Skyverse Technology: The Referees of China's Semiconductor Chipmakers

I. Introduction & Episode Roadmap [00:00 - 08:00]

There is a room in every advanced semiconductor fabrication plant that visitors almost never see. It is not the lithography bay, where a machine the size of a bus prints circuit patterns finer than a virus. It is not the etch module, where plasma carves trenches a few dozen atoms wide. It is a quieter room, full of tools that make nothing at all. They only look.

They look at wafers — 300-millimetre discs of polished silicon — and answer two questions over and over, thousands of times a day: Is there anything on this wafer that should not be there? And is everything that should be there exactly the right size?

Those two questions decide whether a fab makes money.

This is the unusual economic reality at the centre of the semiconductor industry. High-profile capital expenditure flows primarily to machines that create: ASML's lithography scanners, etchers from Lam Research and 中微公司 AMEC, and deposition chambers from Applied Materials. But a fab's profit and loss statement is not governed solely by processing speed. It is governed by yield — the percentage of functional chips per wafer. Yield is ultimately governed not by the tools that create, but by the tools that inspect.

Industry executives refer to this category as "process control," "quality control equipment," or in Chinese, 量检测 inspection and metrology. Its financial impact stems from manufacturing arithmetic. A leading-edge logic process involves close to a thousand sequential steps.² Every step must execute almost flawlessly, because losses compound.

When a process runs more than 500 steps, each individual step must achieve a 99.99% success rate for the finished wafer to yield above 95%. If that per-step figure slips to 99.98% — a difference that sounds like a rounding error — final yield collapses to roughly 90%.² A tiny fractional drop in step-level accuracy doubles total yield loss.

That is why fabrication plants invest heavily in tools designed solely to catch defects early. It also explains how American supplier KLA Corporation built one of the most profitable franchises in capital equipment, operating within a global inspection-and-metrology market that reached roughly US$19.1 billion in 2025, where the top five foreign vendors control more than 70%.²

Against this entrenched backdrop stands 深圳中科飞测科技股份有限公司 Skyverse Technology Co., Ltd. (ticker 688361 on the 上海证券交易所科创板 Shanghai Stock Exchange STAR Market). Established in December 2014 as a spin-out from the 中国科学院微电子研究所 Institute of Microelectronics of the Chinese Academy of Sciences (IMECAS), the company was founded by a physicist who spent five years in KLA's research and development organisation.¹⁵¹⁷ By 2026, Skyverse had emerged as China's broadest domestic supplier of front-end optical yield-control tools, shipping over 400 systems in 2025 across thirteen product families.⁶

Yet the company is currently generating net losses at an unprecedented rate for its public history.

This financial tension forms the core of the Skyverse story. In the first half of 2026, Skyverse expanded revenue by 34.01% to RMB 941.02 million while posting a net loss attributable to shareholders of RMB 127.44 million — its largest half-year net loss since listing.¹ Management attributes the loss to aggressive growth investments: expanded research and development projects, headcount growth, and introductory pricing on initial tool evaluations at customer fabs. For long-term investors, the critical question is whether these expenditures secure a durable competitive position or highlight the steep ongoing costs required to bridge the remaining technology gap.

The story unfolds across several key areas. First, it examines the physics and economics of yield control and why process-control tools exert outsized financial leverage. Next, it traces the company's origins: a founder from the 中国科学技术大学 University of Science and Technology of China's Special Class for the Gifted Young, a decade of experience across American metrology firms, and a return to China that led to a Shenzhen startup. The analysis then evaluates segment economics, identifying where current revenues originate and distinguishing commercially proven product lines from those undergoing evaluation.

It then addresses capital allocation, evaluating two equity raises totaling RMB 4.388 billion and an organic expansion strategy that contrasts sharply with the acquisition-led growth of global incumbents.⁹ Subsequent sections test Skyverse's competitive moat, customer relationships, and reliance on state subsidies, followed by a competitive breakdown, key risks, bull and bear cases, and the core metrics tracking execution.

The central thesis to test is clear: Skyverse has established a rare domestic position in semiconductor metrology, but market valuations may assume its primary hurdles are cleared. The evidence indicates the most challenging phase of technical scaling and margin recovery lies ahead.

---

↑ Back to Top

II. The Physics & Economics of Semiconductor Yield: Why Metrology Matters [08:00 - 22:00]

Picture a silicon wafer as a sprawling city seen from an aeroplane at night. Every building is a transistor, and there are tens of billions of them on a single disc. Now imagine the job is to fly over that city and locate every cracked window in under an hour without landing or missing one. That is inspection.

Now imagine a second job: not finding broken structures, but measuring them. How thick is the roof on each building? Are the walls separated by exactly 14 nanometres or 14.3? Is the second floor aligned precisely on top of the first, or is it offset by two nanometres? That is metrology.

The semiconductor industry treats these as distinct disciplines because they rely on different optical physics and address different manufacturing challenges. 检测 Inspection hunts for non-deterministic defects — unwanted particles, scratches, unintended short circuits between isolated lines, and voids in metal fills — scanning the wafer because their locations cannot be predicted. 量测 Metrology measures expected physical properties: film thickness, critical dimensions, etch depth, surface topography, and alignment between successive lithography layers.²

Within inspection, a clear technical hierarchy maps directly onto equipment pricing and profit margins. Unpatterned wafer inspection serves as the entry level: tools inspect blank, mirror-smooth silicon discs for surface contamination before circuit printing begins, monitoring chamber cleanliness across the fab line.² While technically demanding, the uniform surface makes signal processing manageable.

Patterned wafer inspection is significantly more complex because tools must distinguish minute, unintended anomalies from intricate circuit patterns. Here, the technology divides by illumination method. Darkfield inspection sweeps the wafer with a deep-ultraviolet laser at an angle and collects light scattering off surface irregularities — similar to shining a torch across a dusty table where particles illuminate while the polished surface remains dark. It offers high throughput and exceptional sensitivity to small surface particles.²

Brightfield inspection instead floods the wafer with broadband light, directly imaging reflections and comparing adjacent die patterns. Although optically complex and slower, brightfield systems detect low-contrast and buried defects in deeper wafer layers that darkfield tools miss.

Brightfield represents the highest-value category in process control. Industry data from VLSI Research, cited in Skyverse's 2025 annual report, shows brightfield patterned-wafer inspection generated US$2.79 billion globally in 2024 — accounting for 19.4% of all inspection and metrology spending, the largest single product category. Darkfield generated US$1.05 billion (7.3%), while unpatterned inspection reached US$1.44 billion. On the metrology side, optical critical dimension tools accounted for US$1.30 billion, and overlay metrology tools reached US$0.91 billion.²

This distribution highlights a key aspect of Skyverse's commercial positioning: the industry's largest single market segment, brightfield patterned inspection, is a category where Skyverse has yet to establish high-volume commercial shipments. Subsequent sections examine this product gap in detail.

Metrology features its own critical sub-segment: overlay measurement. Overlay tools determine whether successive lithography layers are correctly aligned. Modern integrated circuits are built like multi-layer structures, placing transistors at the base, contacts above them, and up to fifteen layers of metal wiring on top. If layer nine shifts relative to layer eight by a few nanometres, vertical interconnect vias miss their contact pads, rendering the chip non-functional.

