National Silicon Industry Group: China's Semiconductor Substrate Gamble

I. Introduction & Episode Roadmap

There is a moment in every semiconductor fab that nobody photographs. Before lithography scanners hum, before etchers bite, and before metrology tools measure features thinner than a virus, a robotic arm reaches into a plastic cassette and lifts out a mirror-polished disc of silicon, 300 millimeters across and less than a millimeter thick. It looks unremarkable. It costs a fraction of what the finished chips printed onto it will fetch. Yet if its crystal lattice is misaligned by even a few atoms per billion — if a stray oxygen precipitate or a microscopic void sits where a transistor gate must be formed — the entire wafer, along with thousands of dollars of downstream processing, becomes scrap.

That disc is the product National Silicon Industry Group (NSIG, 上海硅产业集团股份有限公司, ticker 688126.SS, known in Chinese markets as 沪硅产业) exists to manufacture. As the largest and most internationalized semiconductor silicon wafer producer in mainland China, the company has been listed on the Science and Technology Innovation Board (STAR Market) of the Shanghai Stock Exchange (上海证券交易所科创板) since April 2020.¹ NSIG operates at the base of the electronics supply chain — a segment long overlooked by mainstream chip investors, yet so strategically sensitive that Chinese policymakers spent a decade and tens of billions of renminbi attempting to secure it.

A central paradox defines NSIG's performance. On the operational scorecard, the company achieved a major technical milestone. In 2014, mainland China produced virtually no 300mm prime silicon wafers for advanced logic and memory chips. By mid-2026, NSIG's subsidiary Shanghai Zingsemi (上海新昇半导体), combined with a new production base in Taiyuan (太原), reported a joint 300mm capacity of one million wafers per month, with management targeting 1.2 million by the end of 2026.² Shipment volume in the first half of 2026 grew by more than 90% year over year.²

On the financial scorecard, however, this expansion has produced severe losses. Revenue in 2025 reached CN¥3.716 billion, up 9.69%.³ Yet the net loss attributable to shareholders widened to CN¥1.508 billion, compared with CN¥971 million in 2024.⁴ Excluding government grants and other non-recurring items, NSIG's 2025 net loss reached CN¥1.774 billion.⁴ Consolidated gross margin — measuring the raw spread between wafer selling prices and direct manufacturing costs — dropped from positive 8.98% in 2024 to negative 17.06% in 2025.⁵ In the first half of 2026, even as volume nearly doubled, net losses widened further to CN¥965 million.²

A company can expand unit volume by 90% while incurring deeper losses. This outcome reflects the structural economics of silicon wafer manufacturing.

The analysis unfolds across seven parts. First, why raw silicon wafers represented China's fundamental bottleneck beneath all other semiconductor chokepoints. Second, the 2015 establishment of NSIG through state capital and its cross-border transactions — including the acquisition of Finland's Okmetic, controlling equity in Shanghai Simgui Technology (上海新傲科技), and a stake in France's Soitec — designed to compress decades of technical development. Third, the extreme engineering involved in converting quartz sand into single-crystal silicon purified to eleven nines. Fourth, commercial qualification at major domestic foundries, including SMIC (中芯国际), YMTC (长江存储), and CXMT (长鑫存储). Fifth, operational scaling under Dr. Chiu Tzu-Yin (邱慈云) and the STAR Market public listing. Sixth, the financial impact of the ongoing depreciation cycle, including an official inquiry letter from the Shanghai Stock Exchange requiring management to explain the widening losses. Seventh, the ultimate evaluation: whether NSIG represents a commercially viable enterprise or a strategic utility funded in part by public market capital.

This examination also tests key assumptions surrounding NSIG. The Okmetic acquisition, often cited as a benchmark for Chinese outbound technology investments, resulted in a goodwill write-down in 2025. Goodwill associated with Simgui was written off entirely. The National Integrated Circuit Industry Investment Fund (the Big Fund), whose backing was widely viewed as an implicit guarantee, has reduced its equity stake. Furthermore, NSIG's claim of domestic market dominance in 300mm wafers faces competition from a Xi'an-based producer that listed in late 2025 and reported to regulators that it held the largest 12-inch wafer manufacturing capacity in mainland China.

The analysis begins with the foundational bottleneck that preceded the broader semiconductor trade conflict.

---

↑ Back to Top

II. The Silicon Substrate Bottleneck & The Birth of NSIG (2014–2016)

In the mid-2010s, Chinese policy documents outlined strategic vulnerabilities across the semiconductor value chain, including lithography systems, electronic design automation software, and advanced logic fabrication. Yet beneath these high-profile gaps lay an equally fundamental challenge: the supply of raw silicon substrates.

China had become the world's largest consumer of semiconductors, yet domestic suppliers produced virtually no 300mm prime polished and epitaxial wafers for advanced logic and memory manufacturing. Global supply of semiconductor-grade silicon was—and remains—highly concentrated. Japan's Shin-Etsu Chemical (信越化学) and SUMCO (胜高), Taiwan's GlobalWafers (环球晶圆), Germany's Siltronic (世创), South Korea's SK Siltron, and France's Soitec together accounted for roughly 80% of the global market in 2024, representing total industry shipments of about 201 million wafers.⁶

This market concentration created substantial operational risk for domestic foundries. Wafer substitution requires extensive, costly requalification of established fab process recipes. Consequently, Chinese fabs constructing 300mm capacity in 2014 faced a long-term reliance on five international suppliers, four of which were based in jurisdictions allied with the United States. Interruption in supply—whether from trade restrictions or commercial reallocation—threatened to halt fab operations entirely.

These dynamics made silicon substrates a foundational chokepoint in China's industrial policy. While lithography attracted public attention, substrate development received sustained state capital investment.

Building a platform, not a company

National Silicon Industry Group was incorporated on December 9, 2015, as a state-backed aggregation platform rather than a conventional commercial startup.⁷ Designed to consolidate fragmented domestic silicon assets, integrate national research programs, and manage capital-intensive expansion, NSIG was structured to withstand the industry's prolonged investment cycles.

Its founding shareholders represented key entities in state semiconductor funding: the National Integrated Circuit Industry Investment Fund (国家集成电路产业投资基金, known as the Big Fund), Shanghai Guosheng Group (上海国盛集团), Shanghai Wuyuefeng Integrated Circuit (上海武岳峰集成电路), Shanghai Novel Microelectronics (上海新微电子), and Jiading Industrial Zone Development Company (上海市嘉定工业区开发有限公司).⁷ NSIG operated without a single controlling shareholder (实际控制人)—a corporate governance structure that distributed political risk across multiple state entities, while also leaving the company without a single obligate backstop.

The platform's primary technical asset was Shanghai Zingsemi Technology (上海新昇半导体科技), formed to execute China's national 02 Special Project (02专项) for 300mm silicon wafers. NSIG acquired control of Zingsemi in 2016.⁸ Zingsemi's early general manager was Dr. Richard Chang (张汝京), an industry veteran who had founded SMIC in 2000. Under Chang's leadership, Zingsemi pulled its first 300mm single-crystal ingot in 2016. After establishing initial production milestones over roughly three years, Chang transitioned management to NSIG's executive team to pursue new ventures.

This leadership transition highlighted a central operational challenge in substrate manufacturing. Successfully growing an initial single-crystal ingot represents a laboratory milestone; manufacturing hundreds of thousands of wafers monthly at low defect rates and competitive cost requires distinct operational disciplines and long-term capital deployment.

The mandate, and what it did not include

Group leadership passed to Yu Yuehui (俞跃辉), who assumed the chairmanship in April 2019.⁷ Yu brought institutional experience from state research organizations, having served as a researcher, deputy director, and Party secretary at the Shanghai Institute of Microsystem and Information Technology (SIMIT, 中国科学院上海微系统与信息技术研究所) under the Chinese Academy of Sciences, while also founding Shanghai Novel Microelectronics.

NSIG's foundational strategy combined greenfield domestic 300mm development with targeted overseas acquisitions to accelerate learning in crystal growth, polishing chemistry, and metrology.

However, the state mandate prioritized strategic supply security over traditional equity return thresholds. NSIG was established primarily to ensure that domestic fabs had access to 300mm substrates. Whether the platform earned a return above its cost of capital remained a secondary concern for its state sponsors—a policy orientation that directly influenced its financial trajectory by 2026.

With capital committed and its corporate framework established, NSIG initiated its acquisition strategy, turning first to Northern Europe.

