Hwatsing Technology: The Chokepoint King of Chinese Chipmaking

I. Introduction & Episode Roadmap

On April 23, 2026, company secretary Chen Zhenyin dialed into a conference bridge on the Shanghai Stock Exchange roadshow platform to report strong headline numbers: 2025 revenue rose 36.46% to ¥4.648 billion, marking a fifth consecutive year of rapid top-line expansion.¹ But the key question from analysts on the call focused on profitability rather than growth. Question 5 on the transcript was direct: the company's gross margin has now declined three years running — is this because you lack bargaining power with your downstream customers?²

That question compresses the fundamental tension of Hwatsing's business into a single sentence.

Hwatsing Technology Co., Ltd. (华海清科, 688120.SS, listed on the Shanghai Stock Exchange's STAR Market) manufactures chemical mechanical polishing (CMP) equipment. These tools planarize silicon wafers to atomic smoothness between the dozens of layers required in modern semiconductor fabrication. According to company disclosures, Hwatsing is the only domestic manufacturer to commercialize 300mm CMP tools at volume, accounting for more than 90% of domestic CMP equipment sales.² By June 30, 2026, cumulative shipments of its core CMP tools surpassed 1,000 units.³

That market share represents one of the highest concentrations of domestic market dominance achieved by a Chinese semiconductor equipment supplier in any major process step. Yet even with that position, the company's financial economics face mounting pressure.

The financial results over the last three reporting periods highlight this divergence between volume growth and margin capture. In 2024, top-line revenue expanded 35.82% to ¥3.406 billion while net profit grew faster, rising 41.40% to ¥1.023 billion.⁴ In 2025, top-line growth remained steady at 36.46%, pushing revenue to ¥4.648 billion, but net profit growth slowed to 5.89% at ¥1.084 billion as gross margin slipped to 41.81%.¹ In the first half of 2026, revenue increased 35.58% year-over-year to ¥2.643 billion, while net profit grew 11.44% to ¥563 million; cost of sales rose 43.34%, outpacing top-line growth and contracting gross margin to roughly 43.0% from about 46.1% in the prior-year period.³

While top-line growth has remained consistent, profitability has decoupled from revenue expansion. The central investment debate is whether this margin compression represents a temporary phase of qualifying new product lines at customer fabs—as management asserts—or a structural limitation inherent to selling capital equipment to a concentrated group of powerful domestic buyers.

The company's evolution spans five distinct phases. It began in a tribology laboratory at Tsinghua University, where professor Lu Xinchun researched surface friction and mechanical wear. It moved to the fab floor, where the first Chinese-built 300mm CMP tool underwent extended customer qualification while engineers addressed particle-count criteria. It accelerated through the geopolitical shift between 2018 and 2020, which elevated a loss-making equipment vendor into a critical national supplier. It reached a public-market milestone with its 2022 IPO, followed by platform expansion into wafer thinning, wafer reclaim, wet cleaning, and ion implantation. By 2026, it faces the core valuation question: the economic durability of a domestic near-monopoly selling into an equally concentrated and strategically vital customer base.

Scale provides essential context for Hwatsing's market position. Relative to global equipment peers, the company remains compact. Total headcount reached 2,654 at year-end 2025, including 876 research and development personnel—representing about one-third of total staff—and 820 production workers.¹⁷ By mid-2026, total assets stood at ¥14.4 billion and shareholders' equity was ¥7.9 billion.³ Hwatsing's advantage stems not from sheer size, but from strategic position: it holds a dominant domestic share in a critical process step while remaining small enough that capital expenditure changes at a single major customer can materially impact earnings.

That market position reflects a strategic paradox. Hwatsing operates with significant policy protection alongside substantial market exposure. Trade restrictions and localization mandates create sustained domestic demand for Chinese CMP tools, reinforced by the technical requirement for wafer planarization in semiconductor manufacturing. Conversely, those same trade dynamics restrict Hwatsing's international reach, leaving mainland China to generate 99.08% of its 2025 revenue.¹¹ Furthermore, its domestic client base consists primarily of major state-backed semiconductor manufacturers with strong bargaining leverage over equipment suppliers. Hwatsing demonstrates how a supplier can maintain an indispensable technical role while remaining economically constrained by customer concentration.

The analysis evaluates the primary arguments supporting Hwatsing's valuation against its financial disclosures and operational history. The narrative follows a chronological structure: the academic origins and thirteen-year development timeline preceding commercial sales; the qualification process at Semiconductor Manufacturing International Corporation (SMIC); the impact of industrial policy shifts; the allocation of proceeds following the IPO; the unit economics and pricing dynamics of the CMP market; the commercial progress of product diversification; the ownership and management control structure; and an analytical framework detailing the key operational metrics that govern future performance.

---

↑ Back to Top

II. Tsinghua Lab to Industrial Reality: The Genesis of Chinese CMP (2000–2013)

Start with the physics, because without it the rest of the manufacturing challenge makes little sense.

A modern microchip is not flat; it resembles a microscopic skyscraper. A dozen or more layers of metal wiring are stacked above a foundation of transistors, with each layer separated by insulating dielectric and linked by vertical conductive plugs. Whenever a new layer is deposited, the surface underneath becomes uneven: metal fills trenches and overflows, while dielectric accumulates unevenly. Attempting to print subsequent circuit patterns onto an uneven surface using a lithography scanner results in optical distortion. Lithography lenses operate with an extremely narrow depth of focus—holding only a few hundred nanometers of vertical margin in sharp resolution. Printing across microscopic hills and valleys causes portions of the circuit pattern to fall out of focus and blur.

Planarization between layers addresses this issue through Chemical Mechanical Polishing (CMP). In this process, a silicon wafer is pressed face-down against a rotating polyurethane pad while flooded with an abrasive slurry of nanoscale particles suspended in reactive chemistry. The chemical action softens the top atomic layers of the film, while the mechanical action of the abrasive slurry sweeps the softened material away. When executed precisely, material is removed at a controlled rate, yielding a disc surface flat to within a few nanometers across a 300mm wafer—a level of precision analogous to flattening a football field to within a single millimeter.

Conversely, an improper CMP process can scratch the wafer, deposit particulate contaminants, cause uneven dishing in soft copper lines, or polish through a dielectric layer entirely. Because a single polishing error can ruin an entire lot of wafers representing weeks of cumulative fabrication, fab operators view CMP as one of their most conservative procurement decisions. This high bar for reliability historically concentrated the global CMP market into the hands of just two suppliers.

CMP also resisted early replication due to its physical complexity. Unlike process steps governed by predictable physical models—such as deposition rates, etch selectivities, or ion ranges—CMP involves a complex interplay of chemical, mechanical, and fluid dynamics. The polishing pad surface degrades with wear, slurry particles can agglomerate unpredictably, the wafer flexes under pressure, and local removal rates constantly change alongside changing surface topography. Because the process cannot be derived purely through theoretical modeling, optimal recipes must be discovered empirically across thousands of test wafers and encoded into hardware controls. As a result, the competitive moat in CMP has traditionally been experiential rather than theoretical, making a tribology laboratory a logical starting point.

The underlying academic research began around 2000, when Luo Jianbin (雒建斌)—later an academician of the Chinese Academy of Engineering—and Lu Xinchun began studying CMP at Tsinghua University's tribology laboratory.⁵ Tribology, the study of friction, wear, and lubrication, provided the framework for understanding how polishing pads, chemical slurries, and silicon wafers interact under pressure. Lu, born in 1966, was a professor in Tsinghua's Department of Mechanical Engineering.⁶ Over several years, the research team focused on fundamental primitives: nanoscale polishing, ultra-clean nanoparticle removal, nanometer-precision in-line film thickness measurement, and the control algorithms required to coordinate the system.⁷

Initial financial backing was provided primarily by state research grants. Starting in October 2008, the Tsinghua group secured project funding under National Science and Technology Major Project 02 (国家科技重大专项02专项), China's dedicated initiative for semiconductor equipment and materials.⁵ By the time of Hwatsing's initial public offering, the R&D team had led or co-led two Major Project 02 mandates alongside three other national-level research projects.⁷ Non-dilutive state grants funded years of early prototype development that private venture capital would have been unlikely to finance.