Overlay systems measure misalignment at sample sites across the wafer and feed real-time correction data back into lithography scanners before subsequent batches are exposed.² Operating inside this direct feedback loop makes semiconductor fabs cautious when qualifying new vendors. Consequently, Skyverse's rollout of third-generation overlay tools across major domestic chipmakers represents a significant commercial milestone for the firm.⁵

Two structural shifts in chip design have increased measurement complexity and expanded tool intensity per wafer. First, memory architecture has shifted vertically. 3D NAND flash piles hundreds of physical layers, requiring high-aspect-ratio channel etches so deep that standard optical signals cannot reach the bottom. Skyverse developed a specialized X-ray measurement tool to address these deep-structure challenges, which is currently operating on a domestic advanced-memory production line.⁶

Second, advanced packaging techniques for high-bandwidth memory (HBM) rely on bonding stacked memory dies using copper-filled through-silicon vias. Verifying that these internal copper fills are free of voids requires penetrating solid silicon, shifting measurement requirements from light-based systems to X-ray inspection.² This expanding array of physical measurement challenges rewards equipment vendors with broad tool portfolios while increasing capital requirements for smaller firms.

Equipment options also vary by underlying detection technology. Optical instruments use light and dominate high-throughput production lines. Electron-beam systems scan focused electron beams to achieve higher resolution, though at significantly lower speeds — with optical tools offering more than a thousand times the throughput of e-beam systems.² X-ray tools provide non-destructive subsurface inspection for features inaccessible to optical instruments. In 2025, optical tools represented 76.6% of global inspection and metrology revenues, e-beam accounted for 17.7%, and X-ray tools held 3.2%.² Skyverse remains primarily an optical equipment company while expanding into e-beam and X-ray product lines.

The financial leverage of process control stems from manufacturing lead times. Advanced logic wafers require approximately three months to move through complete fabrication cycles. If a deposition module drifts out of specification on day four without detection, three months of processing work across subsequent wafer lots can be lost. Yield-control tools compress this diagnostic feedback loop from months to hours, sustaining equipment demand even during industry downturns.

This equipment category also carries high customer switching costs. Inspection tools rely on proprietary algorithms that process engineers calibrate over time. Fabs develop customized recipe parameters tailored to a specific tool's optical properties, background noise levels, and defect classification criteria, building historical databases aligned with that machine. Replacing a tool invalidates these historical baselines.

Skyverse management emphasized this dynamic in investor disclosures, noting that high qualification costs, line management overhead, and historical defect databases mean customers "typically will not choose too many suppliers on the same production line," especially at advanced process nodes.⁶ This dynamic protects incumbent vendors where established toolsets exist, presenting an entry barrier for new suppliers.

Skyverse's disclosures highlight a market "highly monopolised by overseas equipment companies," with KLA maintaining the leading market share in China.² Industry estimates from mid-2026 indicate KLA held approximately 58% of the global market, while China's domestic localization rate in process control equipment stood at roughly 16% as of September 2025.¹³ While precise figures vary, the overall market structure is clear: after extensive domestic import-substitution efforts, more than four out of five yield-control tools in Chinese fabs continue to be supplied by foreign vendors.

Technical barriers in this sector include sub-nanometre optical resolution, ultra-stable deep-ultraviolet light sources rated for continuous operation, real-time image processing architectures handling high data throughput, and libraries of defect classification recipes built over decades. While hardware components and processing pipelines can be engineered, classification libraries require extensive operational access to active production lines. Expanding this customer footprint remains central to Skyverse's long-term strategy.

This challenge leads directly to the background of the founder who set out to build China's domestic metrology champion.

↑ Back to Top

III. Origins & Key Inflection Points (2014–2023): From CAS Lab to STAR Market IPO [22:00 - 38:00]

In 1992, 陈鲁 Chen Lu, a student from Tianjin's Yaohua Middle School, passed the entrance examination for the 少年班 junior class at the University of Science and Technology of China — a selective programme established in 1978 to admit gifted teenagers into university ahead of the standard academic track.¹⁷ After completing his undergraduate degree in physics, Chen moved to the United States at age twenty to attend Brown University, where he earned a doctorate in physics in 2003.¹⁵¹⁷

Following his doctoral studies, Chen chose semiconductor process control over academia or chip design. From 2003 to 2005, he worked as a systems scientist at Rudolph Technologies, a specialist American yield-control firm. He then spent five years, from 2005 to 2010, as a senior scientist at KLA, focusing on next-generation optical inspection systems for integrated circuits.¹⁵¹⁷

This five-year tenure inside KLA's research and development organisation provided Chen with direct experience in the core optical architectures that dominate modern semiconductor yield control, establishing the technical foundation for Skyverse's future product development.

In March 2010, Chen and his wife 哈承姝 Ha Chengshu returned to China after more than a decade in the United States.¹⁷ Chen joined IMECAS as a researcher under the Chinese Academy of Sciences' 百人计划 Hundred Talents Programme, serving as a doctoral supervisor until August 2016.¹⁵¹⁷ Ha brought complementary corporate and financial expertise, holding a doctorate in law from an American university, prior experience as a senior tax analyst at Deloitte in the U.S., and five years in venture capital at 金沙江创投 GSR Ventures before joining Skyverse full-time in 2016.¹⁷ This combination created a distinct division of leadership between technical architecture and corporate finance.

On 31 December 2014, the founders established Skyverse in Shenzhen, with IMECAS participating as an institutional co-founder.¹⁷ Shenzhen offered established supply chains, skilled precision-manufacturing labour, and municipal backing for semiconductor equipment. At the time, Chinese domestic self-sufficiency in front-end optical process control was near zero, and the sector attracted minimal private capital-market interest.¹⁷ Skyverse entered a market dominated by established foreign incumbents, led by KLA, which held a multi-decade technological lead.

Skyverse reached its first major operational inflection point by 2016, when its initial defect-inspection tool entered customer validation.¹⁷ Vendor evaluation in front-end semiconductor manufacturing involves significant operational risk, as chipmakers rarely qualify equipment based on technical specifications alone.

Instead, fabs must allocate cleanroom floor space, engineering capacity, and active production wafers to test a tool over several months against established benchmark systems. As Skyverse management noted, customers exercise extreme caution during tool qualification because the process consumes substantial internal research, process engineering, and fab production resources.⁵ Consequently, qualification cycles routinely extend over several years, requiring substantial capital reserves to sustain operations.

Skyverse successfully navigated these early qualifications, converting initial trials into repeat orders between 2017 and 2020 for unpatterned wafer defect inspection and optical film-thickness metrology tools across major domestic fabrication lines. While Skyverse does not explicitly name customer accounts in regulatory filings — stating only that its top five customers are leading domestic front-end chipmakers and advanced-packaging manufacturers⁵ — industry analysts widely identify 中芯国际 SMIC and 长江存储 YMTC as core customer accounts.¹³

This initial phase imposed severe financial demands. A single evaluation unit costing millions of yuan in components and extensive engineering labor must be delivered, installed, and supported on-site for a year or more without revenue recognition. During validation, the vendor bears full field engineering costs and absorbs requested hardware modifications. If a system fails qualification, the investment is written off entirely.

Even a successful evaluation yields only an initial single-unit purchase order, often negotiated at low initial prices to establish a customer reference. Scaling this model across multiple tool families requires years of continuous funding before achieving commercial self-sustainability — a structural barrier that explains why process-control startups depend heavily on strategic venture funding and government support during early development.

To sustain operations through this extended validation cycle, Skyverse assembled a cap table backed by major state-affiliated and industry-aligned investors, including 深圳市创新投资集团 Shenzhen Capital Group, 哈勃投资 Hubble Technology Investment (Huawei's chip-focused investment arm), 中芯聚源 SMIC-Fortune, and 国投集团 SDIC.⁷ Beyond providing growth capital, strategic investors like Hubble played a pivotal role after 2019 by facilitating customer access across domestic semiconductor supply chains.