↑ Back to Top

III. Cross-Border M&A & The Tech Foundation: Okmetic, Simgui, & Soitec (2016–2018)

Vantaa, a municipality outside Helsinki known for hosting Finland's primary airport, is also the headquarters of Okmetic Oyj. Operating with a workforce of several hundred employees, Okmetic specialized in silicon wafers that were modest in volume but offered attractive profit margins. Rather than competing in the capital-intensive 300mm market for logic and memory chips dominated by major Japanese suppliers, Okmetic produced 200mm and smaller wafers engineered for micro-electromechanical systems (MEMS) sensors, high-voltage power devices, and cavity silicon-on-insulator (SOI) structures. These specialty products served automotive and industrial customers with strict qualification requirements and high switching costs.

In the spring of 2016, NSIG launched a takeover bid for the Finnish manufacturer.

The deal

On April 1, 2016, NSIG announced a voluntary recommended public tender offer for all shares and option rights in Okmetic at €9.20 per share in cash.⁹ Because Okmetic's board had proposed a dividend of €0.65 per share at its annual general meeting, NSIG agreed that the dividend would not reduce the offer price. Okmetic shareholders ultimately received €9.85 per share in total value, representing a premium of roughly 29.6% over the pre-announcement closing price on Nasdaq Helsinki.⁹ Okmetic's board unanimously recommended the offer. The transaction secured regulatory approvals, gained acceptance from 95.72% of shareholders, and valued Okmetic at approximately €170 million including debt.¹⁰

The acquisition did not provide a shortcut into 300mm manufacturing, as Okmetic produced no 300mm wafers. Instead, the purchase delivered an operational European manufacturing footprint with established process discipline, access to tier-one industrial and automotive customers across Europe and North America, and international corporate standing.

NSIG expanded this strategy in June 2016 by announcing a voluntary public offer for Denmark's Topsil Semiconductor Materials A/S to acquire its float-zone silicon business.¹¹ The approach focused on acquiring small, specialized wafer manufacturers in Nordic jurisdictions at a time when regulatory approval for Chinese outbound technology investments was still attainable.

Simgui and the Soitec bargain

The domestic foundation of NSIG's strategy rested on Shanghai Simgui Technology (上海新傲科技), founded in 2001 as mainland China's primary manufacturer of silicon-on-insulator wafers. Unlike conventional bulk silicon wafers where current can leak into the underlying substrate, SOI wafers incorporate a thin layer of insulating oxide directly beneath the active silicon surface. This structure isolates transistors, reducing electrical leakage and parasitic capacitance. These characteristics make SOI substrates particularly useful for radio-frequency switches in smartphones, high-voltage power components, and silicon photonics.

However, manufacturing SOI wafers is technically complex and costly. The industry standard process, known as Smart Cut, involves bonding two wafers and cleaving one along a hydrogen-implanted plane, a technique patented by France's Soitec. Simgui was one of a small group of global manufacturers licensed to use Smart Cut technology, and the only licensed producer in mainland China.

To strengthen this relationship, NSIG acquired a 14.5% equity stake in Soitec in 2016, which it trimmed to approximately 12% the following year. This marked the first outbound investment by a Chinese entity in semiconductor materials, driven by Soitec's positions in fully depleted SOI (FD-SOI) and radio-frequency SOI (RF-SOI). Soitec and Simgui subsequently expanded their operational ties, signing agreements to scale 200mm SOI wafer production in China and designating Simgui as Soitec's exclusive sales agent in the mainland market.¹²

This structure combined an equity position in the technology licensor, exclusive domestic distribution rights, and sole domestic manufacturing licensing for Smart Cut SOI technology.

Testing the SOI optionality moat

Subsequent financial performance, however, revealed structural limits to this specialty strategy.

Because Simgui relied on Soitec's intellectual property for its advanced SOI production, its economic margins remained constrained by licensing terms and technology roadmaps controlled by the French licensor. Pricing power in advanced SOI accrued primarily to Soitec rather than its domestic licensee.

Market demand also fell short of expectations. In its 2025 annual report, NSIG recorded CN¥411 million in goodwill impairment losses across Simgui and Okmetic, the group's two primary specialty wafer subsidiaries.³ This included CN¥280 million against Simgui, which fully wrote off the goodwill associated with its acquisition, and CN¥130 million against Okmetic.¹³ In a written disclosure to the Shanghai Stock Exchange, NSIG attributed the Simgui impairment to weak demand for Soitec-related foundry services, noting depressed overall capacity utilization, underutilized epitaxial lines, and soft demand for Soitec-linked SOI processing.¹³ Okmetic's polished wafer operations maintained moderate utilization, but its SOI lines operated significantly below capacity.¹³ NSIG cautioned regulators that the remaining CN¥374 million of goodwill on Okmetic's balance sheet remains vulnerable to further write-downs if radio-frequency and automotive demand fails to recover through 2026 and 2027.¹³

These disclosures challenged the premise that Simgui and the Soitec alliance represented an immediate, high-margin asset. Nearly a decade after the initial transactions, the specialty segment yielded impairments rather than steady profit growth. Nevertheless, Okmetic and Simgui provided NSIG with established specialty manufacturing capabilities, international customer relationships, and production capacity for a potential demand recovery. By the first half of 2026, NSIG's total group SOI capacity exceeded 65,000 wafers per month, though foundry service revenue declined by approximately 22% due to sustained weakness in demand.¹⁴ Future performance will depend on capacity utilization metrics across 200mm and SOI lines, as well as whether remaining goodwill withstands impairment reviews in 2026 and 2027.

While Okmetic represented a smoothly integrated cross-border transaction, the subsequent write-downs underscored the long-term cyclical risks facing specialty wafer assets.

Meanwhile, NSIG gradually monetized its equity investment in Soitec. The company announced successive share sales through Euronext Paris in 2023, 2025, and 2026 to reduce its holding.⁸ Originally acquired as a core strategic alliance, the Soitec stake increasingly served as a financial liquidity reserve for NSIG's broader capital requirements.

The strategic M&A foundation laid between 2016 and 2018 ultimately served a broader objective: developing mainland China's domestic capacity for high-volume single-crystal ingot growth and 300mm wafer fabrication.

↑ Back to Top

IV. The Physics, Economics, & Geopolitics of Semiconductor Wafers

In every crystal-growing plant, operators monitor screens tracking the meniscus—the bright ring of molten silicon where the liquid melt meets the solidifying crystal. The orange, grainy image appears uninformative to an outsider, but it reveals the entire state of production. A slight wobble signals diameter drift; a subtle shift in brightness indicates a moving thermal field. Pull rates are adjusted in microns per minute over a single growth run lasting most of a day, where a flaw introduced in the third hour remains hidden until slicing and inspection days later.

This precise manufacturing reality defines the operations of National Silicon Industry Group. Understanding why silicon substrate production presents such steep technical and commercial hurdles requires examining a process where almost nothing is intuitive.

From sand to a single crystal

The manufacturing sequence begins with quartz sand refined into electronic-grade polysilicon with a purity of 99.999999999%—commonly known as eleven nines. That purity level corresponds to roughly one foreign atom per hundred billion silicon atoms, an environment comparable to identifying a single misplaced grain of sugar dissolved within an entire swimming pool of sugar.

The refined polysilicon is placed into a quartz crucible and melted at approximately 1,425°C. Production then relies on the Czochralski process. A single-crystal seed of silicon with a precise atomic lattice orientation is lowered into the melt and slowly withdrawn while rotating. As molten silicon freezes onto the seed, it adopts the seed's underlying structure. Under tightly managed thermal gradients and pull rates, the liquid grows into a single continuous crystal ingot—an uninterrupted lattice weighing hundreds of kilograms, measuring 300 millimeters in diameter, and reaching two meters in length.

Flaws in crystal growth occur at the atomic level and remain undetectable until they degrade customer manufacturing yields. Excessive pull speeds cause crystal dislocations to propagate. Unbalanced thermal profiles cause atomic vacancies to cluster into microscopic voids. Oxygen from the quartz crucible dissolves into the melt and forms precipitates inside the crystal. While controlled oxygen precipitates can beneficially trap metallic impurities, unmanaged precipitates ruin device yields if located where transistor gates are constructed. Maintaining a precise oxygen profile across a two-meter ingot represents a core proprietary discipline in substrate manufacturing that cannot be reverse-engineered from a finished wafer.

