Victory Giant Technology: The PCB Backbone of the AI Revolution

I. Introduction & Episode Roadmap (15 min)

On the evening of May 17, 2025, in a private dining room in Taipei, Jensen Huang hosted the executives who physically build Nvidia's computing hardware. Around the table sat the leaders of Taiwanese electronics manufacturing—foundries, chip designers, and server assemblers. Among them was a single mainland Chinese executive: Chen Tao (陈涛), a former soldier from Xinjiang who had taught himself printed circuit board engineering from textbooks in his thirties.¹²

That seating arrangement captured an industrial shift in miniature. Printed circuit boards—the green fiberglass slabs foundational to electronic hardware—are traditionally considered computing's least glamorous component. Long defined by commoditization, the sector features thousands of manufacturers, intense price competition, and gross margins in the teens. Yet by 2026, Victory Giant Technology (HuiZhou) Co., Ltd. (胜宏科技) had emerged as one of only two or three suppliers globally capable of mass-producing the specialized circuit boards required by Nvidia's Blackwell-generation AI racks at volume and at acceptable yield rates.

The financial impact was dramatic. In 2025, Victory Giant reported revenue of ¥19.29 billion, a 79.8% surge year-over-year, while net profit attributable to shareholders jumped 273.5% to ¥4.31 billion.¹ For context, in 2022 the company operated with gross margins near 18%, characteristic of a conventional commodity manufacturer.¹² Three years later, its gross margin expanded to 35.2%, surpassing the margins of most semiconductor packaging firms.³

Capitalizing on that momentum, Victory Giant listed H-shares on the Hong Kong Stock Exchange (02476.HK) on April 21, 2026, pricing at the top of its range at HK$209.88 per share. The stock surged 57.2% on its debut to close at HK$330.00, pushing the company's valuation to approximately HK$315 billion.² After fully exercising the over-allotment option the next day, total proceeds reached roughly HK$23.1 billion—or about US$2.95 billion—making it the largest Hong Kong initial public offering of the year to that date.¹³

Over the following three months, however, the stock fell more than 40% from its late-May peak.¹⁶

This steep decline highlights why Victory Giant's story demands rigorous scrutiny rather than simple celebration. The company represents one of the most remarkable industrial transformations in modern Chinese manufacturing, yet it operates with an unusually concentrated set of risks. A single end-customer likely generates the majority of its high-margin volume; its chairman's personal affairs triggered market volatility; its controlling shareholder sold nearly ¥1.7 billion in stock just months before raising public capital; and its order book for 2027 and 2028 remains contingent on strategic decisions made in Santa Clara rather than Huizhou.

The core question: is Victory Giant's advantage a durable process-power moat, or is it a well-timed position on a cyclical wave that competitors are now spending tens of billions of renminbi to reach?

The roadmap:

1. The Pearl River Delta crucible and the founding years (2003–2014)
2. The ChiNext listing and the multilayer ceiling (2015–2018)
3. The 2019 HDI gamble—the decision that made everything else possible
4. The MFS Technology acquisition and what the regulator asked about it (2023)
5. The AI supercycle and the Nvidia relationship (2024–2026)
6. Product mix, unit economics, and what the company does not disclose
7. Competitive structure through Helmer and Porter
8. Management behavior, governance, and the capital-allocation record
9. Bull case, bear case, and the two or three numbers that actually matter

The story begins where capital was scarce.

Subscribe on LinkedIn

↑ Back to Top

II. Founding & The Pearl River Delta PCB Crucible (2006–2014) (20 min)

The founder's biography does not read like a technology origin story. Chen Tao was born in 1972 and enlisted in the military at seventeen. After his discharge, he took a stable government-adjacent job, but a business trip to Guangdong—the great southern manufacturing sprawl of post-reform China—redirected his life. He moved into sales, worked his way into management within four years through self-study, and in 2003 founded Shenghua Electronics (胜华电子), the direct predecessor of Victory Giant.⁵

The most revealing detail from that period is that Chen had neither the technical background to design PCB processes nor the capital to hire people who did. He studied for and obtained professional engineering certification himself, eventually qualifying as a senior PCB engineer.⁵ This was not a trivial biographical flourish. Printed circuit board manufacturing is a chemistry-and-tolerances business—copper plating baths, drill registration, lamination pressure profiles—where a founder's ability to personally argue with process engineers about scrap rates determines whether a factory makes money. Chen's later willingness to bet the balance sheet on a manufacturing process most of his peers considered too difficult becomes far clearer given his ability to read process sheets.