Skyverse reached its second major milestone on 19 May 2023 with its initial public offering on the STAR Market. The company issued 80 million new shares priced at RMB 23.60 per share, raising RMB 1.888 billion in gross proceeds against an initial RMB 1.0 billion target, yielding RMB 1.698 billion in net capital after underwriting fees.⁷⁹ Investor demand drove a strong trading debut: shares opened at RMB 66.00 and closed up 189.62% on day one, giving Skyverse a market capitalization of approximately RMB 21.87 billion.⁷

This initial market response reflected investor focus on import substitution rather than near-term profitability. In 2022, the year preceding its listing, Skyverse generated RMB 509 million in revenue and RMB 11.74 million in net profit.⁷ Public market valuation was driven primarily by Skyverse's strategic position as China's leading domestic player in process control — a market dynamic that has continued to influence its trading valuation and price volatility.

Following its IPO, Skyverse delivered rapid top-line growth, expanding revenue from RMB 890.90 million in 2023 to RMB 1.380 billion in 2024, and reaching RMB 2.053 billion in 2025 — representing a two-year increase of more than 130%.² However, top-line expansion did not translate into sustained net profitability, setting up the key financial and operational challenges confronting the firm.

↑ Back to Top

IV. Segment-Level Economics, Product Line Breakdown & Materiality [38:00 - 54:00]

Understanding what Skyverse actually sells requires looking past marketing taxonomy and following the revenue dynamics across its product lines.

In 2025, inspection equipment generated RMB 1.364 billion in revenue—a 38.52% increase—at a gross margin of 53.29%. Metrology equipment generated RMB 623 million, surging 72.71%, but at a lower gross margin of 41.91%. Software and services accounted for the remaining 3% of revenue, carrying a gross margin near 54.6%. Together, these segments yielded a blended gross margin of 49.93%, an increase of roughly one percentage point.⁸

Three key conclusions emerge from these financial results.

First, inspection serves as the profit engine, while metrology acts as the growth engine. Because metrology grew nearly twice as fast while carrying a gross margin roughly eleven percentage points lower, product mix represents the primary structural driver of corporate margins. Consequently, rapid top-line growth can coincide with squeezed overall profitability, creating a potential trap for investors who focus strictly on headline revenue figures.

Second, the software business remains strategically notable but financially minor. Skyverse offers three intelligent software families: a yield management system for root-cause analysis, virtual metrology, and yield prediction; an automatic defect classification system that ingests defect data from all inspection tools on a line to categorize anomalies by size, shape, position, and clustering; and a lithography overlay analysis and feedback system that integrates data across multiple tool types and brands to correct scanner alignment.² Strategically, software integrated across a fab's entire defect dataset creates strong customer lock-in that is harder to displace than individual hardware units. However, contributing roughly 3% of total revenue, the software segment remains an emerging opportunity rather than a core financial driver.

Third, Skyverse maintains a broad product portfolio relative to its overall operational scale. During an April 2026 investor call, management detailed thirteen equipment series and three software series spanning optical, electron-beam, and X-ray detection technologies.⁶ In 2025 alone, the company introduced wafer flatness metrology, electron-beam critical-dimension metrology, and two specialized X-ray instruments: one targeting high-aspect-ratio etch structures in advanced memory, and another designed to inspect copper-fill voids in through-silicon vias for high-bandwidth memory (HBM) and advanced packaging.²⁶

Portfolio breadth offers distinct competitive advantages in semiconductor equipment. Management asserts that customers adopting a full product suite gain purchasing efficiency, faster service response, and lower vendor-management overhead, while the supplier secures deeper customer integration.⁶ This logic mirrors the strategy pursued by global incumbents like KLA. However, broad product expansion entails substantial upfront capital and engineering demands: each of Skyverse's thirteen tool series requires dedicated optical engineering, mechanical design, algorithm development, and lengthy customer qualification cycles before generating revenue.

Beyond new product launches, product iteration serves as a crucial driver of near-term revenue. In its 2025 annual report, Skyverse attributed top-line growth not only to its new darkfield series, but also to fourth-generation unpatterned wafer defect inspection systems and third-generation overlay metrology tools designed for finer process nodes.² Incremental upgrades allow the company to leverage established platforms, software architectures, and field-service teams.

Iterative upgrades extend existing customer relationships and open doors to more advanced manufacturing lines at a lower cost per revenue unit than developing entirely new product categories. However, this dynamic raises questions regarding elevated research expenditure: if revenue growth stems primarily from refining existing platforms, research and development intensity would typically be expected to moderate rather than expand.

A key forward-looking demand indicator appears on the balance sheet: contract liabilities—representing customer prepayments for equipment pending delivery or customer acceptance—reached RMB 881 million at year-end 2025, a 55.8% increase.⁸ Because equipment revenue recognition depends on final customer acceptance, contract liabilities offer a direct signal of customer commitments ahead of formal accounting recognition. The fact that contract liabilities grew faster than recognized revenue indicates an expanding order book, supporting volume growth expectations even as margin pressures persist.

However, long-term growth prospects depend heavily on higher-value product categories.

In darkfield nano-patterned wafer defect inspection, progress has been substantial. By April 2026, management stated that darkfield tools had completed production-line validation across multiple major domestic logic and memory manufacturers and achieved volume sales, serving as a key revenue driver for 2025 and the first quarter of 2026 while capturing a growing share of new orders.⁶ Darkfield inspection represents a US$1.05 billion global market, where Skyverse has established a meaningful domestic footprint.²

By contrast, commercial adoption in brightfield patterned inspection—the industry's largest segment at US$2.79 billion globally—remains a longer-term objective. In April 2025, management disclosed that initial brightfield systems had shipped in small volumes for customer evaluation, noting that positive technical feedback was balanced by significant engineering challenges that would delay meaningful revenue contributions.⁵ By April 2026, management reported that validation on customer production lines had progressed over an extended period with improved stability, projecting further milestones later in the year.⁶ In its August 2026 half-year commentary, management highlighted both darkfield and brightfield tools as expected volume drivers for the second half of the year, anticipating final customer qualifications before year-end.³

While optical inspection development programs typically require five to seven years from prototype to commercial adoption, three consecutive reporting cycles of optimistic management updates have yet to yield disclosed volume orders. Skyverse has demonstrated that its brightfield tools can undergo extended evaluation at leading fabrication plants, but it has not yet established that these systems can command high-volume commercial orders at gross margins comparable to its established inspection lines.

In contrast to brightfield development, advanced packaging and HBM product lines provide a more immediate commercial catalyst. Management states that Skyverse's qualified equipment suite addresses comprehensive inspection and metrology requirements for domestic HBM and 2.5D/3D packaging customers, encompassing darkfield and unpatterned inspection, patterned wafer inspection, 3D topography, overlay, and dielectric film thickness.⁶ Demand momentum was evident in the first quarter of 2026, when advanced packaging revenue temporarily exceeded 30% of total revenue following HBM orders placed in 2025.⁶ Additionally, a wafer flatness metrology system shipped to a leading domestic HBM manufacturer in the first half of 2026.⁴

However, management also noted during investor communications that the elevated revenue contribution from advanced packaging would normalize as front-end equipment sales resumed their typical weighting, adding that advanced packaging and mature-node tools yield lower gross margins than front-end advanced-node systems.⁶ As a result, current HBM-driven growth serves primarily as a volume driver while exerting near-term downward pressure on blended gross margins.

In summary, Skyverse has successfully expanded beyond its foundational unpatterned-inspection and film-metrology tools into a broader equipment platform, securing viable market positions in darkfield inspection and overlay metrology alongside an expanding, albeit lower-margin, HBM customer base. However, commercial penetration into brightfield inspection—the industry's largest and highest-value market—remains incomplete, shaping the financial performance examined in subsequent sections.

↑ Back to Top

V. Capital Allocation & M&A Strategy: Pure-Play Organic R&D vs. Global Comps [54:00 - 66:00]

Virtually every global player competing in semiconductor process control grew through acquisitions.