Once pulled, the single-crystal ingot is sliced by wire saws into individual discs, lapped, etched, and polished using chemical mechanical planarization until surface flatness varies by less than a nanometer across 300 millimeters. If a 300-millimeter wafer were expanded to the size of Shanghai, its surface elevation would vary by only a few centimeters. Chemical mechanical planarization uses an abrasive slurry within a reactive chemical bath, pressed by a rotating pad to strip surface material through simultaneous mechanical abrasion and chemical dissolution. Achieving uniform material removal across a disc the size of a dinner plate—without rounding the perimeter or introducing subsurface stress—requires decades of refined process know-how.

For advanced logic and memory chips, manufacturers deposit a final epitaxial layer of ultra-pure single-crystal silicon onto the polished surface. While polished bulk wafers retain the oxygen and defect history of the original crucible, an epitaxial layer grown from high-temperature gas inherits the underlying crystal orientation without its bulk impurities. The process effectively provides a pristine foundation atop the substrate, for which semiconductor fabs pay a premium. Because each stage—ingot pulling, slicing, lapping, polishing, and epitaxial deposition—relies on specialized equipment and independent yield metrics, a disc qualifies as a prime wafer only if every step executes without defect.

Why the economics are unforgiving

Translating these physical realities onto a corporate balance sheet exposes severe financial pressures. Constructing a modern 300-millimeter wafer manufacturing plant requires billions of dollars in capital expenditure before initial shipments begin, followed by aggressive equipment depreciation over five to seven years. Consequently, fixed overhead accounts for the overwhelming majority of unit production costs, incurred long before market pricing is established.

This cost structure creates three distinct operational dynamics that dictate NSIG's financial performance.

First, capacity utilization determines unit profitability. Because fixed depreciation costs remain constant regardless of output, unit costs drop sharply as production scales and surge when lines sit idle. A manufacturing plant operating at 90% capacity and one operating at 50% capacity do not merely experience a 40 percentage-point difference in unit costs; they frequently operate on opposite sides of cash profitability.

Second, customer qualification yields operate with binary risk. While a substrate manufacturer's internal yield determines how many prime wafers are produced per ingot, the foundry's chip yield on those substrates determines contract retention. Even minor increases in substrate defect density can lead foundries to cancel qualification entirely rather than reduce order volumes proportionally.

Third, long-standing incumbents hold structural cost advantages. Global market leaders like Shin-Etsu Chemical and SUMCO have operated 300-millimeter manufacturing lines for over two decades. With fully depreciated equipment assets and mature process recipes, these established suppliers can set selling prices that cover cash operating expenses while maintaining profitability—pricing levels that force newer entrants bearing full depreciation schedules to record net accounting losses on every wafer sold.

This structure also explains the divergence between NSIG's net income and operational cash flows. Because depreciation represents a non-cash accounting expense, a substrate manufacturer scaling new production capacity can report substantial net losses while maintaining positive operating cash flows, provided revenues cover direct cash operating expenses and interest obligations. For scaling materials producers, financial viability depends less on immediate net profitability and more on sustaining cash liquidity until fixed asset depreciation cycles mature.

The oligopoly, the newcomer, and the equipment problem

Market concentration among global substrate suppliers remained exceptionally stable through 2026. Japan's Shin-Etsu Chemical and SUMCO together supplied roughly half of global 300-millimeter wafer volume. Alongside GlobalWafers, Siltronic, SK Siltron, and Soitec, the top six international producers accounted for approximately 79.88% of the global market in 2024.⁶ As legacy 200-millimeter fabrication lines retire, 300-millimeter wafers represent the vast majority of total semiconductor substrate demand.

Evaluating NSIG's competitive standing requires distinguishing between global market share and domestic market dynamics. NSIG holds a low-to-mid single-digit share of the global 300-millimeter market. Within mainland China, the industry remains fragmented across multiple domestic suppliers. NSIG, Zhonghuan Leading (中环领先), Lion Micro (立昂微), Hangzhou Zing Semiconductor (杭州中欣晶圆), Shanghai Advanced Silicon (上海超硅), Xi'an ESWIN Material (西安奕材), and Shanghai Hejing (上海合晶) collectively generated roughly 86% of domestic output in 2024, with no single firm establishing market dominance.⁶ This domestic market fragmentation represents a primary competitive challenge facing NSIG.

The market is also shaped by a specific supply security dynamic. Silicon substrates represent a unique fab input that cannot be easily substituted, stockpiled in large quantities, or bypassed through process re-engineering. Foundries can qualify alternative photoresists or adjust chemical gas parameters, but chip manufacturing cannot proceed without substrates. Holding excessive wafer inventory is economically unfeasible due to high storage costs, surface degradation risks, and node-specific tuning requirements. These characteristics—non-substitutable, perishable in long storage, and concentrated among a small group of global suppliers—made silicon wafers a primary focus of Chinese industrial strategy. However, these same factors also lead foundries to maintain dual-sourcing strategies, ensuring that domestic fabs will continue to split wafer orders among multiple qualified vendors rather than relying on NSIG alone.

Finally, supply chain vulnerabilities persist in production equipment. NSIG relies on international vendors for advanced manufacturing tools, including high-precision crystal pullers, planarization equipment, and metrology systems used to certify defect rates. In its 2026 interim report, the company cited trade friction and equipment sourcing constraints as core risk factors, along with domestic market competition and technical gaps relative to global industry leaders.¹⁴ For an enterprise scaling 300-millimeter production capacity by roughly 200,000 wafers per month within a single calendar year, equipment delivery delays or customs restrictions present immediate operational risks.

The physical requirements illustrate why substrate manufacturing is technically complex, while the financial structure demonstrates why scaling is capital-intensive. The subsequent phase of NSIG's expansion determines whether operational execution could overcome these fundamental industry hurdles.

↑ Back to Top

V. Scaling 300mm: The Dr. Chiu Tzu-Yin Era & The STAR Market IPO (2019–2023)

By 2019, NSIG had a crystal, a customer pipeline, a Finnish subsidiary, and a problem. Zingsemi had proven it could make 300mm wafers. What it had not proven was that it could make them consistently enough, cheaply enough, and in enough volume for a fab to build a production plan around. That is a manufacturing problem, and manufacturing problems are solved by manufacturing people.

Hiring the operator

In April 2020, NSIG's board appointed Dr. 邱慈云 Chiu Tzu-Yin as president — the 总裁, functionally chief executive of operations.¹⁵ He had joined Zingsemi as general manager the previous year.

Chiu's résumé is close to the perfect fit for the specific gap NSIG had. He holds a doctorate in electrical engineering from the University of California, Berkeley, and an executive MBA from Columbia.¹⁵ More to the point, he had spent his career on the factory side of the Chinese semiconductor industry, including senior roles at 华虹NEC Hua Hong NEC and, most consequentially, the chief executive role at SMIC — where he is widely credited with steadying an operationally chaotic foundry and returning it to profitability. His reputation in the industry is not for vision or dealmaking; it is for yield, discipline, and the unglamorous grind of turning a process that works in a pilot line into a process that works on a shop floor at 3 a.m.

The market read the appointment exactly that way. Coverage at the time noted NSIG's market value jumping sharply on the news.¹⁵ Whether that enthusiasm was warranted is a question the 2024–2026 financials will answer, and the honest answer is: partly.

The STAR Market listing

NSIG went public on the STAR Market on April 20, 2020, offering 620.07 million shares at CN¥3.89 each — 25% of post-offering share capital — and raising gross proceeds of approximately CN¥2.412 billion.¹⁶ The listing made it the first mainland company to have achieved large-scale 300mm wafer sales to reach the public market.¹⁷

The debut was extraordinary and, in hindsight, instructive. The stock surged roughly 180% on its first day, and over the following weeks the gain from the offer price reached multiples of the issue level.¹⁸ The fundamental basis for that enthusiasm was thin: in the four years before listing, cumulative net profit had been minimal. What investors were buying was not earnings. It was a call option on Chinese semiconductor self-sufficiency, priced with the exuberance of a market that had just been given a Nasdaq-style venue and a national policy narrative to trade.

The use of proceeds was concrete. CN¥1.75 billion was earmarked for the second phase of the 300mm wafer technology development and industrialization project, adding 150,000 wafers per month of capacity on top of the 300,000 per month line Zingsemi had already built.¹⁶ NSIG returned to the market in 2021 for a follow-on that raised net proceeds of about CN¥4.946 billion, and again in 2025 for approximately CN¥2.079 billion.¹⁹ It is worth noting plainly: this company has been a repeat and substantial consumer of external equity, not a self-funding compounder, and any characterization of its capital story should start there.