This state-supported capital model illustrates how domestic equipment initiatives crossed early technical hurdles. While venture capital funds typically operate on five-to-seven-year return horizons, CMP required roughly thirteen years from initial Tsinghua research to its first fab installation, and another seven years to reach an installed base of approximately 50 units. State grants absorb this multi-year timeline because the primary return metric is strategic capability rather than near-term financial yield. However, this same non-dilutive funding mechanism remains available to future state-backed entrants, creating potential long-term domestic competition.

Hwatsing (华海清科) was incorporated on April 10, 2013, in Tianjin's Jinnan District by Luo and Lu to commercialize the laboratory's intellectual property.⁵ Originally organized as Tianjin Hwatsing Clearing Technology Co., Ltd. (天津华海清科机电科技有限公司), the business reorganized into a joint-stock company in March 2020 ahead of its public listing.³

The initial product scope extended beyond basic polishing tools. By the time of its IPO prospectus, Hwatsing had developed volume-production equipment for non-metal dielectric CMP, metal film CMP, and silicon CMP, addressing three distinct process families with unique chemical formulations and failure modes.⁷ The company accumulated several domestic technical recognitions, including a first prize at the Tianjin Science and Technology Awards, a special prize from the China Machinery Industry Science and Technology Awards, China Semiconductor Innovation Product and Technology awards in 2018 and 2019, and a China Good Design Gold Award.⁷ By year-end 2020, Hwatsing held 173 granted patents, including 101 invention patents.⁷

The uncomfortable base rate

Despite these academic and developmental milestones, commercial adoption presented a separate challenge. While early intellectual property established a foundational starting point, disclosures in the company's prospectus highlighted the distance between laboratory success and commercial market share.

In its 2021 prospectus, Hwatsing noted that it lagged global incumbents "by a considerable gap" in product maturity, technology, and market presence. The company disclosed that Applied Materials (应用材料) and Ebara Corporation (荏原製作所) together controlled more than 90% of the global CMP equipment market, and served as the exclusive tool suppliers for advanced production lines below 14nm.⁷ As of December 31, 2020—seven and a half years after incorporation and two decades after Tsinghua research began—Hwatsing had delivered a cumulative total of 58 CMP tools, with an order backlog of 35 units.⁷

A cumulative shipment total of 58 tools over seven years underscores the timeline required to convert academic research into commercial equipment revenue. Solving the underlying physics was a prerequisite, but commercial progress ultimately depended on fab operators willing to qualify unproven machinery on active production lines.

That initial willingness was secured on a fab floor in Beijing.

↑ Back to Top

III. The Crucible of Fab Qualification & The SMIC Breakout (2013–2018)

On August 2, 2015, a truck delivered Hwatsing's first 300mm CMP tool—the Universal-300—to Semiconductor Manufacturing International Corporation (SMIC).⁸ It marked the first time a domestically built CMP machine entered a Chinese fab for validation on an active production line. For a long stretch afterward, however, the machine remained idle as qualification got underway.

Understanding that delay requires examining what a fab operator actually purchases. A fab buys not just hardware, but a process recipe—a fine-tuned set of operational parameters, verified over months, that reliably transforms incoming wafers while maintaining strict targets for thickness, defectivity, and surface uniformity. The machine is merely the delivery vehicle; the validated recipe is the asset, and every recipe is developed for a specific tool architecture.

Hwatsing's initial public offering prospectus detailed this timeline with unusual candor. A first demonstration tool installed for a new customer running a new process typically required 12 to 18 months from shipment to final acceptance, the point at which revenue could finally be recognized. By contrast, follow-on tools for established recipes required only three to six months.⁷ That dynamic defined the company's early cost structure: Hwatsing had to manufacture tools, ship them, deploy resident field engineers, and wait more than a year to confirm whether the customer would accept the machine and pay for it.

SMIC formally accepted the Universal-300 in 2016. In 2017, Hwatsing delivered an upgraded model, the Universal-300 Plus, to SMIC for a second round of validation. That same year, the tool cleared multiple process qualifications, including oxide and tungsten CMP.⁸ These process steps represent the core workhorses of wafer fabrication: oxide CMP planarizes dielectric insulation between conductive layers, while tungsten CMP polishes the vertical metal plugs that connect adjacent circuit layers.

The economics of waiting

This prolonged validation cycle placed severe demands on the company's early income statement. Hwatsing relied exclusively on direct sales, securing contracts through commercial negotiations or competitive tenders.⁷ Once a contract was signed, each tool was custom-built and configured. According to company disclosures, manufacturing concluded with specialized software configuration and full system testing against customer specifications before the equipment could be warehoused and shipped.⁷ After shipment, service engineers stationed permanently at the customer's fab spent months installing, calibrating, and testing the tool. Only after the client issued formal acceptance did Hwatsing recognize revenue.⁷

As a result, Hwatsing's working capital cycle ran in reverse. Raw materials, manufacturing labor, and dedicated field engineering teams required immediate cash outlays, while revenue lagged by a year or more. At the same time, research and development spending could not slow down; equipment qualified for a 28-nanometer process node could not automatically process 14-nanometer wafers. In its prospectus, Hwatsing framed this structural challenge as a core risk: high initial R&D intensity, low sales volume, high unit manufacturing costs, and elevated expense ratios without the benefits of scale produced net losses in 2018 and 2019, delaying net profitability until 2020.⁷

Under these conditions, on-site technical support functioned as a strategic pillar rather than an operational cost center. Having resident engineers available around the clock to resolve defectivity spikes meant the difference between retaining a customer and losing future orders to incumbent suppliers like Applied Materials. Hwatsing's stationed-engineer model—deploying dedicated personnel to handle installation, debugging, warranty service, repairs, and technical consulting on-site—provided the operational foundation for its subsequent domestic market share gains.⁷

What certification actually bought

Industry commentary often assumed that clearing SMIC's qualification bar immediately granted Hwatsing customer lock-in and pricing power. However, the financial disclosures from that period paint a starkly different picture.

While revenue grew rapidly from ¥35.66 million in 2018 to ¥210.93 million in 2019 and ¥385.89 million in 2020, underlying profitability remained fragile. Net profit excluding non-recurring items was negative ¥67.84 million in 2018 and negative ¥47.72 million in 2019, before turning slightly positive at ¥14.61 million in 2020.⁷ In 2020, although headline net profit attributable to shareholders reached ¥97.79 million⁷, underlying core operations generated only ¥14.61 million. Government grants accounted for roughly six-sevenths of reported net income that year, leading Hwatsing to explicitly list subsidy dependency as a key risk factor, noting that any significant reduction in government support would severely impact financial performance.⁷

Customer concentration was equally extreme. The company's top five customers accounted for 99.09% of total sales in 2018, 94.96% in 2019, and 85.71% in 2020.⁷ Effectively, Hwatsing's commercial survival depended on just three or four chipmakers. Furthermore, company filings showed high levels of overdue accounts receivable, explaining that major state-backed customers—including Hua Hong Group, YMTC, and SMIC—typically settled payments on their own extended timelines.⁷

These financial dynamics reflected a supplier operating at the leverage of its buyers rather than a vendor exercising commercial pricing power.

Fab qualification represented a formidable technical barrier, and Hwatsing paid the steep operational entry fee that no other domestic peer had managed to clear. Yet entering the market was not equivalent to controlling it. Through 2018, qualification demonstrated hardware performance, but commercial viability remained vulnerable to external forces beyond the company's direct control.

That external catalyst arrived when trade policy shifts began reshaping the semiconductor industry landscape.

↑ Back to Top

IV. Geopolitical Catalyst & The Domestic Substitution Surge (2018–2022)

While Hwatsing's initial R&D established technical feasibility, external trade actions fundamentally altered its market trajectory. In April 2018, after the U.S. Department of Commerce cut ZTE off from American suppliers, procurement managers across Chinese semiconductor fabs began an overhaul of their supply-chain risk models.