KLA formed through the 1997 merger of KLA Instruments and Tencor Instruments before acquiring Orbotech in 2019 to add printed-circuit and display inspection. Onto Innovation emerged from the 2019 merger of Rudolph Technologies — Chen Lu's former employer — and Nanometrics. Nova Ltd. acquired Ancosys to expand into in-line chemical metrology. Across the industry, breadth has historically been bought rather than built, because each measurement modality presents a distinct physics challenge that requires years to solve independently.

Skyverse has pursued the opposite path. Company filings emphasize a commitment to 自主研发、自主创新 independent R&D and independent innovation since its founding, utilizing a four-stage internal development framework — concept and feasibility, Alpha, Beta, and volume production — alongside a dedicated research platform for forward-looking technology and key module development.² All thirteen of its core product series were developed entirely in-house.

This organic model carries clear trade-offs. On one hand, available domestic acquisition targets consist primarily of sub-scale startups with single product lines and limited customer footprints, where acquisitions would introduce integration risks without securing established fab relationships. On the other hand, developing thirteen tool families organically within a single decade requires immense capital — an expense reflected directly across Skyverse's income statement.

Skyverse's financing history illustrates the scale of this capital effort. Since its listing, the company has tapped equity markets twice. Its May 2023 initial public offering raised RMB 1.888 billion in gross proceeds. In October 2025, it completed a private placement of 28.57 million shares at RMB 87.50 per share to 23 institutional investors, bringing in RMB 2.500 billion gross and RMB 2.481 billion net to fund the industrialization of brightfield and darkfield nano-patterned inspection and optical critical-dimension metrology systems.¹⁰ Across both offerings, total gross proceeds reached RMB 4.388 billion.⁹ By mid-2026, the main IPO production facility was fully operational with 89.90% of its funds deployed, while 43.46% of the private placement capital had been utilized, supplemented by credit facilities of up to RMB 6 billion.⁴

These capital raises indicate a business aggressively expanding rather than harvesting established profits. They also challenge any view of Skyverse as strictly capital-disciplined. When the company priced its private placement at RMB 87.50 per share in October 2025, it issued equity at a fraction of the share price, which climbed as high as RMB 459 over the following twelve months.¹³ Existing shareholders suffered substantial dilution relative to the market valuation achieved soon after. While management secured liquidity to finance R&D, the execution highlights the steep cost of dilutive financing during rapid expansion.

Operating results reflect the financial impact of this aggressive spending. In the first half of 2026, research and development expenditures absorbed 40.51% of total revenue, reaching RMB 381.00 million — a 33.55% increase year over year. Skyverse expanded its R&D headcount from 577 to 820 personnel, with doctorates comprising 19.63% of the team and master's degrees representing 62.56%. The company held 848 intellectual property assets, including 361 invention patents.³⁴

Compounding these research costs was introductory pricing on initial tool evaluation units. Management acknowledged that initial tools delivered to customer fabs are often "priced slightly below normal selling price." Consequently, Skyverse posted a net loss attributable to shareholders of RMB 127.44 million on revenue of RMB 941.02 million in the first half of 2026, widening significantly from an RMB 18.35 million net loss in the prior-year period.¹³ Operating cash flow reached negative RMB 190.28 million over the same period.¹

The historical falsification test on capital allocation. Evaluating these financial outcomes requires testing management's ongoing assertion that heavy R&D spending is a temporary phase that will normalize as revenue scales. A review of disclosures across consecutive reporting cycles offers a clear benchmark for assessing this guidance.

In April 2025, following 2024 annual results, executive guidance indicated that R&D expense growth would stabilize in 2025, the R&D expense ratio would gradually decline, and full-year profitability would improve.⁵ Skyverse initially met these targets: the R&D expense ratio declined from 36.07% in 2024 to 31.95% in 2025, while reported net income swung from a net loss of RMB 11.53 million to a net profit of RMB 58.65 million.²

However, in April 2026, following 2025 annual results and a loss-making first quarter, management issued similar guidance, projecting R&D growth below revenue expansion, a continuing downward trend in the R&D ratio, stable gross margins, and further profitability gains.⁶ Four months later, the 2026 half-year report revealed an R&D ratio of 40.51% and an expanded net loss, with management citing introductory pricing on new product evaluations and product mix for the margin compression.¹³

This outcome represents a clear departure from stated targets. While management has provided detailed operational justifications for spending increases — including concurrent development programs, staff expansion, and evaluation placements — the record demonstrates a pattern of guiding toward margin normalization before re-accelerating expenditure. For investors, management's R&D normalization guidance reflects strategic intent rather than a reliable financial forecast until sustained full-year moderation materializes.

A second, sharper test: what does the profit line actually contain? Examining the composition of earnings reveals further nuances in Skyverse's performance. In 2023, reported net profit stood at RMB 140.34 million, but net profit excluding non-recurring items was RMB 31.69 million. In 2024, reported net profit was negative RMB 11.53 million, while adjusted net profit dropped to negative RMB 124.10 million.

In 2025, reported net profit returned to positive territory at RMB 58.65 million, but net profit excluding non-recurring items remained negative at RMB 122.55 million.² For the first half of 2026, the reported net loss of RMB 127.44 million widened to an adjusted loss of RMB 191.76 million.¹

These disclosures indicate that Skyverse's core operations have not generated an annual profit excluding non-recurring items since 2023. The gap between reported and adjusted earnings — RMB 109 million in 2023, RMB 113 million in 2024, and RMB 181 million in 2025 — consists largely of government grants and R&D subsidies, which represent the largest non-operating income category for Chinese equipment makers. Consequently, the net profit turnaround in 2025 was driven primarily by non-operating subsidies rather than core operational profitability.

Government subsidies are common across the semiconductor capital equipment sector, providing policy support for domestic development. However, evaluating Skyverse's long-term commercial sustainability depends on core operational profitability — specifically, whether adjusted net income excluding non-recurring items can transition into positive territory.

Cash flow trends reinforce this operational picture. Operating cash flow deteriorated from negative RMB 52.04 million in 2023 to negative RMB 312.70 million in 2024, reaching negative RMB 862.17 million in 2025 — expanding even as top-line revenue nearly tripled.² Balance sheet data through mid-2026 shows inventory expanding by 49.84% to RMB 3.401 billion, alongside a 195.68% surge in accounts receivable to RMB 803 million.³ While high inventory levels are typical in semiconductor equipment as evaluation tools undergo customer validation prior to revenue recognition, a tripling of accounts receivable points to growing working-capital demands that warrant close monitoring if fab customers extend payment cycles.

Additionally, Skyverse's revenue recognition exhibits pronounced fourth-quarter seasonality that complicates mid-year performance assessments. In 2025, revenue expanded progressively through the year: RMB 294 million in the first quarter, RMB 408 million in the second, RMB 500 million in the third, and RMB 852 million in the fourth.

Profitability followed an even steeper curve: Skyverse recorded a cumulative net loss over the first three quarters of 2025 before generating RMB 73.35 million in net profit in the fourth quarter alone, securing a full-year net profit of RMB 58.65 million.²⁹ This pattern occurs because customer acceptance sign-offs — which trigger final revenue and gross profit recognition for delivered tools — cluster near year-end.

This seasonal concentration carries two key implications. First, judging annual performance based on first-half results provides an incomplete view of full-year trajectory. Second, reported annual profitability remains sensitive to the precise timing of customer sign-offs in December. A delay in customer acceptances across a quarter boundary can shift full-year results from net profit to loss, emphasizing the importance of evaluating performance on a trailing twelve-month basis.

Ultimately, Skyverse's internal R&D strategy has built technical product coverage across key process control segments. However, achieving sustainable unit economics remains an ongoing effort. Long-term margin recovery will depend on converting current customer tool evaluations into recurring commercial orders at standard market pricing — a transition that hinges on product competitiveness and customer adoption dynamics.