Victory Giant Technology (胜宏科技) was formally established in Huizhou, Guangdong, in 2006.⁵ Huizhou sits on the eastern arc of the Pearl River Delta, an hour and a half from Shenzhen—close enough to the electronics supply chain to recruit engineers, but far enough out that land and labor were cheap. In the mid-2000s, this region held the densest concentration of electronics manufacturing capacity in human history, and its PCB tier was among the most fragmented. Thousands of shops competed with almost no differentiation.

Understanding the physics of a circuit board explains why everything downstream matters. A manufacturer starts with fiberglass cloth soaked in epoxy resin, laminates copper foil onto both faces, and produces copper-clad laminate—the core raw material. Etching away unwanted copper creates a wiring pattern. Stacking several of these sheets with insulating layers between them, pressing them together under high heat and pressure, and drilling holes plated with copper connects one layer's wiring to another. That creates a multilayer PCB. In 2006, a four-to-eight-layer board of this type was a genuine commodity: any competent shop could produce one, leaving price and delivery speed as the primary competitive variables.

The economics were correspondingly grim. Gross margins across the Chinese PCB industry in that era clustered in the low-to-mid teens. Input costs—copper foil, fiberglass cloth, epoxy resin—were exposed to commodity price swings that manufacturers could rarely pass through to customers. A shop that won business on price and then faced a spike in copper prices could easily lose money on a full order book.

This unforgiving industry structure stems from three characteristics that combine poorly. First, PCB manufacturing is capital-intensive, requiring fixed costs to be covered regardless of order volume. Second, it is technically legible: the underlying processes are well-documented, equipment is sold openly by Japanese and German vendors, and any competitor with sufficient capital can replicate a given specification within a few years. Third, manufacturers sell to buyers—such as set-top box makers, monitor assemblers, and appliance brands—who operate on thin margins themselves and pass every cent of price pressure upstream. Capital intensity, high replicability, and powerful buyers create a textbook recipe for an industry that grows dramatically while generating meager economic returns for its participants.

The only exit from this trap is migrating to specifications that competitors cannot economically match. Victory Giant eventually executed that shift, but doing so required eight years of accumulating operational discipline before the company was capable of attempting it. The unglamorous founding era was not a blank space in the company's history; it was the prerequisite foundation.

Victory Giant's early differentiation was mundane and expensive: it kept buying advanced equipment. Where many contract shops sweated aging assets and competed purely on labor costs, Victory Giant invested heavily in automated production lines and process controls. That strategy depressed early capital returns and did not create an immediate economic moat—plenty of Chinese manufacturers over-leveraged during this period and went bankrupt. However, it built an organizational discipline for running capital-intensive, tightly specified production, establishing the operational foundation required for its later high-density interconnect (HDI) business.

By 2011, the company had established a district-level engineering research center. Through the early 2010s, its customer mix shifted from fragmented, small-batch orders toward higher-volume contracts with major display and computing-hardware manufacturers.⁵ The strategic value lay in operational discipline rather than sheer scale: high-volume contracts with demanding original equipment manufacturers force a factory to maintain yield and delivery precision within narrow tolerances month after month. Companies that master that transition build durable operational capabilities; those that fail disappear.

What this period means for an investor: virtually nothing in Victory Giant's first decade predicted its eventual ascension in AI hardware. It operated as a competent, capital-hungry, sub-scale participant in a commodity market, distinguished mainly by a founder who understood process technology and persistently reinvested in machinery. The primary insight from this era is negative rather than positive: the competitive moat that exists today was not visible in the business model, market position, or margins of 2014. It had to be acquired, and the critical purchase order would not be signed until 2019.

First, however, the company needed access to public capital markets.

↑ Back to Top

III. Going Public & The Multilayer Bottleneck (2015–2018) (20 min)

In June 2015, Chinese equity markets were in the terminal phase of one of the most spectacular retail-driven bubbles in modern financial history. The Shanghai Composite peaked that month before losing roughly a third of its value in three weeks. Into that volatile window, on June 11, 2015, Victory Giant listed on the ChiNext board of the Shenzhen Stock Exchange under ticker 300476, pricing 36.67 million shares at ¥15.73 each and raising gross proceeds of roughly ¥577 million.⁶

By the standards of what followed—most notably its 2026 Hong Kong offering that was nearly forty times larger—this initial raise was modest. Yet it proved transformative for the Huizhou manufacturer, offering the first clear test of management's capital-allocation priorities.