Subsequent U.S. trade enforcement actions followed in rapid succession: Huawei's addition to the Entity List in 2019, SMIC's listing in December 2020, and expanding export controls targeting advanced logic and memory makers. Inside domestic fabs, procurement priorities inverted. Prior to 2018, purchasing domestic equipment meant accepting technical risk to achieve lower capital costs. After 2018, relying exclusively on foreign vendors introduced supply-chain risks capable of idling a multi-billion-dollar fabrication facility. Consequently, domestic substitution (guocan tidai) shifted from a long-term policy objective into an immediate operational mandate.

This policy shift altered fab operations by reallocating scarce qualification slots. Validating a new tool on an active production line requires substantial engineering resources and dedicated wafer runs that otherwise produce saleable chips. On a 28-nanometer logic line, fab operators had previously allocated minimal qualification capacity to unproven domestic vendors. As second-sourcing every critical process step became a strategic requirement, fabs opened validation slots to domestic equipment suppliers. For Hwatsing, whose primary operational bottleneck had been securing extended tool testing in customer fabs, this adoption barrier effectively dissolved.

Hwatsing's customer base expanded rapidly as a result. By the end of 2020, its tools were installed at major domestic chipmakers, including SMIC, YMTC, Hua Hong Group, CXMT (长鑫存储), United Semiconductor Xiamen (厦门联芯), CanSemi Guangzhou (广州粤芯), Shanghai Jita (上海积塔)—and Intel.⁷ Intel's inclusion was notable because the vendor operated outside Chinese domestic policy mandates. Although Hwatsing has not disclosed the volume or financial scale of that relationship, Intel's presence on the customer roster served as a rare third-party reference point from an international manufacturer.

The financial results reflected this market shift. Revenue rose from ¥385.89 million in 2020⁷ to ¥4.648 billion in 2025¹—representing a twelvefold expansion over five years. In a July 2024 interview, then-general manager Zhang Guoming (张国铭) cited a five-year compound annual growth rate of 86%, while asserting that key CMP tool components had become substantially self-controllable.⁹

By 2026, company disclosures stated that Hwatsing's tools accounted for more than 90% of domestic CMP equipment sales, covering logic, 3D NAND, and DRAM process platforms while increasing penetration on domestic 300mm advanced lines.² Customs and trade data mirrored this trend: China's CMP equipment imports totaled $793 million in 2025, with the foreign-supplied market share dropping to 63%—indicating that domestic equipment suppliers had captured nearly 40% of the total domestic market, up from near zero a decade earlier.¹⁰

Was it the product or the politics?

Distinguishing between technical capability and geopolitical drivers is critical to evaluating Hwatsing's long-term competitive position.

Geopolitical policy created the commercial opportunity, while Hwatsing's previous decade of R&D provided the baseline capacity to capture it. Policy directives could not force a fab to run production wafers through non-functional hardware. Conversely, technical capability alone would unlikely have produced a twelvefold revenue surge in five years against global incumbents controlling over 90% of the market, absent an exogenous catalyst that constrained access to foreign equipment.

However, available evidence does not support the assertion that Hwatsing displaced incumbents such as Applied Materials and Ebara based on superior total cost of ownership. Hwatsing has not published comparative performance data regarding wafer throughput, mean time between failures, or cost per polished wafer against global peers, and its initial prospectus explicitly acknowledged a significant technology and scale gap.⁷ In the absence of published comparative metrics, technical parity cannot be assumed.

Hwatsing's geographic revenue distribution provides further context regarding its global competitiveness. In 2025, mainland China generated 99.08% of total revenue.¹¹ If the company's tools offered superior overall economics on open market terms, commercial adoption would likely extend into major international semiconductor hubs such as Taiwan, South Korea, or Singapore. Instead, export restrictions and domestic focus leave the equipment's competitiveness untested in markets where buyers retain full access to foreign incumbents.

Hwatsing's initial growth expansion was policy-catalyzed and capability-enabled. Its long-term defensibility depends on whether it can maintain market share as domestic customers complete their initial substitution phase and shift focus from supply security toward unit cost and hardware performance. The early evidence regarding that transition appears in the company's gross margin performance.

First, the money.

---

↑ Back to Top

V. STAR Market IPO, M&A, & Capital Deployment (2022–2024)

On June 8, 2022, Hwatsing listed on Shanghai's STAR Market at ¥136.66 per share, issuing 26.67 million shares. The offering generated ¥3.644 billion in gross proceeds and ¥3.490 billion net—far exceeding its original ¥1.0 billion target. By bringing in ¥2.49 billion in surplus capital, the company ended up raising roughly three and a half times its stated requirement.¹² Market enthusiasm was immediate: the stock opened 72% above its offer price and ended its trading debut with a 64% gain.¹³¹⁴

An over-subscription of that scale reflected market conditions rather than fundamental quality alone. Hwatsing stood as the sole listed pure-play domestic CMP manufacturer at a moment when substantial domestic capital was directed toward semiconductor self-sufficiency, while the STAR Market's pricing mechanism allowed intense demand to set the clearing valuation. Consequently, management suddenly held nearly ¥3.5 billion in cash—far more than requested—and faced the immediate challenge of allocating that windfall.

In its prospectus, Hwatsing outlined four intended uses for the funds: expanding high-end CMP equipment manufacturing capacity, funding advanced semiconductor equipment R&D, constructing a wafer reclaim facility, and supplementing working capital.¹²

Where the capital went

Three primary capital deployments defined the post-IPO period.

Capacity. The company expanded its physical footprint. By April 2025, management confirmed that the Beijing subsidiary's high-end equipment R&D and manufacturing hub, along with the Tianjin phase-two expansion, were both complete and ramping up production.⁴ By mid-2026, a Shanghai integrated-circuit equipment R&D and manufacturing facility and a Kunshan wafer reclaim expansion had cleared investment filings and environmental approvals.³

Wafer reclaim. Hwatsing entered the services market by processing used monitor and dummy wafers—stripping and repolishing them for reuse in customer fabs. Leveraged directly from the company's CMP technical foundation, this service created recurring customer relationships; by 2026, Hwatsing reported long-term bulk orders from several major chipmakers. Its Tianjin facility was operating at its full capacity of 200,000 wafers per month, while a second facility in Kunshan was under construction to add 400,000 wafers of monthly capacity, aiming for a combined monthly output of 600,000 units at full ramp.²³

Ion implantation through acquisition. On December 24, 2024, Hwatsing announced it would spend up to ¥1.005 billion in cash to acquire the remaining 82% equity interest in Xinyu Semiconductor (Shanghai) Co., Ltd. (芯嵛半导体), an ion implantation tool developer in which it held a minority stake, turning it into a wholly owned subsidiary.¹⁵¹⁶ Xinyu was unprofitable at the time, posting net losses attributable to owners of ¥24.42 million in 2023 and ¥15.16 million over the first five months of 2024. The transaction included explicit revenue targets: audited consolidated revenue of at least ¥45 million in 2024, ¥105 million in 2025, and ¥340 million in 2026, with at least 80% of each annual target derived from customer-accepted equipment rather than spare parts, maintenance, or materials.¹⁶³

That 80% customer-acceptance requirement introduced key structural discipline to the acquisition. By tying earn-out milestones specifically to accepted equipment rather than consumables or maintenance services, the structure prevented revenue targets from being inflated through non-equipment channels.

Testing the capital-allocation claim

While optimistic market narratives held that IPO proceeds were deployed seamlessly into high-return core R&D, the financial record demonstrates a more nuanced reality.

On one hand, the expansion into wafer reclaim—frequently viewed with skepticism as a lower-margin distraction—proved to be the company's most profitable segment. In 2025, Hwatsing's semiconductor equipment gross margin fell 1.94 percentage points to 40.88%, whereas its semiconductor services gross margin rose 1.81 percentage points to 48.21%, driven by a 61.23% increase in service revenue to ¥594 million.¹⁷¹ Rather than diluting overall profitability, the services division became the sole segment expanding its gross margins.

On the other hand, overall capital efficiency contracted as the balance sheet expanded. Return on equity stood at 14.53% for 2025.¹ In the first half of 2026, weighted average return on equity fell to 7.24%, compared to 7.50% in the prior-year period.³ By June 2026, total assets grew 9.62% in six months to reach ¥14.402 billion, primarily driven by rising inventory levels.³ Across its expanded asset base, Hwatsing required progressively more capital to generate each unit of net profit.