↑ Back to Top

VI. Strategic Position, Competitive Dynamics & 7 Powers / 5 Forces Analysis [66:00 - 80:00]

Evaluating Skyverse's strategic position reveals an unusual competitive structure: the company faces a dominant incumbent it cannot match on pure technology, operating in a market where state policy is systematically redirecting demand toward domestic suppliers.

Hamilton Helmer's 7 Powers.

Process Power — moderate, and improving. Skyverse's advantage lies in accumulated engineering know-how: multi-mode bright- and dark-field illumination systems, variable magnification optics, high-speed autofocus tailored for warped wafers, and classification algorithms that separate genuine defects from background noise.² The company's 361 invention patents signal ongoing R&D activity rather than an impregnable moat.

Stronger evidence comes from customer behaviour: leading domestic fabs have retained Skyverse tools on active production lines for years. The boundaries of this power are equally apparent from market performance: third-party reporting, citing company comparisons against KLA's portfolio, highlights a technology gap of two to three process generations on leading-edge applications.³ Trailing by two generations represents meaningful progress for a domestic entrant, but it does not confer pricing power.

Switching Costs — high, and the company's most durable competitive asset. Every process recipe calibrated, every defect classified, and every month of historical baseline data accumulated on a Skyverse tool raises the cost of replacing it. Management's disclosure that first-mover tools established on a customer's line create deep integration, and that fabs deliberately restrict the number of vendors per line, underscores why early qualification delivers asymmetric long-term rewards.⁶ However, switching costs protect the installed base rather than guaranteeing future wins; every new fabrication line and node transition triggers a fresh qualification cycle.

Scale Economies — emerging but currently negative. Fixed R&D expenditures are currently amortised across a revenue base of roughly RMB 2 billion, whereas global incumbent KLA spreads larger research budgets across a revenue base an order of magnitude larger. Skyverse remains on the disadvantageous side of this scale dynamic, making management's margin thesis dependent on expanding revenue fast enough to absorb its elevated cost structure.

Cornered Resource — moderate, and tied to key personnel. Founder Chen Lu's five years within KLA's R&D organisation and his IMECAS lineage represent scarce technical assets, with IMECAS maintaining a 2.85% equity stake.² However, relying on institutional memory concentrated in leadership introduces key-person risk—a factor reflected in Chen's 2025 compensation of RMB 5.2858 million, which stood at 11.22 times the company average.¹⁴

Counter-Positioning — essentially absent. Skyverse functions as a fast-follower executing import substitution along a technical roadmap established by KLA. The business model mirrors the incumbent's approach, seeking to offer comparable tool functionality at lower domestic price points.

Branding and Network Economies — immaterial in capital equipment markets characterized by a small group of highly specialized corporate buyers.

Porter's Five Forces.

Buyer power — high and structurally entrenched. Skyverse depends on a concentrated group of large, state-supported Chinese fabs. Customer executives understand the domestic procurement mandate, recognize Skyverse's need for production-line access, and leverage their position to negotiate introductory pricing.

Management's disclosure that initial tool placements are priced below standard commercial rates provides concrete evidence of customer bargaining power.³ Furthermore, domestic fabs pursue deliberate multi-vendor strategies—distributing orders among Skyverse, 上海精测 Shanghai Jingce, 睿励科学仪器 Raintree Scientific, 东方晶源 Dongfang Jingyuan, and accessible foreign vendors—to avoid single-supplier dependence.¹⁶ Import substitution expands the total domestic market, but it does not guarantee exclusive market share to any single vendor.

Supplier power — moderate to high, with geopolitical exposure. Critical optical components, deep-ultraviolet light sources, precision lenses, and specialized sensors originate from a narrow supply base in the United States, Japan, and Germany. This component reliance exposes domestic toolmakers to the input side of the same export-control environment that drove their initial market opportunity.

Rivalry — intense across two fronts. Against KLA and foreign incumbents, Skyverse competes on equipment pricing, field-service responsiveness, system customization, and supply-chain security—as well as data confidentiality, which management identifies as a growing priority for advanced-node customers.⁵ Against domestic peers, competition centers on securing limited qualification slots at customer facilities.

Among domestic competitors, Shanghai Jingce is widely viewed as KLA's closest domestic rival in optical inspection, backed by established relationships in memory fabrication; Raintree focuses on optical film metrology and inspection; and Dongfang Jingyuan specializes in electron-beam inspection and critical-dimension scanning electron microscopes (CD-SEM).¹⁶ Skyverse's primary differentiator is portfolio breadth across multiple optical and X-ray modalities rather than single-tool performance superiority.

Threat of new entrants — low in the near term, moderate in the medium term. Prolonged customer qualification cycles create high entry barriers that protect incumbents while delaying revenue realization for new tool introductions.

Substitutes — low. Advanced semiconductor manufacturing cannot operate without yield control and metrology instrumentation.

The critical strategic variable: how the incumbent responds.

Analyzing market dynamics as a static erosion of KLA's market share understates the incumbent's strategic options. An established market leader facing share pressure in a key growth region can deploy multiple counter-measures beyond price adjustments. KLA can expand domestic application engineering teams to match local service responsiveness, bundle brightfield inspection tools with metrology systems on favorable terms, or accelerate its technology roadmap to maintain a multi-generation lead as domestic competitors advance.

However, global incumbents cannot bypass government export restrictions, making the boundaries of international trade policy a primary determinant of Skyverse's addressable market. Consequently, long-term valuation models for domestic equipment vendors remain intrinsically tied to regulatory outcomes.

A potential trade de-escalation presents a distinct risk scenario: if export controls were relaxed on contested tool categories, domestic vendors would face a well-capitalized incumbent with decades of proprietary defect libraries, operating among customers seeking to maintain flexible supply options. In that environment, Skyverse's defensibility would rely primarily on the switching costs embedded in its existing installed base—making the rate of tool placements over the coming years a critical metric for long-term positioning.

In summary, Skyverse's competitive advantage centers on switching costs within established customer production lines, supported by domestic procurement policies and constrained by fabs seeking multi-vendor supply chains. While this position grants a foot in the door at major domestic fabs, current market valuations appear to price in a level of market dominance closer to a domestic equivalent of KLA than its current financial and technical profile reflects.

↑ Back to Top

VII. Management Credibility, Governance & Falsification Stress Test [80:00 - 90:00]

Governance at Skyverse is straightforward on paper and warrants detailed examination in practice.

Founder Chen Lu and Chairman Ha Chengshu serve as joint actual controllers, holding an aggregate concert-party voting stake of roughly 21%. At year-end 2025, 苏州翌流明光电科技有限公司 Suzhou Yiliuming Photoelectric held 10.80%, 深圳小纳光实验室投资企业 Shenzhen Xiaonaguang held 5.38%, and Ha personally held 4.79%. Ha owns 31.00% of Suzhou Yiliuming and holds an indirect majority of Xiaonaguang through 横琴承心 Hengqin Chengxin.² For 2025, audit firm 容诚会计师事务所 Rongcheng CPAs issued a standard unqualified opinion, with full director attendance at the annual board meeting and no dual-class share structures or special governance mechanisms in place.²

An aggregate 21% stake does not confer absolute voting control. While sufficient to direct corporate strategy with board consensus, it leaves management vulnerable to a coordinated shareholder bloc—a notable condition given recent shifts among major equity holders.

Institutional share reductions. SDIC's Shanghai fund (国投（上海）创业投资), which entered as an 11.40% pre-IPO investor, has systematically trimmed its position since December 2024. Between 24 November 2025 and 22 January 2026, the fund sold 2.35 million shares for roughly RMB 374 million. From 23 January to 16 February 2026, it sold an additional 643,900 shares for approximately RMB 123 million. In total, SDIC realized RMB 1.08 billion across 9.20 million shares, followed by an April 2026 plan to divest another 10.50 million shares that would elevate total proceeds toward RMB 2.77 billion.¹¹¹² By year-end 2025, SDIC's holding had decreased to 8.30%.²

While pre-IPO fund liquidations are standard following lock-up expirations, the duration and scale of SDIC's exits are noteworthy. The fund has systematically sold into strong share price appreciation over an eighteen-month period, monetizing gains rather than maintaining long-term exposure.