Victory Giant deployed the capital into expanding capacity, building an engineering research center, and constructing what it branded a "smart factory"—a digitized, heavily automated campus in Huizhou that reached full production in 2018.⁵ The timing of these investments was telling. Between 2015 and 2018, mainstream Chinese PCB manufacturers primarily targeted the mobile supply chain, pouring capital into smartphone flexible circuits and LED lighting boards. Victory Giant instead directed its resources toward factory automation and higher-layer-count rigid boards for telecom and server applications.

This capital deployment reflected a pragmatic response to structural industry dynamics rather than pure foresight. Traditional multilayer PCB manufacturing features a modest economic ceiling. Because the underlying processes are standardized and production equipment is commercially available, pricing power rests with system integrators and device assemblers rather than board vendors. A manufacturer can expand revenue by adding capacity, but each incremental square meter yields low-margin returns. Compounding this challenge, key end markets were maturing: global personal computer shipments had plateaued, and consumer electronics suppliers faced relentless annual price reductions from buyers.

By 2017, Chen Tao faced a strategic choice common to industrial suppliers: continue competing inside a structurally low-margin commodity market, or commit substantial capital to transition up the technology curve.

The company initiated its shift in 2017, committing roughly ¥1 billion toward new-energy vehicle and IoT-related applications while establishing formal technical research into high-density interconnect technology.⁵ At the time, HDI fabrication was dominated by established Taiwanese and Japanese manufacturers, including Unimicron (欣興電子) and Compeq (華通電腦). Chinese PCB firms functioned largely as consumers of HDI design standards rather than high-yield volume producers.

The core investor lesson from this phase lies in what steady top-line growth concealed. Between its 2015 IPO and 2018, Victory Giant reported solid operational metrics: revenue expanded and production facilities modernized without showing financial strain. However, the business was continuing to accumulate assets in product categories where long-term returns on incremental capital were fundamentally constrained. A financial analysis focused strictly on short-term income statement growth would have seen a steady mid-tier manufacturer; a structural analysis of the industry revealed a company approaching an economic wall.

Management appeared to recognize that limit. The aggressive strategy the company adopted in 2019 appeared, at the time, to be a high-risk gamble.

↑ Back to Top

IV. The Great Pivot: High-Stage HDI & The AI Infrastructure Bet (2019–2022) (25 min)

In August 2019, Victory Giant's high-density interconnect (HDI) division began production. Phase one opened with a monthly capacity of roughly 45,000 square meters, averaging about three stages.¹² Understanding why this constituted a high-stakes gamble rather than a routine expansion requires clarifying what "stages" signify in circuit board manufacturing.

A conventional multilayer board functions like an apartment building where every floor is connected by a single elevator shaft running top to bottom. A manufacturer drills one hole through the entire stack and plates it with copper. The process is inexpensive and reliable, but inefficient: that vertical shaft consumes valuable surface area on every level regardless of whether that floor requires a connection.

High-density interconnect replaces that single shaft with a network of tiny local staircases. A laser burns a microscopic hole—a micro-via, often under 100 microns wide, thinner than a human hair—that connects only layer one to layer two. A subsequent lamination cycle adds more layers, and another set of micro-vias connects those. Each of these build-up cycles represents one "stage." A six-stage board undergoes six sequential rounds of lamination, drilling, plating, and imaging; a ten-stage board undergoes ten.

The economic challenge lies in compounding yield loss. Every stage must align its micro-vias onto conductive pads from the previous stage with registration accuracy measured in single-digit microns—all on a laminate sheet that expands and contracts under heat and pressure. Misalignment breaks the connection. Crucially, a defect introduced at stage nine destroys all value added in stages one through eight. Because yield loss compounds multiplicatively, a ten-stage process running at a 98% yield per stage results in an overall board yield of roughly 82% before final testing. At a 95% yield per stage, overall production yield drops to 60%. Consequently, mastering high-stage-count HDI is not merely a purchasing decision; a manufacturer cannot buy its way to volume production. It is a steep learning curve paid for in scrapped panels.

Victory Giant absorbed those learning costs between 2019 and 2022. Capital spending targeted laser drilling systems, laser direct imaging exposure equipment, and vacuum lamination presses—the most expensive categories of PCB capital equipment. Annual R&D expenses climbed from ¥221 million in 2020 toward ¥450 million by 2024.¹² Meanwhile, financial performance reflected the drag of these investments: gross margin stood at roughly 18.1% in 2022, essentially unchanged from commodity-industry norms despite years of heavy capital deployment.¹²

For three years, the initiative appeared to be a failing bet. Capital had been committed, depreciation expenses were mounting, and gross margins remained stagnant. Chinese retail investors, who dominate ChiNext trading, were not rewarded for patience. Consumer demand was soft, the smartphone cycle was decelerating, and the ultimate payoff market—hyperscale AI accelerators—did not yet exist at scale.