Qualification, and what qualification is worth

The commercial breakthroughs came in sequence. Zingsemi's wafers passed validation at 上海华力微 Shanghai Huali Microelectronics and SMIC in 2018, and in 2019 its 28nm logic and 3D NAND prime wafers were qualified at YMTC.¹⁷ By September 2021, cumulative Zingsemi 300mm shipments had exceeded 3.4 million wafers.²⁰ Zingsemi's subsidiaries led or participated in seven national 02 Special Project programs, including the 40–28nm and 20–14nm 300mm wafer industrialization efforts.¹⁷

Each qualification is a genuine achievement, and it is worth being concrete about what one involves, because the word does a lot of work in bullish commentary. A fab does not simply try a new wafer and decide it likes it. It runs the candidate substrate through the full process flow alongside its incumbent supplier's material, measures electrical results device by device, tears wafers apart for cross-sectional analysis, tracks defect maps against yield loss signatures, and repeats the exercise across multiple production lots and multiple crystal batches to confirm the supplier can reproduce the result rather than having got lucky once. Only then does the wafer enter the approved vendor list, and even then it typically starts with a small allocation that grows as confidence builds. The process routinely takes twelve to eighteen months and consumes expensive test wafers that produce no saleable chips. Once complete, the fab has a real incentive not to repeat the exercise.

That incentive is the source of everything bullish anyone says about this company. It is worth testing whether it does what the bulls claim.

Testing the fab qualification moat

So does qualification create pricing power? This is the load-bearing question for the bull case, and the company's own subsequent disclosures answer it.

They answer it no.

In its reply to the Shanghai Stock Exchange's inquiry on the 2025 annual report, NSIG disclosed that the average selling price of its 300mm wafers fell 8.83% in 2025 versus 2024, and that because the price decline exceeded the reduction in unit cost, the 300mm gross margin contracted by 5.19 percentage points.¹³ For 200mm and smaller products, gross margin fell 11.9 points.¹³ Management attributed the losses to a lagging industry recovery, intensifying domestic competition, and its own capacity expansion.⁵ Prices across all wafer types declined in both 2024 and 2025.¹³

Read that carefully. NSIG was qualified at the major domestic logic and memory fabs, was growing volume rapidly, and still absorbed a high-single-digit price cut on its flagship product. That is what a market with high switching costs but abundant competing qualified supply looks like. Switching costs bought NSIG allocation — a guaranteed seat at the table. They did not buy the right to set the price at that table, because the fabs had multiple qualified vendors, both foreign incumbents with depreciated assets and a growing set of domestic rivals, and used them.

There is a structural reason this was always likely, and it is worth stating because it will not change with the cycle. Qualification is expensive for the fab, which is why fabs do it rarely. But a fab that has already paid that cost at three or four suppliers holds a permanently strong hand: the sunk expense is behind it, and the marginal cost of shifting volume between already-qualified vendors is close to zero. Switching costs, in other words, are a barrier at the moment of entry and an irrelevance thereafter. The supplier pays to get in; the buyer harvests the option forever after. NSIG spent a decade and enormous capital buying its way onto those vendor lists, and what it purchased was the right to compete on price against everyone else who had done the same.

The calibrated conclusion: the switching-cost moat is real but does its work on volume, not on price. It protects revenue continuity and makes NSIG hard to displace. It does not, on nine years of evidence, deliver pricing power in a downcycle. The falsifying test going forward is straightforward — in an up-cycle with tight supply, does NSIG's realized 300mm ASP rise in line with the global price increases the industry has seen in 2026, or does it lag?

That question runs directly into the financial reality of the past three years.

---

↑ Back to Top

VI. Segment Breakdown, Financial Anatomy, & The Depreciation Storm (2024–2026)

There is a specific kind of regulatory notice that no executive team welcomes: an official inquiry letter requiring a company to justify its expanding losses. NSIG received such a demand from the Shanghai Stock Exchange following its 2025 annual report and submitted its detailed reply in late July 2026.⁵ That filing provides rare transparency, forcing management to quantify in granular detail the operational dynamics behind its widening deficit.

The shape of the business

NSIG's operations split across three primary product lines and a minor specialty business. The 300mm segment—encompassing Zingsemi in Shanghai and the new manufacturing base in Taiyuan—generated CN¥2.439 billion in 2025, up 15.80% year over year, driven by a 27.01% increase in sales volume.³ This volume expansion lifted the 300mm division's share of total group revenue from 63.26% in 2024 to 66.28% in 2025.¹³ The 200mm-and-below segment, managed by Okmetic and Simgui, serves power discrete, sensor, and automotive markets; once the group's steady cash generator, it has faced weakening end-market demand. Within this division, silicon-on-insulator (SOI) products expanded their share of consolidated revenue from 4.75% to 6.47%.¹³ A fourth, minor line covers piezoelectric thin-film substrates.¹⁹

These figures reveal an underlying commercial strain: a 27% increase in physical unit shipments yielded only a 15.8% gain in revenue. Realized prices per wafer fell even as shipping volume expanded, demonstrating that unit growth came at the expense of average selling prices.

Three years of deterioration

A clear examination of NSIG's financial trajectory shows that the operational downturn began earlier than simple headline figures suggest.

In 2022, revenue reached CN¥3.600 billion with net profit attributable to shareholders of CN¥325.0 million; however, net profit excluding non-recurring items stood at just CN¥115.2 million.²¹ In 2023, revenue contracted 11.39% to CN¥3.190 billion, and reported net profit fell 42.61% to CN¥186.5 million. Crucially, net profit excluding non-recurring items reversed into a net loss of CN¥165.9 million.²¹

That 2023 divergence marks a critical turning point. Although NSIG recorded a positive headline profit, its core operations were already generating losses, with net earnings sustained primarily by government grants and non-operating subsidies. The financial squeeze did not start in 2024; rather, 2024 was when expanding depreciation expenses outgrew state subsidies.

In 2024, revenue rebounded to CN¥3.388 billion, but NSIG posted a net loss of CN¥971 million, which widened to CN¥1.24 billion when excluding non-recurring items.⁵ Gross margin remained positive at 8.98%.⁵

By 2025, revenue rose 9.69% to CN¥3.716 billion, yet net losses attributable to shareholders widened to CN¥1.508 billion, or CN¥1.774 billion excluding non-recurring items.⁴ Gross margin turned negative, dropping to negative 17.06%.⁵ The fourth quarter alone accounted for a net loss of CN¥876 million on revenue of CN¥1.07 billion, driven by end-of-year impairment charges.⁴ Full-year operating cash flow fell to negative CN¥559 million, and diluted earnings per share reached negative CN¥0.4561.⁴ Meanwhile, research and development expenses rose 32.63% to CN¥354 million, representing 9.52% of revenue.³

Impairments for 2025 totaled approximately CN¥694 million. This included CN¥411 million in goodwill write-downs across Simgui and Okmetic, alongside CN¥261.4 million in inventory write-downs against total inventory of CN¥2.308 billion, resulting from declining average selling prices and prolonged inventory destocking by customers.¹³⁵

The balance sheet question

Total assets reached CN¥33.787 billion by the end of 2025, a 15.4% increase, while equity attributable to shareholders expanded 37.6% to CN¥16.925 billion, driven primarily by shares issued to fund major asset acquisitions.⁴

Regulators also questioned NSIG's capital structure, specifically why the company maintained CN¥6.074 billion in cash alongside CN¥9.788 billion in interest-bearing debt.⁵ Maintaining substantial liquid reserves while incurring net interest expense creates a negative carry on capital. In response to the exchange, management stated that these cash holdings serve as designated reserves for ongoing facility construction, working capital, and debt-service buffers.⁵ While this provides a logical rationale for a capital-intensive expansion phase, it also underscores that NSIG relies on external financing to fund growth that internal operations cannot currently support.

2026: more volume, more loss, better cash

The first half of 2026 provided the initial empirical test for NSIG's operating leverage thesis, yielding mixed financial results.