Insider transactions. Executive share sales present a distinct signal. On 22 November 2025, Skyverse disclosed planned share sales by Ha Chengshu (up to 125,000 shares), Chen Lu (up to 62,500 shares), and board secretary 古凯男 Gu Kainan (up to 9,375 shares), with all shares originating from equity incentive plans and attributed to personal financing needs.⁹ The absolute monetary value remains modest; Chen's planned sale represented roughly RMB 13.25 million, or approximately 2.5 times his annual salary.¹⁴

However, the timing of these sales creates optics challenges. Executive disposals—even minor ones—occurred two months after Skyverse raised RMB 2.5 billion in private equity and immediately prior to reporting its largest half-year net loss since listing.

Dividend policy execution. Skyverse's 2024–2026 shareholder return plan outlines a commitment to distribute at least 10% of annual distributable profits as cash dividends.⁴ For 2025, despite the parent entity recording no accumulated losses, the board proposed no cash dividend, stock dividend, or capital reserve conversion, citing current development stage, ongoing R&D requirements, and long-term capital allocation priorities.² While cash preservation is aligned with an operating cash outflow of RMB 862 million, skipping dividend distributions in the first full year of the return plan highlights a gap between declared policy commitments and operational execution.

Disclosure quality and transparency limits. Skyverse maintains detailed investor relations practices relative to STAR Market hardware peers. The company conducted over fifty investor meetings during the first half of 2026, maintaining a reported 100% response rate to submitted inquiries, publishing comprehensive meeting transcripts, and securing an ESG rating upgrade from BB to AA.⁴ Detailed disclosures in April 2025 and April 2026 regarding tool validation timelines and gross margin drivers have allowed external analysts to evaluate management guidance against subsequent reported results.

However, material analytical gaps remain. Skyverse omits customer identities from public filings, identifying its primary accounts only as leading domestic front-end chipmakers and advanced packaging firms.⁵ Interim disclosures omit customer concentration metrics, brightfield tool unit shipments, order book valuations, and specific commercial deployment targets, leaving brightfield progress descriptions qualitative across three consecutive reporting cycles.

Management attributes these omissions to heightened customer confidentiality requirements, particularly among advanced-node fabrication clients.⁵ Regardless of the underlying cause, external evaluation of Skyverse's primary growth catalyst relies on limited quantitative data.

Falsification stress tests.

Test one: The moat thesis. The thesis asserts that Skyverse possesses a comprehensive yield-control platform comparable to global incumbents. However, management's disclosures regarding brightfield inspection show an extended timeline: favorable progress updates in April 2025 and April 2026 were followed by an August 2026 expectation of passing customer qualification before year-end.³⁵⁶ In parallel, third-party reports cite a two-to-three-generation technological gap compared to KLA on advanced node applications.³ Verdict: Evidence narrows the scope of Skyverse's competitive moat.

Skyverse maintains established positions in unpatterned wafer inspection, darkfield inspection, overlay metrology, optical film metrology, and advanced packaging tools. However, it has not achieved high-volume commercial adoption in brightfield inspection, the industry's largest product segment. A confirming event would consist of disclosed volume orders for brightfield tools at a tier-one fab at pricing that preserves segment gross margins.

Test two: The domestic localization guarantee. The thesis posits that China's semiconductor self-sufficiency policy guarantees expanding market share for Skyverse. However, procurement patterns indicate that domestic fabs maintain multi-vendor sourcing strategies across domestic and non-sanctioned foreign equipment suppliers, while management confirms that customers cap vendor counts per line to control integration costs.⁶¹⁶ Industry localization and single-vendor share expansion represent distinct variables.

Verdict: The localization thesis holds in modified form. China's mainland metrology market expanded to US$4.45 billion in 2025, representing a 19.46% five-year compound annual growth rate, but market share within this addressable market remains competitive.² The thesis would be disproved if a leading domestic fab qualified a competing domestic supplier for a tool category where Skyverse previously held sole domestic vendor status.

Test three: Balance sheet strength. Skyverse closed the first half of 2026 with total assets of RMB 8.316 billion, shareholders' equity of RMB 5.082 billion, no corporate bonds, and access to RMB 6 billion in credit facilities.¹⁴

However, balance sheet expansion reflects capital raising rather than operational cash generation. Skyverse accumulated RMB 4.388 billion in gross equity proceeds over twenty-nine months, while shareholders' equity declined 0.73% during the first half of 2026 due to net losses.¹⁹ Furthermore, the 2025 operating cash outflow of RMB 862 million absorbed over one-third of the October 2025 private placement proceeds.²⁴ Verdict: Skyverse's balance sheet reflects substantial liquidity but does not demonstrate self-sustaining cash flow.

Growth has been financed through external equity offerings and debt facilities rather than internal cash generation. If operating cash outflows do not narrow alongside revenue expansion, additional equity dilution remains a key financial risk. The primary indicator of financial independence is the trajectory of operating cash flow relative to revenue growth.

Test four: Operating profitability. As detailed in Section V, adjusted financial results indicate that Skyverse's core operations have not generated net profit excluding non-recurring items since 2023, with 2025 net income relying on government subsidies.¹² Verdict: Claims of commercial profitability in 2025 are contradicted by adjusted financial disclosures. The alternative thesis—that Skyverse is approaching operational breakeven at scale—remains unverified, with net income excluding non-recurring items serving as the decisive metric.

↑ Back to Top

VIII. Material Risk Radar [90:00 - 98:00]

Three primary risks are material enough to alter the long-term investment case. Everything else is secondary noise.

1. Geopolitical and supply-chain restriction risk. Export controls on advanced foreign equipment represent a double-edged sword at the core of the investment thesis. Trade restrictions created Skyverse's market opportunity: every yield-control tool a Chinese fabrication plant cannot purchase from KLA is a system it must source domestically or manage without. However, Skyverse's own bill of materials depends on the same restricted global supply base for deep-ultraviolet laser sources, precision optics, high-sensitivity detectors, and specialized components required to achieve sub-nanometre optical resolution.

The company's filings acknowledge that the industry requires multidisciplinary integration across optics, algorithms, software, and electromechanical control.² The specific exposure lies in procurement: if trade controls tighten on sub-tier optical components or laser sources at a level of granularity matching previous lithography restrictions, Skyverse's technical roadmap toward brightfield parity could be constrained by component procurement rather than internal engineering capability. Public filings disclose no explicit hedge against component restrictions, and few domestic alternatives exist.

2. Fab capex cyclicality and mature-node overcapacity. Skyverse's top-line revenue operates as a direct derivative of Chinese fabrication plant construction. SEMI data cited in the 2025 annual report put mainland China semiconductor equipment sales at US$49.31 billion in 2025 — over 36% of the global market and the sixth consecutive year as the world's largest buyer, against a global total of US$135.06 billion growing more than 15%, with forecasts of US$145 billion in 2026 and US$156 billion in 2027.² This macro expansion provides a strong industry tailwind.

The risk lies in capex distribution: a substantial portion of domestic capacity expansion has targeted mature nodes (28nm, 55nm, and 90nm) where global overcapacity presents an active concern. Price erosion on mature chips eventually leads fabrication plants to defer or cancel equipment orders. Skyverse has partially insulated itself: advanced-process work exceeded 60% of 2025 revenue and over 70% of orders, per management.⁶ However, this insulation concentrates remaining commercial exposure onto a smaller group of leading-edge domestic fabrication accounts, shifting cyclical risk into customer concentration risk.