What the company accumulated during those years was not immediate margin expansion, but customer qualification. In high-performance computing, a PCB supplier cannot simply submit a commercial bid. The board functions as an active element of the electrical system: at the signal speeds required for modern GPU interconnects, a copper trace acts not as a simple wire but as a transmission line. Its dielectric properties, surface roughness, and impedance consistency dictate whether data arrives intact. Securing inclusion in a reference platform requires months of joint electrical characterization alongside silicon vendors and system integrators. Victory Giant spent 2019 through 2022 building those technical relationships and qualifying its processes—work that generates no upfront line item on an income statement.

By the time the AI buildout arrived, the company had crossed a threshold that mattered more than any single product win: it had become one of a small group of manufacturers globally with credible mass-production capability in six-stage-plus HDI. The company stated it was among the first globally to mass-produce six-stage 24-layer HDI, and that it possesses eight-stage 28-layer and 16-layer any-layer HDI capability.⁷

There is a second, subtler asset that a manufacturer acquires during an extended yield ramp: tacit process knowledge. Yield improvement at high stage counts is rarely driven by a single technological breakthrough. Instead, it relies on thousands of minor operational adjustments, most of which are never formally documented—such as identifying which lamination press runs half a degree hot, which laser drill head drifts after nine hours of operation, which supplier's laminate lot expands differently along the machine direction, or which panel-handling technique prevents micro-scratches that cause electrical opens three stages later. This knowledge resides with line engineers and within the calibration culture of a specific factory floor. A competitor can purchase identical equipment from identical vendors and still operate at half the yield for two years because the machinery is relatively accessible, whereas process refinement is costly and time-intensive. This dynamic creates genuine process power—though it also dictates its eventual erosion, as competitors with sufficient production volume and scrapped panels will eventually accumulate similar refinements.

The other notable aspect of this period is what management chose not to pursue. Victory Giant did not attempt to diversify into downstream assembly, acquire its way into advanced technology, or convert into an integrated device manufacturer. The 2019 mandate remained narrowly focused: excel at one specific, complex manufacturing process. That narrow focus enabled its technical success, while simultaneously creating the customer concentration risk that characterizes the business today.

The analytical conclusion here is central to the overall investment case. The 2019 pivot demonstrates that management was willing to accept three years of depressed returns and shareholder skepticism to secure a position in high-complexity manufacturing. That commitment reflects genuine capital-allocation discipline and represents the single strongest operational evidence in the company's favor. However, this history should be evaluated objectively rather than mythologized. No evidence suggests Victory Giant anticipated the generative AI boom in 2019; the stated rationale focused on broad high-end market positioning across smartphones and automotive applications. The company effectively purchased an option on process complexity, and the AI supercycle provided the unexpected payoff that made that option immensely valuable. Strategic courage and predictive foresight are distinct attributes, and only one of them is repeatable.

Before that payoff materialized, management executed a second, far more debatable transaction—this time committing cash across an international border.

↑ Back to Top

V. Cross-Border M&A: The MFS Technology Acquisition (2023) (20 min)

In July 2023, Victory Giant announced an all-cash transaction to acquire 100% of a holding company that owned Singapore-headquartered MFS Technology (S) Pte Ltd and its subsidiaries. The deal carried a total cost not exceeding $460 million, comprising roughly $365 million in equity value plus up to approximately $95 million in debt and transaction-related expenses.⁸

The strategic logic was straightforward. Victory Giant was primarily a rigid-board manufacturer, whereas MFS brought flexible printed circuits (FPCs)—the bendable ribbon circuits used in camera modules, battery management systems, and folding hinges—alongside rigid-flex hybrids. In 2022, MFS generated approximately $261 million in revenue, with $197 million derived from FPC products and $64.7 million from PCB products.⁸ The group operated four manufacturing facilities: one in Malaysia and three across Changsha and Yiyang in mainland China.⁸ Management framed the deal as a means to accelerate diversification into automotive and medical end markets.⁹

An important distinction clarifies a widely repeated narrative regarding this transaction: the MFS footprint acquired in 2023 included only the Malaysian and mainland Chinese facilities. Victory Giant's manufacturing projects in Vietnam and Thailand were separate, later, greenfield developments funded by domestic equity raises—a distinction that matters when evaluating how much geopolitical supply-chain diversification the acquisition actually purchased.²⁵