Revenue rose 36.51% year over year to CN¥2.317 billion, but the consolidated net loss widened to CN¥965 million, compared with CN¥366.5 million in the prior-year period.² Growth was led by the 300mm segment, where shipment volumes expanded by more than 90% and revenue grew 57%.² By contrast, the 200mm-and-below line increased volume by approximately 16% while revenue rose only 5%, reflecting continued pricing pressure.²¹⁴ Revenue from foundry services declined by roughly 22%.¹⁴

Operating expenses below the gross line exacerbated losses. R&D investments grew significantly, asset impairments rose due to lower market pricing and legacy inventory, and financial expenses expanded as euro currency fluctuations impacted Okmetic's European balance sheet.² In the first quarter of 2026 alone, financial expenses surged more than sixteenfold year over year to CN¥142.6 million due to foreign exchange losses and higher debt servicing costs, while R&D expenditure grew 79.81%, reaching 13.15% of quarterly revenue.²²

Conversely, cash generation showed improvement, with operating cash flow turning positive at CN¥179 million for the half, compared to negative cash flow in the same period of 2025.² Total installed monthly capacity across Shanghai and Taiyuan reached one million 300mm wafers, alongside 200mm polished and epitaxial capacity exceeding 500,000 wafers per month and SOI capacity exceeding 65,000.² Intellectual property holdings expanded to 1,014 granted patents, including 697 invention patents.¹⁴

However, nameplate capacity requires careful interpretation. In silicon wafer manufacturing, headline capacity measures installed line and crystal-pulling infrastructure rather than shipped prime wafers. NSIG's filings report underutilization at Simgui and Okmetic, while top-line 300mm revenue indicates that actual shipped volume remains substantially below the one-million-wafer monthly threshold. The one-million figure reflects built production capability; closing the gap between installed capability and realized customer shipments remains essential to absorb fixed depreciation costs.

What management says, and what analysts ask

A contrast remains between management's public statements and market scrutiny. During its April 17, 2026 investor call, executive leadership highlighted efforts to release new capacity, scale production, improve factory utilization and yield rates, and lift gross margins by shifting toward higher-value product lines.²³ Company filings framed market conditions as a structural recovery, citing artificial intelligence and high-bandwidth memory (HBM) demand as counterweights to weakness in consumer electronics, while projecting pricing improvements in the second half of 2026.²¹⁴

Equity analysts have focused primarily on one operational benchmark: the timeline for 300mm gross margins to return to positive territory. Management provided a specific milestone in documentation supporting its 2025 asset acquisition, projecting that the acquired assets would reach consolidated profitability in 2026, identifying an 8.5% gross margin as the operational breakeven threshold.²⁴ This represents a clear, verifiable target for evaluating management's recovery roadmap.

In its inquiry response to regulators, NSIG projected global wafer shipment growth of 5.2% in 2026 and 8.6% in 2027, expressing expectation that market recovery will support margin expansion.⁵ Because these industry-wide demand growth projections are modest, financial recovery will depend principally on NSIG's internal operational execution, unit yields, and product mix rather than broader market tailwinds.

The question of who is accountable for that execution, and how management has deployed shareholder capital, forms the basis of the next analysis.

↑ Back to Top

VII. Management Credibility, Governance, & Capital Allocation Stress Test

On August 15, 2025, NSIG held an extraordinary general meeting to complete a scheduled board transition. Jiang Haitao (姜海涛) assumed the chairmanship, while Yu Yuehui (俞跃辉), who had chaired the group since 2019, stepped down from the board.²⁵ Dr. Chiu Tzu-Yin was re-appointed president, serving alongside a new chief financial officer, Huang Yan (黄燕), and board secretary Fang Na (方娜).²⁵

This leadership change occurred at a critical juncture: mid-way through the largest capital expansion program in the company's history, amid an extended downturn, and roughly nine months before the Shanghai Stock Exchange issued its formal inquiry into NSIG's expanding net losses. Although NSIG framed the shift as a standard board rotation and disclosed no internal disputes, the transition ended executive continuity at the board level precisely when the group's capital allocation strategy faced mounting external scrutiny.

Assessing the two leaders

Jiang Haitao inherits a board role whose primary function at NSIG is institutional rather than operational: securing capital injections from state entities, aligning corporate strategy with national policy, and negotiating municipal support packages such as the Taiyuan expansion. His predecessor, Yu Yuehui, executed this mandate aggressively. Under Yu's tenure, NSIG raised approximately CN¥4.946 billion in 2021 and CN¥2.079 billion in 2025 through equity offerings, executed a CN¥7.04 billion subsidiary acquisition, and structured a CN¥9.1 billion manufacturing expansion in Taiyuan backed by CN¥2 billion in local government co-investment.¹⁹²⁶²⁷ Measured against state policy goals for capacity expansion, this represented effective execution; measured against public equity returns, it produced substantial share dilution during an industry downcycle.

President Chiu Tzu-Yin brings an operational reputation built on his prior leadership of SMIC, where he restored operational discipline and profitability to a struggling foundry. At NSIG, operational metrics under Chiu's direct oversight have progressed, even as financial performance has deteriorated. In the first half of 2026, 300mm wafer shipments grew by more than 90% year over year, Zingsemi facility utilization increased, and operating cash flow turned positive despite ongoing construction. In 2025, NSIG developed 168 new wafer products, advancing 57 into mass production.³ However, negative gross margins and expanding net losses reflect persistent industry pricing pressure and heavy fixed-asset depreciation. The primary operational test for Chiu remains whether unit cost reductions can catch up with market pricing to achieve management's targeted 8.5% breakeven gross margin in 2026.

Incentives and ownership

The standard narrative regarding NSIG’s governance—that executive equity ownership is minimal, governance is dominated by state entities, and compensation is aligned with strategic volume targets rather than equity returns—requires qualification regarding scale. Executive compensation has been substantial: Chiu's disclosed 2023 remuneration reached CN¥12.976 million, attracting domestic financial media scrutiny in 2024.²⁸ While reflecting market rates for seasoned semiconductor executives, this compensation structure is largely uncoupled from net earnings or share price performance during periods of expanding corporate losses.

A fundamental governance characteristic of NSIG is the absence of a single controlling shareholder (实际控制人). The group's primary equity holders have been state-backed entities, led by Shanghai Guosheng Group and the National Integrated Circuit Industry Investment Fund (the Big Fund), with Big Fund Phase II also holding a major position.²⁹ This fragmented state ownership structure grants professional management operational independence, but it also means no single shareholder possesses both the equity incentive and the legal obligation to absorb long-term losses or backstop public equity investors during prolonged industry downturns.

The state backstop, tested

Assumptions that NSIG benefits from an implicit state guarantee that protects public equity valuation are contradicted by the transaction history of its primary state sponsor.

State-backed investment vehicles have steadily reduced their equity positions.

In June 2023, the Big Fund—which then held 567 million shares, representing a 20.76% equity stake—announced an intention to sell up to 3% of NSIG's total share capital, triggering an immediate single-day share price decline of over 8%.³⁰ Although that divestment plan was terminated early,³⁰ share reductions resumed. Between May 6 and June 4, 2026, the Big Fund sold 99.15 million shares—exactly 3% of total share capital—raising approximately CN¥2.626 billion in gross proceeds and lowering its equity stake to 12.49%.²⁹ In July 2026, regulators disclosed another planned share reduction by the fund.³¹

While a 12.49% holding remains substantial and aligns with the Big Fund's broader mandate to recycle capital across the semiconductor value chain, these transactions demonstrate a clear divergence between industrial policy and equity market support. State sponsors have acted as net sellers of public equity during NSIG's deepest loss cycle, generating persistent downward pressure on the stock price. State backing fully supports NSIG's strategic manufacturing capacity, but it has not functioned as a price support for public equity holders.

Capital allocation, examined

The acquisition of Okmetic established NSIG's specialty wafer footprint, though it later required goodwill impairments as market demand softened.

More recent capital deployments have been significantly larger. In December 2023, Zingsemi reached an agreement with the Taiyuan municipal government and the Taiyuan Zhongbei High-Tech Zone to build a 300mm crystal-pulling and wafer-processing facility. The project involved a total planned investment of CN¥9.1 billion—comprising CN¥2 billion from local government entities and CN¥7.1 billion raised by Zingsemi—targeting monthly capacities of 600,000 pulled crystal wafers and 200,000 finished wafers.²⁶ A broader group expansion program totaling approximately CN¥13.2 billion was disclosed during the same period.³²

In May 2025, NSIG announced a transaction to acquire minority equity interests in three key subsidiaries—Zingsemi Jingrui (新昇晶睿), Zingsemi Jingke (新昇晶科), and affiliated entities holding the pulling and processing operations for the 300mm Phase II project. The acquisition was valued at approximately CN¥7.04 billion, structured as CN¥6.716 billion in newly issued equity and CN¥324 million in cash, supported by a private placement raising up to CN¥2.105 billion at CN¥15.01 per share.²⁷ The deal consolidated 100% ownership of the 300mm manufacturing business. However, market observers noted that NSIG paid over CN¥7 billion to buy out minority stakes in loss-making subsidiaries precisely as the parent company's consolidated financial performance was deteriorating.²⁷ Following the transaction, Big Fund Phase II held a 9.36% equity stake in the enlarged company.²⁷

Over a three-year period, NSIG committed roughly CN¥30 billion across greenfield plant construction, minority equity buyouts, and supporting capital raises. This aggressive expansion occurred while average wafer selling prices were declining industry-wide and shortly before regulatory authorities requested justification for the group's expanding operating losses. While rapid capacity expansion is necessary to achieve minimum efficient scale in semiconductor materials manufacturing, the immediate financial consequences for public shareholders have included substantial share dilution, an extended depreciation drag, and negative gross margins.