3. Brightfield execution and qualification risk. As the primary company-specific risk, brightfield inspection represents Skyverse's largest growth catalyst and its biggest execution test. Skyverse has committed private placement proceeds to brightfield industrialisation, maintained engineering teams on the program for years, and repeatedly signaled to the market that qualification results are near.⁶¹⁰ If these tools fail to reach commercial yield-parity qualification on leading-edge production lines, the consequence extends beyond a single delayed product family: Skyverse's addressable market would remain permanently excluded from brightfield inspection, the yield-control industry's largest product category, leaving corporate gross margins anchored to lower-value tool lines. Current market valuations do not reflect the impact of an unsuccessful brightfield rollout.

4. Key personnel concentration risk. Skyverse's technical authority is concentrated in a founder who is simultaneously chairman, general manager, and a designated core technical staff member — three roles that at most listed companies sit with three people.¹⁵ The company has clearly recognized this concentration; equity incentive grants extend across the technical organization, and the R&D headcount expanded from 577 to 820 within a year, with nearly a fifth holding doctorates.³ However, institutionalizing specialized optical design judgment requires significantly more time than recruiting optical engineers, particularly as domestic competitors recruit from the same talent pool. An executive departure at the top of the technical organization would represent a disruption to product roadmaps rather than a routine governance transition.

Secondary working-capital and valuation risks. Two secondary balance-sheet items warrant ongoing monitoring. First, working-capital intensity is reflected in an inventory of RMB 3.401 billion against first-half 2026 revenue of RMB 941 million, pointing to an extended conversion cycle; while much of that inventory represents evaluation systems awaiting customer sign-off, any qualification failure converts inventory into an impairment question rather than a revenue recognition event.¹³ Second, valuation risk operates as an independent risk factor: at roughly RMB 137.1 billion of market value in late July 2026 against trailing revenue of RMB 2.05 billion, the shares traded above 60 times sales, within a 52-week range spanning RMB 81.19 to RMB 459.¹³ For an enterprise whose core operations remain loss-making excluding non-recurring subsidies, premium valuation multiples offer no cushion for execution delays or qualification setbacks.

↑ Back to Top

IX. Playbook: Key Business & Investing Lessons [98:00 - 106:00]

1. The yield referee advantage — and its precondition. Process control represents the highest-leverage niche in semiconductor capital equipment because it sits directly between a fabrication plant's operating costs and its final profitability: accounting for roughly a tenth of the equipment budget while governing the wafer yield that determines overall fab economics. That structural leverage explains the high margins of global incumbents like KLA. However, the investor takeaway is conditional rather than automatic: the referee advantage accrues exclusively to whichever vendor the fab trusts to inspect its production lines.

Trust in this sector is built over years of accumulated defect classification data on a specific tool, transferring to a domestic challenger only when that challenger's diagnostic accuracy is demonstrably equal. Skyverse's darkfield commercial progress alongside its brightfield evaluation delays — coexisting within the same firm under identical market tailwinds — offers a stark natural experiment: matching engineering teams and customer accounts produced vastly different commercial outcomes based entirely on whether the specific tool cleared the customer's technical threshold.

2. The hard-tech qualification paradox. To secure long-term semiconductor tool placements, an equipment vendor must first place evaluation units on customer lines at minimal cost. Skyverse's own disclosure — that initial evaluation units ship below standard commercial pricing — states an industry reality plainly.³ The paradox is that the qualification phase is simultaneously when field support costs peak and revenue quality is lowest, causing an equipment maker's financial performance to appear most strained precisely as customer testing expands.

This dynamic creates an analytical trap in both directions. Investors who dismiss initial operating losses risk ignoring structural cash burn, while those who excuse ongoing losses indefinitely overlook suppliers that never transition past customer evaluations. The discriminating indicator is not the initial net loss, but whether subsequent tool shipments to the same customer command full commercial pricing. Consequently, segment gross margin, rather than headline revenue growth, serves as the critical metric tracking commercial progress.

3. Evaluating import substitution plays. Qualification is not commercialization. Skyverse has demonstrated a rapid pace of technical development — expanding to thirteen product series across three detection technologies, including four new product lines introduced in 2025 alone.²⁶ However, converting those technical milestones into operating profit excluding government subsidies has proven far more challenging.

Both operational realities exist simultaneously, but commercial conversion determines long-term equity value. In evaluating import-substitution investments, tracking technical milestones must be weighed against the proportion of product lines that successfully transition into recurring volume orders at standard market pricing. A company achieving frequent technical milestones without corresponding margin expansion represents a notable engineering achievement that remains an unproven commercial thesis.

A fourth lesson sits underneath all three, centering on buyer leverage. When an equipment vendor's customer base consists of a concentrated group of state-supported fabrication accounts, and the underlying investment thesis relies on domestic procurement mandates, policy backing does not automatically confer pricing power to the supplier. Instead, it provides fabrication plants with a reliable domestic alternative they can leverage during price negotiations. Commercial bargaining power in an import-substitution market flows to whichever party holds broader options, and in Chinese semiconductor capital equipment, that structural leverage remains with the fabs.

↑ Back to Top

X. Bull vs. Bear Case & Key Forward KPIs [106:00 - 114:00]

The bull case rests on three propositions, ordered by the strength of their underlying evidence.

The strongest argument is that the addressable market is expanding rapidly. With mainland Chinese inspection and metrology demand reaching US$4.45 billion in 2025 and growing at nearly 20% annually, combined with a domestic localization rate still in the mid-teens percentage range, the volume of equipment orders shifting to domestic suppliers over the next five years remains vast relative to Skyverse's RMB 2 billion revenue base.²¹³ To succeed commercially, Skyverse does not need to displace KLA globally; it simply needs to capture a meaningful portion of the domestic demand foreign vendors can no longer fulfill.

The second proposition assumes that Skyverse can replicate its success in darkfield inspection and advanced packaging across brightfield tools. Management asserts that Skyverse is the only domestic supplier capable of volume deliveries at advanced nodes across darkfield inspection, unpatterned inspection, overlay metrology, and metal film thickness—a breadth that competitors would require years to duplicate.⁶ Meanwhile, overseas revenue reached RMB 54.93 million in the first half of 2026—a more than fivefold increase from a small baseline, driven by Southeast Asian advanced-packaging orders—providing early evidence of commercial viability outside a protected domestic market.³⁴⁶

The third and weakest proposition centers on margin recovery: the expectation that introductory evaluation pricing will sunset as tools transition into repeat commercial orders, product mix will shift back toward higher-margin front-end advanced nodes, and research spending will moderate as revenue scales. While logically coherent, this margin expansion has not yet materialized, despite management projecting it across consecutive reporting cycles.

The bear case does not assume operational failure, but rather a persistent disconnect between demonstrated financial performance and market valuation.

The primary bear thesis is that Skyverse remains confined to medium-complexity tool tiers. Under this scenario, brightfield inspection fails to achieve volume parity, leaving foreign incumbents with the highest-margin sockets on advanced logic and memory lines. Top-line revenue continues to expand, but growth shifts toward advanced packaging and mature-node tools—categories that management acknowledged carry lower gross margins than front-end advanced systems.⁶ Consequently, revenue scales at 30% or more while blended gross margins decline and adjusted earnings remain in negative territory.

Supporting evidence appears across the financial statements: operating cash outflows widened over three consecutive years even as revenue nearly tripled, accounts receivable expanded by 195.68%, and non-recurring government subsidies accounted for the entirety of reported net profits.¹²³

From a governance perspective, four operational questions remain unresolved. First, why did a shareholder return plan committing to distribute at least 10% of annual profits yield zero cash dividends in its first full year?²⁴ Second, why was a RMB 2.5 billion private placement priced at RMB 87.50 per share when the stock subsequently traded above RMB 400 within months?¹⁰¹³ Third, why has guidance indicating R&D ratio normalization repeatedly preceded cost re-accelerations? And fourth, why have executive controllers engaged in share sales, however modest, during this expansion cycle?⁹ While individually manageable, these issues highlight ongoing execution and capital-allocation trade-offs.