Capital management was further highlighted by a swift shift in financing strategy. In late August 2025, Hwatsing announced plans for an H-share listing in Hong Kong.¹⁸ Yet on April 22, 2026—less than eight months later—management canceled the Hong Kong listing and announced a proposed A-share private placement of up to ¥4 billion, allocating ¥1.342 billion to the Shanghai manufacturing base, ¥445 million to wafer reclaim expansion, and ¥2.213 billion to advanced equipment R&D.¹⁹²⁰ During an April 2026 investor call, executive management cited shifting macroeconomic conditions, capital market dynamics, and corporate strategy to explain the pivot.²

While altering financing plans within eight months does not indicate improper conduct, it raises questions regarding long-term capital planning. If executed, the proposed ¥4 billion placement—combined with the ¥3.49 billion raised in its IPO—would push total equity capital raised past ¥7 billion, compared to cumulative dividend distributions of approximately ¥270 million since listing.¹² For 2025, the company proposed a cash dividend of ¥141 million, representing 13.02% of net profit.³ Hwatsing has consistently prioritized capital retention and equity fundraising over shareholder payouts—a strategic posture common among capital-intensive equipment makers in expansion phases, but one that exposes existing shareholders to equity dilution ahead of realized returns.

Which brings the analysis to the core question of what those returns actually look like.

↑ Back to Top

VI. Core Business Economics: CMP Deep-Dive & Industry Structure

Picture the machine. A 300mm CMP tool is roughly the size of a large van, housing a robotic arm that transfers wafers between load ports, polishing platens, and a cleaning module. The wafer is held face-down in a carrier—the polishing head—which represents the single most valuable piece of intellectual property in the system.

Why the head? A wafer does not polish evenly: the center wears at a different rate than the edge, the polishing pad degrades, and the slurry distributes unevenly. The head is therefore divided into concentric pressure zones that are controlled independently, allowing the system to push harder on the center and lighter on the edge—or vice versa—to compensate. Mastering zone pressure control across thousands of wafers while pad surfaces wear down and slurry chemistry drifts is where two decades of tribology research pays off commercially. Combined with endpoint detection—optical or motor-torque sensors that signal when the tool reaches the layer beneath—these two subsystems separate a working CMP tool from an expensive wafer-damaging machine.

Why CMP volume grows faster than wafer volume

The demand mechanism behind CMP relies on layer count rather than raw wafer volume. Every additional metal layer in a logic chip requires planarization. As process nodes advance, layer counts rise, driving up the number of CMP steps per wafer; mature nodes require substantially fewer planarization passes than advanced nodes, a gap that has widened over the past decade.²¹

The structural curve is even steeper in memory production. 3D NAND architectures expand vertically—from 128 layers to 232 and beyond—fabricated in stacked decks. Each deck requires tungsten CMP to clear excess metal from wordline fills, while each inter-deck transition requires oxide planarization because incoming surface topography at extreme stack heights can exceed 500 nanometers and must be flattened to within a few nanometers before the next deck is deposited.²² Consequently, every added deck increases CMP intensity.

Advanced packaging introduces a third demand driver. Copper-to-copper hybrid bonding—the core technology behind High Bandwidth Memory (HBM) stacks and advanced 2.5D and 3D packages—requires bonding surfaces that are flat and clean at the sub-nanometer scale, ensuring copper pads make direct metallic contact without gaps. This requirement presents one of the most technically demanding CMP challenges in semiconductor manufacturing.

This dynamic gives Hwatsing direct exposure to artificial intelligence infrastructure capex. Management disclosed that several of its advanced-node CMP tools have entered the production line of a major memory manufacturer as baseline equipment for HBM fabrication.²

The product portfolio, as of mid-2026

The company's product lineup has broadened beyond the single Universal-300 delivered to SMIC in 2015. In the first half of 2026, Hwatsing reported that the Universal-S300, featuring a new stacked-layout architecture, cleared customer validation and secured volume orders from multiple clients. The upgraded Universal-S300A shipped its initial unit to a major memory customer for process validation, while the Universal-300FS entered a leading domestic logic customer's supply chain with large-volume purchase orders. Additionally, the Universal-H300 completed volume delivery and customer acceptance. Hwatsing also shipped China's first fully automated panel-level CMP production tool for 510×515 millimeter substrates—the Master-P510APEX—to an advanced packaging customer.³

Stress-testing the moat

A key question for investors is whether controlling over 90% of the domestic CMP market translates into commercial pricing power.

Financial disclosures from recent reporting periods indicate that market share has not yielded expanding margins. In 2025, revenue increased 36.46% to ¥4.648 billion while cost of sales rose 39.80%, pushing gross margin down 1.39 percentage points to 41.81%.¹⁷ Within the core equipment segment, gross margin fell 1.94 percentage points to 40.88%.¹⁷ In the first half of 2026, the gap widened: revenue grew 35.58% year-over-year to ¥2.643 billion, while cost of sales expanded 43.34%.³ Operating expense ratios also increased, with period expenses rising 2.37 percentage points to 22.10% of revenue in 2025, driven by a 37.98% increase in R&D spending to ¥543 million, or 11.69% of revenue.¹¹⁷

Customer concentration remains a defining structural feature of the business. In 2025, the top five customers accounted for ¥2.978 billion, or 64.07% of total revenue. The single largest customer represented ¥1.275 billion, or 27.42% of revenue, while the second largest generated ¥772 million, or 16.60%.¹⁷ Two of the top five customers were new entrants to the list that year, which management cited as evidence of client diversification.¹⁷

This customer structure reflects two distinct dynamics. Customer concentration has decreased from 85.71% of sales in 2020 to 64.07% in 2025, representing a measurable broadening of the customer base.⁷¹⁷ However, the annual report anonymizes these accounts as Customer 1 through Customer 5, making market attributions to specific chipmakers inferential rather than confirmed by disclosure. The report does explicitly disclose that Customer 4—accounting for ¥364 million, or 7.84% of revenue—is a related party, and that former director and general manager Zhang Guoming served as a director of Hua Hong Group until December 2024, with Hua Hong entities listed in the related-party transaction tables.¹⁷ While related-party revenue of 7.84% is modest, it remains a relevant factor when evaluating customer independence.

When analysts directly questioned management regarding margin compression during the April 2026 earnings call, executives attributed the decline to newly commercialized product lines that remain in early market-development phases at lower initial margins. Management stated that the technological advantages of its core products, market competitiveness, and long-term downstream demand trends remained unchanged.² Addressing why profit growth trailed top-line expansion, management reiterated the impact of early-stage product margins alongside rising headcount costs.² On the April 2025 call, management offered a specific structural explanation for the prior period: the 2024 gross margin reduction stemmed primarily from an accounting reclassification under Ministry of Finance Interpretation No. 18, which shifted warranty provisions from selling expenses into cost of sales with retrospective restatement.⁴

The 2024 accounting adjustment mechanically reduced reported gross margins without representing an underlying economic change, complicating multi-year margin comparisons across that reporting boundary. However, subsequent margin compression in 2025 and mid-2026 cannot be attributed to reclassification.

Management's assertion—that gross margin dilution reflects temporary product-mix effects—presents a testable claim. If accurate, gross margins should stabilize and recover as new tool lines transition from initial low-margin validation shipments to higher-margin volume orders. Conversely, if margin contraction reflects structural price concessions demanded by concentrated domestic buyers, margins will continue to slip regardless of product maturity.

Disclosures from the April 2025 investor call shed light on pricing dynamics. Asked whether advanced-node products faced customer pricing pressure, management acknowledged that the company had operated under competitive conditions since launch, noting that price negotiation was a constant factor and that price adjustments remained within a normal range.⁴

Verdict on the moat claim: The high switching costs protecting Hwatsing's installed base are well established, backed by repeat orders, multi-product tool qualifications, and a 12-to-18-month customer validation cycle that creates significant barriers to entry. However, the premise that market dominance grants pricing power is countered by three consecutive years of gross margin contraction alongside 35%-plus top-line growth. The defensible framing is that Hwatsing possesses a durable volume franchise with constrained pricing power, operating as a near-sole-source supplier to buyers that are equally concentrated and strategically dominant. The key metric governing recovery is the semiconductor equipment segment gross margin disclosed in annual filings, distinct from the consolidated gross margin, which is currently supported by an expanding services business.