This capital allocation pattern highlights a fundamental tension between strategic state objectives and public equity returns: whether public stock markets should finance a national security material program with a multi-year payback horizon, and whether buying out minority stakes in loss-making subsidiaries at a CN¥7.04 billion valuation served public shareholders or selling equity holders. In filings submitted to regulators by its financial adviser, NSIG maintained that full subsidiary consolidation improves supply chain integration and projected that the acquired assets would achieve operational profitability in 2026.²⁷²⁴ That milestone now serves as the primary benchmark for evaluating management's capital allocation strategy.

↑ Back to Top

VIII. Helmer's 7 Powers & Porter's 5 Forces Analysis

Strategic frameworks provide value only when allowed to yield uncomfortable conclusions. Applied systematically to National Silicon Industry Group, standard analytical models highlight severe structural pressures.

Hamilton Helmer's 7 Powers

Scale economies — present but insufficient. This represents NSIG's most plausible long-term economic power, yet remains its weakest current attribute. Fixed-cost absorption serves as the primary mechanism for achieving operational profitability as monthly 300mm capacity targets 1.2 million wafers.² However, the relevant competitive baseline is not NSIG's historical output, but global incumbents. Shin-Etsu and SUMCO operate at multiples of NSIG's volume with fully depreciated asset bases, and jointly committed approximately 150 billion Japanese yen in 2025 to add ultra-flat capacity for 2nm and 3nm nodes.²³ Scale economies function as an economic moat only when an enterprise reaches minimum efficient scale before capital reserves deplete—a challenging race while operating with negative gross margins.

Switching costs — real, but limited in economic protection. Requalification requires twelve to eighteen months and substantial test-wafer expenditures, securing customer retention for NSIG across its qualified accounts. However, as 2025 disclosures demonstrated, customer qualification failed to prevent an 8.83% decline in average selling prices for 300mm wafers.¹³ In this market, switching costs protect order allocation rather than pricing power, yielding high customer retention alongside weak pricing leverage.

Cornered resource — low, despite specialty silicon positioning. The Soitec licensing agreement grants subsidiary Simgui a unique position as mainland China's sole Smart Cut technology licensee.¹² However, a license represents rented intellectual property, and the complete goodwill write-off at Simgui demonstrates that technical exclusivity generates little value without sustained end-market demand.¹³ Meanwhile, foundational patents in single-crystal growth, polishing chemistry, and metrology remain concentrated among Japanese, American, and European incumbents.

Process power — credible in technical progress, unproven in cost parity. Proprietary manufacturing expertise in defect density, wafer flatness, warp control, and oxygen profiles represents the primary asset of an established substrate producer. NSIG has built a substantial technical portfolio, holding 1,014 granted patents—including 697 invention patents—while developing 168 new products in 2025, advancing 57 into mass production, and securing qualifications across logic, memory, image sensor, and power applications.¹⁴³ However, the ultimate measure of process power is cost performance. Manufacturing expertise that does not translate into a narrowing unit cost gap against established competitors like Shin-Etsu functions as technical capability rather than economic power.

Counter-positioning — absent. NSIG does not employ a business model that cannibalizes incumbent revenue streams. The company manufactures standard substrates using established industry methods, operating in a different geographic market driven by national policy objectives. Global incumbents face no business-model dilemma in competing directly on price and volume.

Network economies — absent. Silicon substrate manufacturing exhibits no network effects, as an individual fab's wafer adoption provides no inherent utility or value to other semiconductor producers.

Branding — negligible. Substrates are purchased strictly on technical specification, defect metrics, and contract pricing. While foundry qualification establishes operational trust, it does not confer brand equity or premium pricing power.

In aggregate, NSIG relies on one active economic power—switching costs—that protects volume allocation rather than pricing, alongside a prospective power, scale, that remains unachieved. This represents a narrow moat for an enterprise carrying CN¥9.788 billion in interest-bearing debt.⁵

Porter's Five Forces

Threat of new entrants — structurally high via state-backed capital. Traditional economic models suggest that multi-billion-dollar capital requirements and multi-year customer qualification cycles create formidable barriers to entry. In mainland China, state-directed capital has effectively lowered these entry barriers. Xi'an ESWIN Material listed on the STAR Market in October 2025, raising CN¥4.636 billion in the largest semiconductor equipment and materials initial public offering on A-share markets, becoming the first pre-profit enterprise admitted under updated listing rules.³³ ESWIN disclosed to regulators that it held the largest 300mm wafer manufacturing capacity in mainland China and ranked sixth globally, accounting for approximately 6% of global monthly shipments and 7% of global capacity in 2024, while projecting that its second facility would lift capacity to 1.2 million wafers per month, representing over 10% of global output.³³ Simultaneously, Shanghai Advanced Silicon has pursued a STAR Market listing despite holding less than 2% of global 300mm market share and recording accumulated losses approaching CN¥4 billion.³⁴

These competitive dynamics challenge assumptions of domestic market exclusivity for NSIG. ESWIN's regulatory disclosures identify it as a primary wafer supplier to major mainland memory manufacturers and leading logic foundries.³³ With two domestic producers simultaneously expanding toward monthly capacities of 1.2 million 300mm wafers to serve the same customer base, NSIG operates not as a sole domestic champion, but as a participant in a capital-intensive domestic expansion race.

Supplier power — high. Manufacturing equipment—including high-precision crystal pullers, wire saws, chemical mechanical planarization tools, and specialized metrology systems—remains heavily dependent on foreign vendors, leading NSIG to identify equipment sourcing constraints as a key risk factor.¹⁴ Additionally, fluctuations in raw polysilicon input costs have contributed to inventory valuation adjustments.¹³

Buyer power — high and structurally entrenched. Substrate demand is concentrated among a small group of high-volume foundries and integrated device manufacturers, including SMIC, Hua Hong, YMTC, and CXMT domestically, alongside international customers cited in company disclosures such as TSMC, UMC, and STMicroelectronics.¹⁹ Foundries maintain dual-sourcing strategies to preserve purchasing leverage and supply security, a dynamic illustrated by recent price reductions. Domestic industrial policy further reinforces buyer power by encouraging multiple qualified substrate vendors rather than relying on a single supplier.

Threat of substitutes — low. For high-volume 300mm logic and memory fabrication, no viable alternative substrate material exists. While compound semiconductors serve specialized power and radio-frequency applications, bulk single-crystal silicon remains the essential foundation for mainstream semiconductor manufacturing, providing structural demand stability for the wafer industry.

Industry rivalry — extreme across two fronts. NSIG faces simultaneous competition from global incumbents operating depreciated production lines and lower capital costs, alongside domestic peers expanding capacity under identical state policy support. In 2026, global substrate markets initiated price increases, with cumulative gains exceeding 15% across two pricing rounds; standard 300mm wafers rose 5% to 8%, while specialty substrates for artificial intelligence and high-performance computing increased 18% to 22%, driven by high-bandwidth memory requiring roughly three times the silicon area of standard DRAM and advanced 300-layer NAND architecture doubling substrate consumption per unit.²³ Whether NSIG can capture these global price increases or whether expanding domestic capacity depresses mainland pricing remains a central operational question.

↑ Back to Top

IX. Bull vs. Bear Case & Material Risk Radar

Every capital-intensive materials enterprise eventually reaches an operational inflection point where two distinct trajectories diverge. In one scenario, a company expands through an industry downcycle, absorbs heavy fixed-asset depreciation, reaches minimum efficient scale, and establishes a low unit-cost structure that funds long-term returns—the historical path of established global wafer incumbents. In the alternative scenario, an enterprise expands into a structurally oversupplied market, fails to achieve cost parity, and requires repeated capital injections to sustain operations. NSIG faces this structural crossroads, and empirical evidence through mid-2026 does not yet confirm which outcome will prevail.

The bull case: the operating leverage inflection

The bull thesis rests on three sequential operational premises.