Where the frameworks land. Synthesizing the competitive dynamics from Section VI, Skyverse's core strengths stem from switching costs on established production lines and an emerging process-power base that trails market leaders by two to three generations. Scale economies remain negative, while customer fabs retain significant bargaining power. This structure supports a business capable of growing alongside domestic fab expansion and generating modest returns once capital intensity stabilizes, rather than justifying valuation multiples modeled on market-dominant global incumbents.

Myth versus reality, in four lines.

Myth: Skyverse returned to profitability in 2025. Reality: Reported net profit of RMB 58.65 million relied on non-recurring subsidies, while core operations posted an adjusted net loss of RMB 122.55 million, with full-year reported earnings concentrated entirely in the fourth quarter.²

Myth: The company is a full-spectrum KLA equivalent for China. Reality: While offering thirteen tool series with established positions in darkfield inspection and metrology, the industry's largest segment—brightfield patterned inspection, representing 19.4% of global spend—remains in customer validation.²⁶

Myth: HBM and advanced packaging are the high-margin growth engine. Reality: Management confirmed that advanced packaging and mature-node tools carry lower gross margins than front-end advanced-node equipment, expecting the early-2026 packaging revenue spike to normalize over time.⁶

Myth: Localisation policy guarantees share gains. Reality: Government policy expands the domestic addressable market, but fabrication plants maintain multi-vendor sourcing strategies that prevent single-supplier dominance.⁶¹⁶

The KPIs that matter. Three metrics will determine the commercial trajectory:

1. Segment gross margin, particularly inspection. This metric reflects pricing power, customer evaluation discounts, and product mix. In 2025, inspection generated gross margins of 53.29% and metrology reached 41.91%, resulting in a blended gross margin of 49.93%.⁸ A sustained recovery in blended margin toward and above 50% alongside top-line growth would validate management's guidance on introductory pricing. Conversely, persistent margin erosion would confirm the bear case regardless of revenue expansion.

2. Brightfield conversion from validation to disclosed volume orders. Beyond qualitative progress updates, commercial validation requires disclosed volume shipments or customer acceptance milestones at a named tier-one logic or memory customer. This binary outcome determines whether Skyverse can address the industry's largest equipment category, a milestone management has highlighted across three consecutive reporting cycles.³⁵⁶

3. Net profit excluding non-recurring items. Adjusted net profit serves as the definitive measure of core operational viability. Having posted adjusted net losses of RMB 122.55 million in 2025 and RMB 191.76 million in the first half of 2026, achieving positive operational earnings will signal a transition from subsidy-supported development to self-sustaining commercial operations.¹²

↑ Back to Top

XI. Outro & Epilogue [114:00 - 118:00]

A distinct symmetry runs through this story. A physicist spends five years inside KLA learning how leading optical inspection systems are engineered, returns to a Chinese Academy of Sciences institute, and on the final day of 2014 founds a company in Shenzhen to build them domestically.¹⁵¹⁷ Eleven and a half years later, that business ships more than 400 tools annually, spans thirteen product families across optical, electron-beam, and X-ray technologies, and stands as the broadest domestic supplier in a sector China had previously ceded.⁶ Measured strictly as an engineering milestone, the expansion is notable—unfolding faster than standard qualification cycles would suggest.

Measured as a commercial business, however, the verdict remains open. Skyverse's core operations have not generated an adjusted net profit since 2023. Annual cash burn has deepened, while customer fabrication plants retain pricing leverage and actively cultivate competing vendors. Its highest-value product candidate—brightfield patterned inspection—has remained in customer evaluation long enough that the delay itself has become a material signal. Meanwhile, equity markets have assigned the firm a valuation that assumes these operational hurdles will resolve favorably.

What makes the coming eighteen months decisive is that the key outcomes are measurable and near. Brightfield tools will either secure volume commercial orders or fail to qualify. Blended gross margins will either rebound past 50% or remain depressed. Adjusted earnings will either turn positive or continue to sink. These are not questions requiring a decade of patience; they are operational checkpoints with specific reporting dates.

The broader lesson extends beyond a single company. Semiconductor equipment localization is frequently presented to investors across Chinese firms and national initiatives in the language of inevitability. The Skyverse record points to a more precise formulation.

Building technical capability in this industry is achievable—given sufficient talent, capital, and time, a dedicated engineering team can develop instruments that leading fabs will evaluate on their production lines. The far steeper hurdle lies in the final step: converting an evaluation unit that a customer will test into a product that the customer will reorder, at full price, in volume, over multiple years. That transition defines the true economic moat in semiconductor equipment—and it represents the barrier global incumbents like KLA have spent decades fortifying.

Skyverse has achieved the technical milestone that attracts headline attention. It has yet to complete the economic transition that secures durable commercial profitability.

---

References

1. 深圳中科飞测科技股份有限公司2026年半年度报告摘要 — CFi.CN 中财网, 2026-08-25 ↩↩↩↩↩↩↩↩↩↩↩

2. 深圳中科飞测科技股份有限公司2025年年度报告摘要 — 巨潮资讯 CNINFO, 2026-04-24 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

3. 上半年营收增长34.01%仍亏超1亿！中科飞测独家回应：费用上行毛利下行拖累盈利，静待明场及暗场设备放量 — 网易财经, 2026-08 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

4. 深圳中科飞测科技股份有限公司2026年度"提质增效重回报"专项行动方案的半年度评估报告 — CFi.CN 中财网, 2026-08-25 ↩↩↩↩↩↩↩↩↩

5. 深圳中科飞测科技股份有限公司投资者关系活动记录表 编号：2025-001 — 腾讯财经公告 PDF, 2025-04-27 ↩↩↩↩↩↩↩↩↩↩

6. 深圳中科飞测科技股份有限公司投资者关系活动记录表 编号：2026-002 — 东方财富公告 PDF, 2026-04-28 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

7. 中科飞测科创板上市：市值217亿 国投基金与华为是股东 — 搜狐财经, 2023-05-19 ↩↩↩↩

8. 中科飞测 | 2025年报&2026一季报点评：营收持续增长，先进制程量检测设备验证进展顺利 — 网易财经, 2026-05 ↩↩↩

9. 连亏股中科飞测实控人拟减持 2023上市2募资共43.88亿 — 新浪财经, 2025-11-24 ↩↩↩↩↩↩↩

10. 深圳中科飞测科技股份有限公司向特定对象发行股票发行情况报告书 — 巨潮资讯 CNINFO, 2025-10-17 ↩↩↩

11. 中科飞测股东国投创业60天减持234.87万股，套现3.74亿元 — 新浪财经, 2026-01-22 ↩

12. 中科飞测股东国投创业25天减持64.39万股，套现1.23亿元 — 新浪财经, 2026-02-24 ↩

13. 深度分析之中科飞测 — 东方财富财富号, 2026-07-28 ↩↩↩↩↩↩

14. 中科飞测核心技术人员、总经理、董事长陈鲁年薪528.58万，拟减持6.25万股 — 新浪财经, 2026-01-16 ↩↩

15. 陈鲁个人简介 — 中科飞测(688361)高管人员，中商产业研究院数据库 ↩↩↩↩↩↩

16. 半导体量测检测设备行业：市场呈现一超多强格局 国产厂商有望持续取得突破 — 观研报告网 ↩↩↩↩

17. 中科大少年班"学霸"，创业12年，干出1200亿半导体设备龙头 — 网易财经, 2026-07-06 ↩↩↩↩↩↩↩↩↩↩↩