If core equipment margins cannot expand, the company's valuation hinges on whether its newer product categories can deliver higher returns.

↑ Back to Top

VII. The Platform Expansion & Optionality Bets

Every equipment company eventually faces the same wall. You dominate one process step. That step is maybe 6-8% of a fab's tool budget. To keep growing at 35%, you either need the whole industry to grow at 35% forever, or you need more steps.

Hwatsing's answer is what it calls the 装备+服务 "equipment plus services" platform strategy, and by mid-2026 it spanned seven distinct product families beyond core CMP.³ Each is at a different stage of the same journey Hwatsing already made once with CMP — which makes the portfolio a natural experiment in whether the company can repeat itself.

Thinning equipment (减薄装备). Grinding wafers thin — down to tens of microns — is what makes 3D stacking possible. You cannot stack twelve DRAM dies into an HBM cube unless each is thin enough that the stack fits in a package and heat can escape. Hwatsing reported that cumulative shipments of its 12-inch ultra-precision wafer thinning tools exceeded 20 units, that its integrated thinning-and-polishing machine won a China Good Design Gold Award, and that repeat orders had come from multiple leading customers.²¹⁷ In the first half of 2026, it shipped first units of the Versatile-GH300 for advanced memory, the Versatile-GN200 for compound semiconductors, and an 8-inch ring-retained thinning tool, the Versatile-GR200.³ Of all the platform bets, this is the most advanced and the most credible: it is adjacent to CMP technically, sells to the same customers, and rides HBM and advanced packaging demand directly.

Wafer reclaim (晶圆再生). Discussed above — the highest-margin segment in the company, growing 61% in 2025.¹¹⁷ The bear objection is that reclaim has low technical barriers and will attract competition. That is fair as a forward risk. But the 2025 margin data does not yet show it, and the multi-year bulk contracts with leading fabs suggest a stickier position than a pure commodity service.

Wet cleaning (湿法装备). Post-CMP cleaning is the natural extension — every polish must be followed by a clean. In the first half of 2026 the company shipped its first eight-chamber 300mm single-wafer cleaner to a leading domestic silicon wafer manufacturer, and reported that several cleaning tools had completed customer validation with revenue recognized.³ Still early. The competitive problem here is severe: 盛美上海 ACM Research Shanghai is an entrenched domestic specialist in exactly this category with far more installed base.

Dicing and edge polishing (划切装备 / 边缘抛光装备). Both remain in customer validation and early volume. The Master-BN300 edge polisher won repeat orders from a leading domestic chip company in the first half of 2026 — a meaningful signal, since repeat orders are the first real proof of commercial adoption.³

Ion implantation. The dark horse, and the one that most deserves scrutiny.

The optionality claim, tested against its own earn-out

The thesis claim is that Hwatsing will replicate its CMP dominance in ion implantation. The company's own performance undertakings are the cleanest available test, because they were negotiated by insiders who knew the pipeline.

The Xinyu commitments required consolidated revenue of at least ¥45 million in 2024, ¥105 million in 2025, and ¥340 million in 2026.¹⁶³ Actual 2025 ion implanter revenue was ¥116 million — a beat on the ¥105 million hurdle, but on total company revenue of ¥4.648 billion, that is 2.5% of the business.¹¹ The 2026 hurdle of ¥340 million requires roughly tripling in a single year.

On product, the progress is real: Hwatsing reported that its 300mm high-current implanters have been delivered in volume to multiple leading domestic chipmakers, that the iPUMA-LE shipped to a leading advanced memory company, and that the first 300mm low-temperature implanter, the iPUMA-LT, is in customer validation. Management stated it has now achieved full model coverage across the high-current implanter category.²³

But certification is not commercialization, and this company's own history is the proof: 58 CMP tools in seven years after the physics was solved. Ion implantation faces entrenched competition that CMP in 2015 arguably did not — Applied Materials globally, plus domestic players including 凯世通 Kingstone Semiconductor under 万业企业 Wanye Enterprises, and the 中国电子科技集团第48研究所 CETC 48th Research Institute. High-current implant is also the easiest of the implant categories to enter; the harder, higher-value segments are medium-current and high-energy, where Hwatsing has disclosed less progress.

Verdict on the optionality claim: the record does not reject it, but it narrows it substantially. Hwatsing has demonstrated it can convert technical milestones into revenue — the CMP franchise is the existence proof — but it converts them slowly, over half-decade horizons, and the ion implant business is at roughly the stage CMP occupied around 2018. Framing it as a near-term second growth pillar is not supported. Framing thinning equipment that way is much better supported. The observable test for 2026 is straightforward: whether Xinyu clears the ¥340 million earn-out hurdle, with at least 80% from accepted tools. A miss would be the clearest available signal that platform expansion is running behind the plan management sold to its own board.

The people making these calls are, notably, mostly new to their seats.

---

↑ Back to Top

VIII. Management, Governance, & Incentives

On April 1, 2025, Hwatsing's board received two resignation letters at once.²³

Lu Xinchun, the Tsinghua professor whose academic laboratory research launched the company, stepped down as chairman and director at age 58, citing an adjustment of work arrangements. He remained chief scientist and a core technical staff member. Meanwhile, director and general manager Zhang Guoming resigned upon reaching statutory retirement age, staying on as chief adviser.²³

Into both roles stepped Wang Tongqing (王同庆), then 42 years old. Born in 1983, Wang holds a doctorate in mechanical engineering from Tsinghua University, completed postdoctoral research in its Department of Precision Instruments, and served as a research fellow in the Department of Mechanical Engineering. He joined Hwatsing as R&D director before rising through assistant to the general manager and deputy general manager.²⁴¹⁷ Wang was elected chairman, appointed general manager, and named chair of the board's strategy and ESG committee—a concentration of executive authority in a single individual that is common among Chinese listed companies.²⁴

This leadership transition appears logical on paper: the founding scientist returns to research, while an operational executive assumes management of the enterprise. Lu remained the company's highest-paid executive in 2025, earning ¥6.601 million.²⁵ However, this transition leaves an open operational question: the two most consequential capital decisions of the past year—the reversal of the Hong Kong listing and the proposed ¥4 billion private placement—were both enacted under the new leadership within twelve months of taking office.

Who actually controls the company

The ownership structure of Hwatsing is frequently misunderstood.

Tsing-Tech Innovation (清控创业投资有限公司) is the direct controlling shareholder, holding 139,245,242 shares, or roughly 28.1% of the 494.73 million shares outstanding as of June 30, 2026.³ Lu Xinchun personally held 24,351,046 shares, or about 4.9%, while co-founder Luo Jianbin also ranked among the top ten shareholders.³

However, the ultimate controlling entity is not Tsinghua University. In mid-2022, shortly after the company's initial public offering, 100% of Tsinghua Holdings (清华控股) was transferred to the Sichuan provincial state-asset regulator. The entity was renamed Tianfu Qingyuan (天府清源) and subsequently transferred to Sichuan Energy Investment. As a result, Hwatsing's actual controller became the Sichuan Provincial State-owned Assets Supervision and Administration Commission (四川省国资委, or Sichuan SASAC).²⁶ In early 2025, the indirect controlling shareholder shifted again to Sichuan Energy Development Group following a strategic reorganization of state enterprise assets, while the direct controlling shareholder and actual controller remained unchanged.²⁷ The company's interim report for 2026 confirms that the parent company is Tsing-Tech Innovation and the ultimate controller is the Sichuan SASAC.³

This distinction is more than nominal. It means Hwatsing's equity incentive plans require provincial state-asset approval; for example, the 2023 restricted stock plan was formally filed with and approved by the Sichuan SASAC in May 2023.³ Strategic expansion and capital-raising decisions move through the governance channels of a provincial state energy enterprise rather than an academic institution. The designation of Hwatsing as a "Tsinghua spinoff" reflects its technical lineage rather than its ultimate governance control.