First, domestic demand continues to expand as mainland China constructs a substantial share of global new fabrication facilities through 2028, expanding mature-node capacity.²³ Concurrently, structural shifts in silicon intensity—driven by artificial intelligence applications that require roughly triple the silicon area for high-bandwidth memory compared to standard DRAM, alongside doubled substrate consumption for advanced high-layer NAND architecture—steepen the long-term demand curve for 300mm substrates.²³ Having secured product qualifications across domestic foundries and memory producers, NSIG is positioning new capacity directly into this demand shift.

Second, depreciation is a fixed schedule rather than a permanent operational expense. Production tools installed between 2022 and 2026 will gradually complete their heaviest depreciation cycles in subsequent years. If plant utilization approaches full capacity while fixed overhead per unit declines, unit manufacturing costs fall rapidly. In substrate manufacturing, where gross margin represents a narrow spread over high fixed costs, operational leverage creates steep margin swings. Management identified an 8.5% gross margin as the operational breakeven threshold for its acquired 300mm assets, projecting consolidated profitability for these entities in 2026.²⁴ Supporting this trajectory, operating cash flow turned positive at CN¥179 million in the first half of 2026 despite peak construction activity, indicating that core operations are covering cash expenses even as accounting net losses widen.²

Third, global pricing tailwinds may lift realized revenues. If global wafer price increases in 2026 flow through to NSIG's realized average selling prices rather than being eroded by domestic competition, higher real prices will apply directly to a scaled cost base.²³

If all three factors align, financial results can shift rapidly. For an enterprise operating at a negative 17.06% gross margin, reaching profitability requires modest gains in capacity utilization and selling prices sustained over several quarters.

The bear case: the perpetual depreciation sink

The bear case contends that structural competitive dynamics will prevent sustained pricing recovery.

Foreign incumbents with fully depreciated assets can maintain baseline pricing that covers cash operating expenses, capping the prices NSIG can command for equivalent products. Concurrently, domestic rivals—including Xi'an ESWIN Material, Shanghai Advanced Silicon, and Zhonghuan Leading—are expanding capacity toward similar monthly volume targets with state capital backing and limited immediate return requirements.³³³⁴ In a market where multiple state-supported suppliers target the same 1.2-million-wafer monthly capacity for identical domestic customers, pricing power remains suppressed regardless of broader cyclical recoveries.

Furthermore, the fixed-asset depreciation trough may never fully materialize. Rapid technological transitions—such as ultra-flat substrate requirements for sub-2nm logic nodes and increasing NAND layer counts—demand continuous capital expenditure for next-generation crystal pullers, polishing equipment, and metrology systems. A substrate manufacturer required to re-tool constantly cannot harvest fully depreciated production lines. Reflecting this capital intensity, NSIG conducted equity raises in 2020, 2021, and 2025, while issuing CN¥1.2 billion in corporate bonds during the first quarter of 2026 alone.¹⁶¹⁹²²

Finally, geopolitical risks present asymmetric headwinds. Export controls on advanced semiconductor manufacturing equipment directly restrict NSIG's equipment procurement, while restrictions on domestic foundry customers indirectly limit substrate demand by slowing fab expansion.¹⁴

Weighing the two

Evaluating both perspectives requires isolating their core assumptions.

The bull thesis relies heavily on pricing leverage. While demand expansion and depreciation roll-offs are supported by empirical data, realized pricing remains vulnerable. The 8.83% drop in 300mm wafer average selling prices in 2025 occurred while NSIG was qualified, expanding volume, and gaining market share.¹³ This demonstrates that high customer qualification requirements secure order allocation rather than pricing power. Consequently, the bull case represents a bet on a broad cyclical price recovery across the wafer industry rather than superior pricing power specific to NSIG.

Conversely, the bear assumption of perpetual domestic overcapacity overstates long-term market fragmentation. Capital-intensive expansion races typically conclude through financial consolidation, demand growth, or shifts in state industrial policy toward industry rationalization. Given its balance sheet size, broad product portfolio, international operating footprint, and institutional backing, NSIG is positioned to act as an industry consolidator rather than an exit candidate during market rationalization.

The empirical evidence indicates that NSIG can achieve positive 300mm gross margins as volume, plant utilization, and depreciation schedules align with positive operating cash flows. However, the evidence does not yet demonstrate that the business will generate economic returns above its cost of capital on the approximately CN¥30 billion committed since 2023. Realized wafer pricing within a fragmented domestic market remains the decisive variable determining long-term equity returns.

Second-layer diligence and the risk radar

Regulatory Scrutiny and Asset Impairments. The formal inquiry letter from the Shanghai Stock Exchange regarding NSIG's 2025 annual report underscores ongoing governance oversight, requiring management to justify widening net losses, quantify impairment charges, and explain holding CN¥6.074 billion in cash alongside CN¥9.788 billion in interest-bearing debt.⁵ Additionally, management disclosed prospective impairment risks regarding the remaining CN¥374 million in goodwill associated with Okmetic, indicating that asset valuation will remain a key area of audit scrutiny.¹³

Foreign Exchange Exposure. Operating Finnish subsidiary Okmetic introduces euro-denominated currency exposure to a renminbi-denominated balance sheet. Financial expenses expanded more than sixteenfold year over year in the first quarter of 2026 due primarily to foreign exchange losses, with currency volatility cited again as a performance drag in the interim 2026 period.²²²

State Divestment Signals. Equity sales by the Big Fund in 2026 lower its shareholding while generating short-term downward pressure on share prices, reinforcing that state policy backing does not equate to public equity price support.²⁹³⁰³¹

Divergence in Core Earnings. The gap between reported net income and adjusted net income excluding non-recurring items remains substantial. Subsidies accounted for the entirety of reported net profit in 2023—turning an underlying operational loss of CN¥165.9 million into a reported profit of CN¥186.5 million.²¹ In 2025, non-recurring items reduced a CN¥1.774 billion adjusted net loss to a reported loss of CN¥1.508 billion.⁴ Evaluating long-term commercial viability requires analyzing performance net of state subsidies.

Valuation Decoupling. NSIG shares rose 81.51% between April and June 2026, lifting market capitalization back above CN¥100 billion, while domestic peers Lion Micro and Youyan Silicon gained 97.91% and 78.88% respectively.²³ That equity rally occurred alongside expanding corporate net losses, illustrating that public valuations in the substrate sector frequently track policy narrative and spot price headlines rather than realized net earnings.

↑ Back to Top

X. Strategic Playbook & Investing Lessons

Lesson one: national mandates create volume, not necessarily value. NSIG fulfilled its strategic mandate. Mainland China transitioned from producing virtually no domestic 300mm prime wafers to shipping at scale to leading domestic logic and memory foundries, reaching an installed monthly capacity of one million wafers. Yet establishing strategic supply security does not guarantee financial profitability. For equity investors, strategic shipment volume and economic profit represent fundamentally different metrics. An enterprise can maximize domestic volume while generating sustained financial losses across a prolonged investment cycle. The analytical discipline requires tracking whether return on invested capital moves toward the cost of capital over time, evaluating capital efficiency rather than raw unit expansion.

Lesson two: small, capability-focused cross-border deals age better than they look—but they still age. The Okmetic acquisition initially avoided common pitfalls of outbound technology transactions: it was priced reasonably relative to earnings, secured regulatory approval in an open jurisdiction, retained local management, and provided established customer relationships alongside technical credibility. However, nine years after the acquisition, the asset incurred goodwill impairments, with management flagging potential future write-downs as niche end-markets experienced cyclical downturns. Operational integration and long-term investment returns represent separate benchmarks. Cross-border acquisitions must be evaluated on sustained asset carrying value over a decade rather than initial operational integration.

Lesson three: executive leadership operates within structural cost constraints. Transitioning from research-oriented founders to an experienced manufacturing executive was an operational necessity. Under Dr. Chiu Tzu-Yin, NSIG advanced foundry qualifications, improved yield rates, and expanded shipment volumes. However, operational discipline alone cannot overcome severe structural headwinds. An executive team can optimize plant utilization, reduce defect rates, and retain customer allocations, but management cannot alter fixed equipment depreciation schedules, rationalize a fragmented domestic market, or create pricing leverage in an oversupplied industry. When assessing executive leadership in capital-intensive sectors, the decisive factor is whether management controls the specific variables that dictate financial performance.

Lesson four: check who is buying. The most revealing signals regarding a state-supported enterprise frequently appear in the shareholder register rather than public annual reports. When a founding state fund consistently reduces its equity stake during a prolonged loss cycle, that divestment provides critical context. While capital reductions do not undermine the enterprise's underlying strategic mission, they clarify which stakeholders bear the financial burden of multi-year capacity expansion.

