distribution.

Buybacks and dilution: Per the MoatMap database, there were zero buybacks recorded at the ACMR level in the last ~90 days (since 2026-03-28). Looking back further than that 90-day window: I found no evidence of a material open-market repurchase program at the Nasdaq parent ACMR over the last three years; capital-return activity is concentrated at the subsidiary. ACM Shanghai conducted share repurchases of roughly RMB 50 million in 2025, which together with its dividend brought total 2025 cash returns to about RMB 349 million, or ~25% of net profit (ACM Shanghai 2025 distribution filing). At the parent level, shares outstanding have drifted modestly higher over three years from equity compensation (option and RSU dilution), as is typical for a growth-stage equipment company; the insider activity in Section 11 is dominated by option exercises, which create shares. Net of all this, the ACMR share count is flat-to-slightly-growing rather than shrinking.

Verdict: Neutral, tilting toward Hoards Capital at the level a US investor actually owns - the Nasdaq parent pays nothing and does not buy back stock, and the only genuine capital return (a modest dividend plus a small buyback) occurs inside the Chinese subsidiary, where the US holder captures it only proportionally and indirectly.

11. Insider activities

Venue: US (Nasdaq), so the primary source is SEC Form 4 via EDGAR; the MoatMap block (US is an open venue) is the spine, cross-checked against recent EDGAR filings. The block is current as of 2026-06-26 and is not marked stale.

The defining feature of the last 12 months is that there is not a single open-market purchase. Every transaction is either an open-market sale or an option/award exercise (the "Other" rows priced at $1.00, $13.89, or $19.49 are option exercises at the grant strike, almost always paired with same-day open-market sales). Across the window: 0 buys, 37 sells, 11 option-exercise/other events, spread over five insiders, and the selling clusters tightly around a sharp run-up in the stock from the mid-$40s in March 2026 to the high-$80s/low-$90s by early June 2026.

Buys: none. There is no open-market insider buying to read as a conviction signal in this window. That absence is worth stating plainly rather than glossing.

Sells - the why: nearly all of this activity has the clear fingerprint of option-exercise-and-sell. The "Other" lines at a $1.00 strike (Wang) and $13.89/$19.49 strikes (McKechnie, Cheav) are exercises of long-dated, low-strike options, immediately monetised at the prevailing market price; this is liquidation of vested equity compensation, not a directional bet against the business. The activity comes in two waves - mid-March 2026 (stock in the mid-$40s) and early June 2026 (stock in the high-$80s/low-$90s) - consistent with scheduled exercises and routine diversification, very likely under pre-arranged 10b5-1 plans, though the specific plan dates are not disclosed in the MoatMap rows and would need the Form 4 footnotes to confirm. CFO McKechnie's large March exercise-and-sell (~$5m) and CEO Wang's repeated 50,000-60,000-share exercise blocks fit the pattern of executives clearing tranches of expiring options. No filing in the set indicates an estate, charitable gift, or sponsor block trade; the reason for each is option-monetisation/diversification rather than disclosed cause beyond that.

Net assessment: insiders are unambiguously net sellers, but the selling is concentrated in scheduled option exercises by the same handful of executives (Wang most active with 20 transactions, plus McKechnie, the two officer-directors, and one other officer), heavily weighted to selling into a strong price run. There is no cluster buying, and no first-time CEO purchase to flag. Read it as mild concern at most, not a red flag: option-exercise-and-sell into strength is the most common and least informative kind of insider selling, and the complete absence of any open-market buying simply means the strongest bullish insider signal is not present here. Neutral-to-mildly-cautious.

12. Scenarios

Bull case. The AI build-out keeps Chinese fabs and global packaging houses spending hard, and ACM's diversification beyond cleaning pays off. The SPM and supercritical-CO2 tools take real share in the high-value cleaning segment; ECP and panel-level plating ride the advanced-packaging wave as AI accelerators demand ever more copper and ever larger panels; and furnace, Track and PECVD graduate from evaluation into repeat orders, turning ACM from a cleaning specialist into a multi-product wet-and-thermal supplier. The Oregon facility and the Singapore and Korea wins prove the company can sell outside China, de-risking the concentration story, while the Lingang second building and the Hong Kong H-share listing give it the capacity and capital to scale. Gross margin recovers durably above the midpoint of the model as higher-margin new products mix in, and the "temporary" framing from early 2026 is vindicated. China's localisation drive sustains, ACM out-executes its domestic rivals, and the export-control overhang stays contained to the parent's already-disclosed structure. In this world ACM is the dominant domestic wet-process champion plus a credible international challenger.

Base case. Management roughly delivers the 2026 guide of about 25% revenue growth, with the China cleaning core compounding steadily and ECP and advanced packaging providing the growth on top. Gross margin oscillates around the lower-to-middle of the 42-48% band as semi-critical pricing pressure and new-product ramp costs offset the better mix, broadly as the last few quarters have played out. The international diversification advances but slowly: a handful of tools outside China, Oregon coming online, but mainland China still well over 80% of revenue. Export controls stay an ever-present but manageable constraint rather than a step-change. New platforms progress through qualification at a measured pace, contributing modestly. ACM remains a fast-growing, China-centric equipment company whose results track Chinese fab capex and whose margin line is the swing factor quarter to quarter.

Bear case. The specific adverse scenario is a tightening of the regulatory vice combined with a China capex digestion. A further US export-control escalation restricts ACM Shanghai's component access or its non-China business, or restricts what its customer fabs can build, just as the policy-driven Chinese fab build-out pauses to digest the capacity added during the boom. Orders, which are PO-based and lumpy, fall away quickly. At the same time domestic rivals like Naura and Kingsemi, funded by the same subsidies, compete margin out of the semi-critical and even the critical cleaning segments, so the 41% gross-margin print of Q4 2025 turns out to be a sign of things to come rather than a one-off, and the "temporary" promise breaks. New products slip in qualification, removing the higher-margin mix that was supposed to rescue margins. The governance structure compounds the pain: US holders, owning the business only through a minority-protected stake in a founder-controlled, Entity-Listed Chinese subsidiary, find that even decent operating results do not translate into capital returned to them.

13. Further reading

• ACM Research, China's Most Successful Semiconductor Capital Equipment Provider, Wins At SK Hynix and Intel - SemiAnalysis (Dylan Patel), 2022-07-18 [free] - Argues ACM's rise rests on domestic-supplier status, state support and price more than pure technology, using its SK Hynix and Intel-Oregon qualifications as evidence.

Sources: Motley Fool Q1 2026 transcript, Motley Fool Q4 2025 transcript, Seeking Alpha Q3 2025 transcript, Seeking Alpha Q2 2025 transcript, Investing.com Q1 2025 transcript, Seeking Alpha Q4 2024 transcript, ACMR Q1 2026 8-K earnings release (SEC), ACMR 10-K FY2024 (SEC), ACM Research on US export restrictions (IR), ACM Shanghai 2025 dividend/profit-distribution 8-K (SEC), SemiAnalysis on ACM Research, Kerrisdale Capital ACMR thesis, ACM Research - Wikipedia, Lam Research market cap (companiesmarketcap), Tokyo Electron market cap (companiesmarketcap), SCREEN Holdings (Yahoo Finance).