Incentives, and what the targets reveal

The company's equity incentive programs provide insight into internal growth expectations. Under the 2023 restricted stock plan, 3,803,579 shares were granted. However, by the time of the 2025 disclosures, only 936,405 shares had met vesting conditions and completed registration.²⁸ This low conversion rate stems from performance targets established prior to the recent revenue acceleration, alongside share lapses caused by staff turnover and unmet annual performance criteria.

The 2026 incentive plan, revised in July 2026, sets cumulative revenue growth hurdles against the 2024 revenue baseline of ¥3.406 billion: requiring at least 58.53% growth in 2026, 86.42% in 2027, and 117.25% in 2028.²⁹³⁰ These percentages translate into implied annual revenue targets of approximately ¥5.40 billion for 2026, ¥6.35 billion for 2027, and ¥7.40 billion for 2028.

Comparing these hurdles to current operational results highlights a notable gap. Having generated ¥4.648 billion in revenue in 2025 and ¥2.643 billion in the first half of 2026, Hwatsing is tracking well ahead of the ¥5.40 billion full-year target for 2026.¹³ The target for 2026 implies a year-over-year revenue expansion of roughly 16% over 2025—a deceleration for a company that registered annual top-line growth above 35% over the previous three reporting periods. Furthermore, the targets for 2027 and 2028 imply subsequent annual growth rates of approximately 18% and 17%, respectively.

These performance targets can be interpreted in two ways. A supportive interpretation suggests management has set achievable baseline metrics to ensure stability during potential downturns in the semiconductor equipment cycle. A more cautious interpretation suggests that setting hurdles at roughly half the company's recent growth rate indicates management expects top-line expansion to cool from its previous 35%-plus trajectory toward a high-teens annual rate.

Credibility, weighed

Operational execution demonstrates several consistent strengths. Management messaging across 2025 and 2026 investor communications has maintained a coherent narrative regarding its platform strategy, product expansion roadmap, and gross margin headwinds. Rather than attributing margin contraction to customer pressure or macroeconomic conditions, executive commentary has consistently linked lower margins to early-stage product mix and expanded operational investments. Product disclosures in public filings remain detailed and verifiable, citing specific tool models, delivery milestones, and shipment volumes. Technical reinvestment is visible on the payroll: research and development headcount expanded from 634 to 876 personnel within a year—representing roughly 33% of the 2,654-person total workforce—with an average annual R&D compensation of ¥437,000.¹⁷³¹ Cumulative granted patents reached 569, including 343 invention patents.¹⁷ Additionally, in June 2024, the CMP technology initiative co-developed with Tsinghua University received the 2023 National Technology Invention Award (First Prize), with Lu Xinchun recognized as the primary contributor.³²³³

At the same time, several corporate governance factors warrant monitoring. These include the cancellation of the Hong Kong listing within eight months of its announcement, a tendency during investor calls to address margin inquiries with broad statements on market positioning rather than detailed unit economics, and the 2025 reporting reclassification that separated revenues into "semiconductor equipment" and "semiconductor services," which improved platform visibility but reduced direct comparability with prior product-level data.¹⁷ Furthermore, during the April 2026 earnings call, when asked by an investor why the company had not provided video-livestreamed earnings meetings with public replays in 2024 or 2026, management responded that it would "seriously study and consider" the recommendation.²

↑ Back to Top

IX. Playbook: Business & Investing Lessons

There is a photograph-shaped moment buried in Hwatsing's filing history that captures its entire commercial arc. In its 2021 prospectus, the company included a section heading that most issuers would have fought their bankers to delete: the company has a considerable gap versus industry leaders in products, technology, and market share.⁷ Four and a half years later, its chairman was fielding a question on an earnings call about whether three consecutive years of margin compression meant Hwatsing lacked bargaining power.² Between those two disclosures lies a twelvefold expansion in revenue, a national technology award, and a cumulative installed base exceeding one thousand tools. Yet the operational answer to both questions remains closer to "yes" than the top-line growth curve suggests.

That gap between narrative velocity and underlying economic reality yields several transferable lessons for analyzing capital equipment suppliers.

The academic spin-off's primary bottleneck is yield engineering, not fundamental physics. Lu Xinchun's Tsinghua research group mastered nanoscale planarization years before Hwatsing manufactured a tool that fab operators would risk on production wafers. The bridge from laboratory prototype to commercial acceptance required years of stationed field engineers, absorbed qualification costs, and operating losses—funded by non-dilutive state grants rather than commercial cash flows. In hard-technology sectors, investors must evaluate a company's field deployment capability rather than its patent catalog. A patent proves an underlying technical idea; an acceptance certificate proves a customer risked active production wafers on the hardware.

Revenue recognition rules reveal where operational risk accumulates. In businesses where revenue recognition depends on formal customer acceptance rather than tool shipment, the balance sheet leads the income statement. Dispatched equipment sitting in fab inventory awaiting qualification represents a key forward indicator: it will become either future revenue or a future write-down, governed by customer timelines outside the supplier's control. Investors should monitor inventory composition and cash conversion as closely as reported gross margins, because accounting margins are often the last metric to reflect operational delays.

Policy tailwinds compress adoption timelines without creating structural moats. The domestic substitution drive between 2018 and 2022 achieved what internal engineering alone could not: it compelled fab operators to bear qualification risk and reallocate scarce validation slots to domestic tools. However, that policy impulse temporarily masked the underlying economic leverage of selling to a concentrated group of buyers. As initial substitution mandates mature, those client dynamics are re-emerging in gross margin contraction. When policy tailwinds subside, commercial defensibility depends on the market power established while the tailwind blew—and Hwatsing built a durable volume franchise rather than a pricing-power franchise.

Monopsony leverage offsets monopoly market share. Hwatsing commands more than 90% of the domestic CMP tool market, yet its five largest customers generate 64% of total revenue. In this structural configuration, buyers retain pricing leverage regardless of the supplier's domestic market share. A supplier cannot walk away from a customer accounting for more than a quarter of its revenue, whereas a concentrated buyer can qualify a secondary supplier over an eighteen-month validation window. Dominant market share does not translate into pricing power when counterparties are equally concentrated.

Platform expansion obeys historical conversion rates. Hwatsing required roughly seven years to achieve customer acceptance for its first 58 CMP tools. Four years after initial investments in ion implantation, that segment contributes roughly 2.5% of total revenue, whereas wafer thinning tools have progressed faster along the adoption curve. Underwriting platform diversification requires applying a company's historical qualification timelines rather than management's projected target dates. Transaction earn-out structures negotiated by insiders—such as customer-acceptance revenue hurdles—offer a more reliable indicator of commercialization timelines than general investor presentations.

Equity incentive targets reflect management's baseline expectations. Hwatsing's 2026 restricted stock plan sets cumulative revenue hurdles implying annual top-line growth in the high-teens through 2028, even as the company delivered top-line expansion above 35% over the preceding three reporting periods. Board compensation committees rarely establish vesting hurdles they expect an enterprise to miss. When internal performance targets diverge significantly from historical growth rates, the incentive structure typically offers the more realistic guide to long-term operational expectations.

State-controlled champions operate with distinct governance trade-offs. Ownership under a provincial SASAC provides capital access, policy alignment, and preferential standing in domestic procurement workflows. Conversely, it introduces multi-tiered approval processes for equity compensation—such as formal state-asset filings—a conservative stance toward dividend payouts, and corporate objectives oriented around strategic industrial capability rather than per-share capital return. These governance characteristics do not invalidate the investment thesis, but they represent structural factors that belong in the valuation discount rate.

---

↑ Back to Top

X. Analysis: 7 Powers, 5 Forces & Bear vs. Bull Thesis

Hamilton Helmer's 7 Powers

Process Power — strong, and the best-evidenced of the seven. Twenty-plus years of accumulated know-how in multi-zone polishing head pressure control, endpoint detection, and slurry fluid dynamics are recognized at the highest national technical level. More importantly, this operational expertise is validated by more than 1,000 tools running in volume production across logic, 3D NAND, DRAM, silicon wafer, compound semiconductor, CIS, MEMS, MicroLED, and advanced packaging lines.³ Process power remains the hardest of the seven powers to replicate because it resides in tacit engineering expertise and operational procedures rather than published documentation.