---

↑ Back to Top

XI. Epilogue, Key Metrics to Watch & Outro

Return to that robotic arm lifting a mirror-polished silicon disc out of a cassette. In 2014, inside a Chinese fab, that wafer inevitably arrived from Japan, Taiwan, Germany, or South Korea. By 2026, a growing share originates in Shanghai and Taiyuan. Measured against the policy mandate NSIG was established to execute in December 2015, the strategic mission is substantially accomplished.

Measured against shareholder returns, however, it is not. Over the three most recent full fiscal years, revenue expanded from CN¥3.190 billion to CN¥3.716 billion, while net income deteriorated from a subsidy-supported profit of CN¥186.5 million to a loss of CN¥1.508 billion, with underlying operational losses wider still.²¹⁴ Consolidated gross margin turned negative. Goodwill across both primary specialty acquisitions was written down or fully written off. The Big Fund reduced its equity position, and market regulators issued a formal inquiry letter demanding explanations.

This downturn does not guarantee a permanent failure. Capital-intensive materials enterprises frequently inflect when manufacturing volume and fixed depreciation schedules cross, and NSIG's first-half 2026 shift to positive operating cash flow during peak construction provides an initial signal of potential stabilization. Yet an operational inflection remains a hypothesis subject to a clear empirical test: management's projected 8.5% breakeven gross margin and 2026 profitability target for its consolidated 300mm assets.²⁴ The falsification condition is equally straightforward: if 2027 arrives with monthly 300mm capacity reaching 1.2 million wafers and high plant utilization, yet gross margins remain negative, the deficit will prove structural rather than cyclical—confirming that aggressive domestic overcapacity, rather than temporary depreciation schedules, remained the binding constraint.

Evaluating that outcome depends on three core metrics:

One: the 300mm gross margin. This line item captures the fundamental investment thesis, providing a direct measurement of whether fixed-cost absorption is overcoming unit price erosion. Management has established the primary benchmark: reaching and sustaining an 8.5% gross margin.

Two: 300mm shipment volume paired with realized average selling price. Unit volume expansion alone is an incomplete metric; NSIG expanded shipments by 27% in 2025 and over 90% in the first half of 2026, yet revenue lagged and gross margins contracted. Volume and pricing must be evaluated together to determine whether 2026 global wafer price increases improve NSIG's realized average selling price or whether domestic competition absorbs those gains, while also monitoring how quickly installed nameplate capacity translates into actual customer shipments.

Three: net profit excluding non-recurring items (扣非后归母净利润). This figure represents the true test of un-subsidized commercial viability. Non-operating government grants represented the difference between reported net profit and an operational loss in 2023 and have altered headline figures by hundreds of millions of renminbi in subsequent years.²¹⁴ Commercial viability must be demonstrated on this adjusted line item rather than headline figures.

NSIG built the strategic material foundation that China's semiconductor policy required. Whether it built a commercially self-sustaining enterprise is the question the coming quarters will decide—and management has provided the precise metrics by which that outcome will be judged.

---

References

1. National Silicon Industry Group Co., Ltd. 2025 Annual Report (688126) — Shanghai Stock Exchange / CNINFO, 2026-04-17 ↩

2. 沪硅产业上半年营收同比增长36.51% 300mm硅片销量同比增超九成 — 财联社 Cailianpress, 2026-08-19 ↩↩↩↩↩↩↩↩↩↩↩↩↩

3. 沪硅产业：2025年实现营收37.16亿元 300mm半导体硅片全年销量同比增长27.01% — 中国证券报 China Securities Journal, 2026-04-17 ↩↩↩↩↩↩

4. 沪硅产业2025年亏损扩大至15.1亿元 营收同比增9.7% — 新浪财经 Sina Finance, 2026-04-16 ↩↩↩↩↩↩↩↩↩

5. 沪硅产业回复监管问询 详解连续亏损及资产减值原因 — 新浪财经 Sina Finance, 2026-07-28 ↩↩↩↩↩↩↩↩↩↩↩↩

6. 半导体硅片：2025全球市场份额和排名、总体销售和需求预测 — 格隆汇 Gelonghui, 2025 ↩↩↩

7. 硅产业集团 — 公司简介 / About Us, National Silicon Industry Group official site ↩↩↩

8. 上海硅产业集团股份有限公司关于拟减持其他权益工具投资的公告（公告编号：2025-027）— CNINFO, 2025-04-26 ↩↩

9. National Silicon Industry Group Announces a Voluntary Recommended Public Tender Offer for All Shares and Option Rights in Okmetic Oyj — GlobeNewswire, 2016-04-01 ↩↩

10. National Silicon Industry Group (NSIG)'s Acquisition of Okmetic Oyj — Moelis & Company ↩

11. National Silicon Industry Group to Submit a Voluntary Public Offer to the Shareholders and Warrant Holders of Topsil Semiconductor Materials A/S — GlobeNewswire, 2016-06-17 ↩

12. Soitec与新傲科技宣布加强合作关系，扩大中国区200mm SOI晶圆产量 — 半导体芯科技 SiSC Magazine ↩↩

13. 上海硅产业集团股份有限公司关于2025年年度报告的信息披露监管问询函的回复公告 — 搜狐 Sohu, 2026-07-29 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

14. 沪硅产业2026年半年度报告 — 证券之星 Stockstar, 2026-08-19 ↩↩↩↩↩↩↩↩↩

15. 邱慈云任职沪硅产业总裁 — 全球半导体观察 DRAMeXchange China, 2020-04-28 ↩↩↩

16. 沪硅产业科创板上市 — China Law Insight, 2020-07 ↩↩↩

17. 沪硅产业登陆科创板，大陆首家实现300mm大硅片规模化销售 — 电子工程专辑 EET China, 2020-04-21 ↩↩↩

18. 沪硅产业登陆科创板 国产大硅片领军者释放产业效益 — 证券时报 Securities Times, 2020-04 ↩

19. 上海硅产业集团股份有限公司2025年年度报告摘要 — 上海证券报 Shanghai Securities News, 2026-04-17 ↩↩↩↩↩

20. 上海新昇12吋硅片出货已超340万片！12吋SOI衬底已实现自主可控 — 腾讯新闻 Tencent News, 2021-09-25 ↩

21. 上海硅产业集团股份有限公司2023年年度报告 — CNINFO, 2024-04-15 ↩↩↩↩↩

22. 沪硅产业2026年一季报解读：营收增35.22% 研发投入涨79.81% 亏损规模翻倍 — 新浪财经 Sina Finance, 2026-04-30 ↩↩↩

23. 沪硅产业重回千亿市值，半导体硅片行业机会来了？ — 同花顺财经 10jqka, 2026-06-25 ↩↩↩↩↩↩↩

24. 中国国际金融股份有限公司关于上海证券交易所《关于上海硅产业集团股份有限公司发行股份及支付现金购买资产并募集配套资金申请的审核问询函》回复 — Shanghai Stock Exchange, 2025-09-01 ↩↩↩↩

25. 上海硅产业集团股份有限公司关于完成董事会换届选举及聘任高级管理人员、证券事务代表的公告 — 上海证券报 Shanghai Securities News, 2025-08-18 ↩↩

26. 91亿元！沪硅产业牵手太原落地300mm硅片项目 — 21世纪经济报道 21st Century Business Herald, 2024-01-02 ↩↩

27. 超70亿元！沪硅产业拟收购三家亏损子公司股权，自身业绩也遭"滑铁卢" — 腾讯新闻 Tencent News, 2025-05-22 ↩↩↩↩↩

28. 68岁A股公司沪硅产业总裁，获上海公租房认购资格！邱慈云去年薪酬1297.6万 — 新浪财经 Sina Finance, 2024-08-14 ↩

29. 沪硅产业：第二大股东国家大基金5月20日至5月28日减持3305.02万股 — 财联社 Cailianpress, 2026 ↩↩↩

30. 大基金拟减持不超3%股份，半导体硅片龙头跌超8% — 观察者网 Guancha, 2023-06-08 ↩↩↩

31. 大基金拟减持沪硅产业 — 21世纪经济报道 21st Century Business Herald, 2026-07-08 ↩↩

32. 豪掷132亿元！沪硅产业扩产300mm半导体硅片 — 财联社 Cailianpress ↩

33. 西安奕材登陆科创板：新政后"硬科技上市第一股"，开启全球硅片竞争新局 — 每日经济新闻 National Business Daily, 2025-10-29 ↩↩↩↩

34. 国产硅片厂商上海超硅冲刺IPO：300mm全球份额不到2%、亏损近40亿 — 观察者网 Guancha, 2026-01-15 ↩↩