Switching Costs — high, and the primary moat. The 12-to-18-month first-tool customer acceptance cycle and the technical reality that process recipes are developed for specific tool architectures make equipment displacement prohibitively expensive. This dynamic is demonstrated behaviorally through repeat and bulk repeat orders across CMP, wafer thinning, edge polishing, and ion implantation product lines.³

Cornered Resource — moderate and weakening as a differentiator. The Tsinghua University licensing relationship was historically decisive. However, holding 569 patents and national technical awards does not prevent a well-funded domestic entrant from developing competing CMP tools; it merely elevates the initial entry cost. Given the state-directed nature of Chinese semiconductor funding, technical capability gaps are eventually financed and closed.

Scale Economies — moderate. Spreading R&D expenses—which accounted for 11.69% of 2025 revenue—across a dominant domestic volume base provides theoretical operating leverage.¹ In practice, operating leverage has not yet materialized: expense ratios expanded in 2025, while the decline in R&D intensity to 10.49% of revenue in the first half of 2026 resulted solely from top-line expansion outpacing R&D spending rather than expense moderation.³

Counter-Positioning, Branding, and Network Economies — absent. Foreign incumbents face no structural barriers preventing them from competing; capital equipment lacks consumer branding leverage; and network effects do not apply to semiconductor hardware manufacturing.

Porter's Five Forces

Buyer power — very high, and the binding constraint. The top five customers generated 64.07% of 2025 revenue, with the largest client alone contributing 27.42%.¹⁷ These buyers are state-backed national champions running multi-vendor qualification programs as explicit supply-chain policy, granting them substantial leverage over equipment pricing and commercial terms.

Supplier power — moderate. The top five component suppliers accounted for only 17.34% of total 2025 procurement, reflecting a well-diversified domestic supply base.¹⁷ The primary risk stems not from commercial supplier concentration, but from potential geopolitical restrictions on foreign high-precision sub-components inside the tools.

Threat of new entrants — low domestically, but not zero. The 12-to-18-month qualification barrier serves as a formidable operational hurdle. While this defense delays entry, it does not permanently prevent it, particularly given patient state funding for domestic alternatives.

Threat of substitutes — very low. No alternative physical process exists to achieve atomic-level wafer planarization at microchip dimensions, making this force durably favorable.

Rivalry — bifurcated. Within mainland China's core CMP market, rivalry features low market-share competition but high pricing pressure. In adjacent product categories, competition is intense against established domestic specialists: Naura Technology in etch, PVD, and CVD; ACM Research Shanghai in wet cleaning; Piotech in PECVD and ALD; AMEC in etch; and Kingstone Semiconductor in ion implantation. Outside mainland China, Hwatsing remains virtually absent—generating 99.08% of 2025 revenue domestically—constrained by export controls and incumbent vendor relationships.¹¹

Risk radar

Customer capex cyclicality. Management's interim report identifies customer capital expenditure cycles as the primary operational risk: during industry downturns, chipmakers cut equipment budgets sharply, directly depressing tool orders.³ With five buyers generating nearly two-thirds of sales, an investment pause by a single major customer represents a material impact to revenue.

Revenue recognition and working capital. Because revenue is recognized only upon formal customer acceptance, tools shipped but undergoing validation sit on the balance sheet in inventory as dispatched goods. This line item reached ¥1.947 billion at year-end 2025, accounting for roughly half of total inventory of ¥3.989 billion.¹⁷ Accounts receivable expanded 45.18% to ¥963 million in 2025 and reached ¥1.457 billion by mid-2026, representing 55.13% of first-half revenue.¹⁷³⁴ Consequently, net operating cash flow dropped 30.73% in 2025 to ¥799 million, then fell 81.44% year-over-year in the first half of 2026 to ¥73 million.¹⁷³ Management attributed the first-half cash flow contraction to scheduled procurement payments and staff compensation outlays.³ Across two consecutive reporting periods, the divergence between accounting profit and cash generation has widened, creating operational exposure if customer acceptance timelines stretch during an industry slowdown.

Supply chain controls. Expanded export restrictions on high-precision sub-components—such as specialized optics, motors, and pumps—present direct operational risks to assembly. The former general manager's 2024 assertion that CMP tool components have achieved substantial self-sufficiency has not been independently verified.⁹

Margin erosion. As detailed throughout the financial analysis, structural gross margin contraction reflects buyer leverage intersecting with an expanding product mix of lower-margin, first-generation tools undergoing initial customer qualification.

The Bull Case

Hwatsing stands as the sole volume domestic manufacturer of a non-substitutable process step in a market committed to semiconductor self-sufficiency, where the foreign-supplied share of CMP tools has already dropped to 63%.¹⁰ High customer switching costs are validated by repeat tool purchases. Technical demand drivers—including rising 3D NAND deck counts, High Bandwidth Memory stacking, and hybrid bonding in advanced packaging—expand CMP steps per wafer faster than total wafer volumes grow, with advanced tools already selected as baseline equipment for a major memory producer's HBM lines.² The wafer thinning segment provides a viable secondary growth driver, supported by over 20 units shipped, repeat customer orders, and direct alignment with advanced packaging demand. Concurrently, high-margin wafer reclaim operations are expanding capacity toward 600,000 wafers per month. Order backlogs remain strong, with advanced-node and advanced-packaging tools representing an increasing proportion of the mix.² Furthermore, internal performance targets under the equity incentive plan reflect management's baseline confidence in sustaining high-teens annual revenue growth through 2028.

The Bear Case

Three consecutive years of gross margin contraction alongside 35%-plus top-line expansion indicate that concentrated domestic buyers retain pricing power. Net profit growth slowed to 5.89% in 2025 and 11.44% in the first half of 2026—lagging revenue growth in both periods. While the 2025 headline profit growth was suppressed by a high non-recurring income baseline in 2024, the underlying core operational expansion of 12.69% remains modest relative to sales growth.² Operating cash flows have deteriorated for two consecutive periods as inventory and receivables outpaced top-line expansion. The ion implantation business generates only 2.5% of total revenue four years after initial investment and faces a steep hurdle to meet its performance earn-out targets. Customer concentration leaves earnings vulnerable to capital spending cuts by a few major chipmakers. Finally, having raised ¥3.49 billion in net IPO proceeds while proposing an additional ¥4.0 billion private placement, cumulative dividend distributions of ¥270 million demonstrate a capital-allocation model that prioritizes equity dilution over shareholder return.

The activist's question

An institutional investor evaluating Hwatsing's capital efficiency would raise three central issues with executive management.

First, the company should disclose equipment gross margins split between first-generation qualification tools and mature repeat-order systems, allowing shareholders to verify whether margin compression stems from early-stage product introduction rather than pricing concessions.

Second, management should clarify the operational factors that prompted the cancellation of the proposed Hong Kong listing within eight months of its announcement, detailing whether the reversal reflected shifting capital market conditions or internal strategy shifts.

Third, given that Hwatsing has raised ¥3.49 billion in net IPO capital and proposed an additional ¥4.0 billion private placement against ¥270 million in total dividend distributions, management should state the hurdle rate required for the proposed capital placement and establish formal metrics to track subsequent returns on equity.

Clear answers to these questions would provide essential transparency regarding Hwatsing's long-term capital allocation and economic moat.

↑ Back to Top

XI. Key Metrics to Watch & Outro

Three numbers capture whether Hwatsing is compounding economic value or simply scaling operational volume.

1. Semiconductor equipment segment gross margin. The critical indicator is not the consolidated figure, but the equipment line specifically, which Hwatsing discloses annually and which contracted to 40.88% in 2025.¹⁷ Blended gross margins are increasingly buoyed by an expanding, higher-margin services business, masking baseline hardware pricing dynamics. If equipment margins stabilize, management's explanation—that margin dilution reflects initial qualification mix—will be validated, indicating that Hwatsing's competitive moat retains pricing power. If equipment margins continue to decline while top-line revenue expands at 30%-plus, the monopsony thesis holds, confirming a volume-driven enterprise operating with structurally capped economic returns.

2. Ion implantation revenue against the ¥340 million 2026 earn-out. This hurdle provides the clearest benchmark for evaluating whether platform expansion is progressing on schedule. Because insider management established the performance threshold, and because the 80% customer-acceptance clause prevents meeting targets through consumables or service fees, Xinyu Semiconductor's audited revenue delivery represents the most informative operational metric in the upcoming 2026 annual report.

3. Operating cash flow relative to net profit. Following two consecutive reporting periods of cash flow contraction—with dispatched tools awaiting customer acceptance accounting for nearly half of inventory, and accounts receivable reaching 55% of first-half revenue—earnings quality represents a core operational vulnerability. For a capital equipment supplier, uncollected accounting profit reflects pending customer acceptances, an asset category that becomes increasingly vulnerable during industry downturns.

A secondary metric warranting tracking is top-five customer concentration, disclosed annually in financial filings. Concentration fell from 85.71% in 2020 to 64.07% in 2025—a positive trend line, as every incremental reduction weakens buyer monopsony leverage.⁷¹⁷ However, this metric ranks below the top three indicators because the decline is partly arithmetic: a twelvefold top-line expansion naturally expands the revenue denominator. The more critical operational test is whether the largest single customer's share, which stood at 27.42% in 2025, declines, as that specific relationship carries the greatest pricing leverage over Hwatsing.¹⁷

Conversely, several widely cited metrics provide limited diagnostic value for investors: reported top-line revenue growth, headline blended gross margin, and cumulative tool shipments. Revenue expansion has hovered near 35% for three consecutive years, yielding little incremental insight into underlying unit economics. Blended gross margin reflects product-mix shifts—driven by the expanding services segment, which yields roughly seven percentage points higher gross margin than equipment—rather than core hardware pricing power. Finally, cumulative shipments exceeding 1,000 tools reflect past adoption rather than the marginal profitability of future units.

Beyond financial metrics, executive commentary at future investor briefings will signal management's underlying visibility. Because Hwatsing does not break out segment-level equipment margins in semi-annual filings—the 2026 interim report discloses only consolidated revenue and cost of sales, requiring investors to infer blended margins—disclosure depth serves as a qualitative indicator.³ Should management detail a segment-level margin bridge or differentiate between initial qualification tools and repeat orders, it would signal confidence in its product-mix thesis. Conversely, repeating general commentary—such as citing intact competitive positioning, unchanged demand trends, and ongoing cost controls—would suggest reluctance to provide quantitative backing for margin claims.²

---

Hwatsing represents a distinct industrial case study: an enterprise that translated academic tribology research into the sole domestic volume supplier for an indispensable semiconductor manufacturing step. That technical execution is established.

The primary valuation question is the economic durability of that market position. Hwatsing sits at the intersection of two opposing structural forces. On one side, national policy mandates and fundamental technology shifts—such as 3D NAND deck scaling, High Bandwidth Memory, and hybrid bonding—mechanically expand demand for wafer planarization tools. On the other side, a concentrated customer base of powerful domestic chipmakers exerts substantial commercial leverage over pricing, while platform expansion into adjacent tool markets has progressed faster than segment profitability.

Recent financial results illustrate this tension: top-line revenue expansion reflects policy and technical tailwinds, while contracting gross margins reflect buyer concentration and execution costs. The equipment segment gross margin in the 2026 annual report will provide the critical empirical test of which force ultimately governs Hwatsing's long-term returns.

---

References

1. 华海清科(688120.SH)：2025年年报净利润为10.84亿元、同比较去年同期上涨5.89% — 新浪财经, 2026-04-23 ↩↩↩↩↩↩↩↩↩

2. 华海清科股份有限公司投资者关系活动记录表 编号：2026-01（2025年年度暨2026年第一季度业绩说明会） — 华海清科, 2026-04-24 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

3. 华海清科股份有限公司2026年半年度报告 — 上海证券交易所披露, 2026-08-22 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

4. 华海清科股份有限公司投资者关系活动记录表 编号：2025-01（2024年度业绩说明会） — 华海清科, 2025-04-30 ↩↩↩↩

5. 国产CMP设备龙头厂商——华海清科介绍 — 搜狐 ↩↩↩

6. 高端装备界面科学与技术全国重点实验室路新春教授和雒建斌院士团队项目获国家技术发明奖一等奖 — 清华大学摩擦学国家重点实验室 ↩

7. 华海清科股份有限公司首次公开发行股票并在科创板上市招股说明书 — 上海证券交易所, 2021-07-01 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

8. 清华实控、国内唯一，12英寸CMP设备商科创板上会通过 — 腾讯新闻, 2021-06-17 ↩↩

9. 华海清科张国铭：近五年营收复合增长率86%，CMP零部件已基本实现自主可控 — 央广网, 2024-07-25 ↩↩

10. 我国化学机械抛光（CMP）设备行业市场规模高速增长 国产化率逐步提升 — 观研报告网 ↩↩

11. 华海清科Q1净利增速大幅跑输营收 终止港股上市计划转向A股定增 — 财联社, 2026-04-23 ↩↩↩↩

12. 华海清科上市4年拟不超40亿定增 IPO募36亿分红共2.7亿 — 中金在线, 2026-04-23 ↩↩↩

13. 半导体设备商华海清科上市首日大涨64% 预计上半年净利润过亿 — 财新网, 2022-06-08 ↩

14. 清华系再出半导体 IPO，华海清科开盘猛涨 72% — IT之家, 2022-06-08 ↩

15. 华海清科股份有限公司关于收购芯嵛半导体（上海）有限公司剩余股权的公告 — 上海证券交易所, 2024-12-25 ↩

16. 10亿收购上海芯嵛82%股权 华海清科加大离子注入机业务 — 新浪财经, 2024-12-25 ↩↩↩

17. 华海清科股份有限公司2025年年度报告 — 上海证券交易所披露, 2026-04-23 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

18. 华海清科股份有限公司关于筹划公司在香港联合交易所有限公司上市的提示性公告 — 巨潮资讯, 2025-08-29 ↩

19. 华海清科：终止筹划港股上市改为定增A股 — 新浪财经, 2026-04-22 ↩

20. 华海清科终止筹划H股上市、转道A股募资，首季营收增长32% — 腾讯新闻, 2026-04-23 ↩

21. CMP in Advanced Nodes: Challenges at 7nm, 5nm & Beyond — JEES ↩

22. Advanced Packaging CMP: Slurry Requirements for 3D NAND & TSV Processes — JEES ↩

23. 华海清科股份有限公司关于变更公司董事、高级管理人员的公告 — 上海证券报, 2025-04-02 ↩↩

24. 华海清科启动港股IPO前换帅，42岁清华博士王同庆升任董事长 — 腾讯新闻, 2025-09-01 ↩↩

25. 华海清科高管平均年薪292.31万：首席科学家路新春年薪660.10万最高 — 新浪财经, 2026-04-23 ↩

26. 华海清科：实控人变更为四川省国资委 — 财联社 ↩

27. 华海清科间接控股股东将变更为四川能源发展集团 — 腾讯新闻, 2025-02-27 ↩

28. 华海清科股份有限公司关于2023年限制性股票激励计划相关事项的公告（公告编号：2025-070） — 华海清科, 2025-10-31 ↩

29. 华海清科股份有限公司关于2026年限制性股票激励计划（草案）及相关文件的修订说明 — 上海证券报, 2026-07-11 ↩

30. 华海清科股份有限公司2026年限制性股票激励计划（草案）摘要公告 — 上海证券交易所披露, 2026-04-16 ↩

31. 东海研究 | 深度：华海清科（688120）：先进CMP设备放量突破，平台化战略引领协同发展 — 新浪财经, 2026-05-12 ↩

32. 华海清科：荣获国家技术发明奖一等奖 — 中证网, 2024-06-25 ↩

33. 2023年度国家技术发明奖获奖项目目录（通用项目） — 中华人民共和国科学技术部, 2024-06-20 ↩

34. 华海清科2026年上半年营收净利双增 应收账款占比超五成 — 新浪财经, 2026-08-21  ↩