The Shenzhen Stock Exchange reacted with skepticism. Regulators issued a formal letter of concern demanding that Victory Giant justify the deal structure, prompting domestic media to question whether the company was paying $460 million for a holding vehicle without substantive direct operations. Victory Giant responded by clarifying that the target entity held 100% of MFS and reiterating that the transaction advanced its diversification strategy.⁹ Following these disclosures, the relevant board resolutions and transaction filings were completed through the exchange in December 2023.¹⁰

An objective valuation assessment reveals a more nuanced picture than standard deal summaries suggest. Victory Giant did not disclose the EV/EBITDA multiple for the transaction, making any published profit multiple speculative. However, the top-line multiple is easily calculated: a total cost of roughly $460 million against $261 million in trailing revenue equates to approximately 1.8 times sales. For a flexible-circuit business—a sector defined by structurally lower gross margins than high-end rigid HDI and a history of severe customer concentration in mobile programs—that was a full valuation, though not an unreasonable price.

The more useful analytical question is not simply whether Victory Giant overpaid, but what the transaction actually delivered over the subsequent three years:

What worked. The acquisition gave Victory Giant immediate flexible-circuit capabilities, a functional manufacturing base in Malaysia, and an international corporate architecture—including a Singapore subsidiary structure, cross-border financing channels, and overseas management. That global plumbing proved valuable when Victory Giant later expanded into Thailand. When the company injected $250 million into its Thai facility, the capital flowed through its Singapore subsidiary.¹⁵ That corporate framework provides tangible option value as Western customers increasingly demand supply chains outside mainland China.

What remains unproven. Victory Giant has not provided public disclosures isolating MFS's contribution to group profit, and flexible circuits were not the primary driver of the company's 2024–2026 financial expansion. That earnings surge stemmed overwhelmingly from high-layer-count rigid boards and high-stage HDI required for AI infrastructure. Connecting the MFS purchase to subsequent financial performance confuses correlation with causation, while market claims that the deal became accretive within twelve months remain unverified by disclosed figures.

What it cost. The transaction removed nearly half a billion dollars in cash from the balance sheet just as the company needed substantial capital for advanced HDI capacity. By 2025, financial leverage had increased noticeably: Victory Giant's debt-to-asset ratio stood at 53.8%, with combined short- and long-term borrowings of roughly ¥3.9 billion against under ¥1.2 billion in cash prior to subsequent capital raises.¹² This capital outlay contributed directly to the company's subsequent equity raises, including a ¥1.9 billion domestic placement and the HK$23 billion Hong Kong initial public offering in 2026.

For investors, the MFS transaction should be evaluated as a defensible strategic purchase of international infrastructure and product optionality rather than a proven driver of high-return growth. The deal illustrates management's willingness to execute cross-border acquisitions and operate with elevated balance-sheet leverage—factors that create both strategic options and financial risk.

Then the world changed.

↑ Back to Top

VI. The AI Supercycle: Becoming Nvidia's Tier-1 Nerve Center (2024–2026) (25 min)

The mechanism by which generative AI transformed a Huizhou circuit board factory is worth spelling out precisely, because it is not simply "more computers were sold."

A traditional data centre server is a pizza-box chassis with a motherboard, some memory, and a network card. An AI rack in the Blackwell generation is a different object: dozens of accelerators in a liquid-cooled cabinet, all of them wired to each other through a switching fabric running at speeds where the physical board becomes an active participant in signal integrity. The accelerator modules themselves need boards with extreme layer counts and stacked micro-vias to route thousands of connections out of a single package. The switch trays need ultra-high-layer-count boards with tightly controlled impedance. And the whole rack needs backplanes that carry those signals between trays without degradation.

The result is that PCB content per unit of compute rose sharply — not incrementally, but by an order of magnitude versus conventional servers. And critically, the boards required sat at exactly the point on the difficulty curve that Victory Giant had spent 2019–2022 climbing.

Consider the shape of the change more concretely. In a conventional two-socket server, the motherboard is a large but relatively ordinary multilayer board, and the value of PCB content in the whole machine is modest — a low-hundreds-of-dollars item in a machine costing several thousand. In a dense liquid-cooled AI cabinet, the picture inverts. There are accelerator boards, one per GPU module, each of them a high-stage HDI part. There are baseboards that host several accelerators. There are switch trays whose boards must carry differential pairs at frequencies where a few microns of copper surface roughness measurably degrades the signal. There are backplanes and mid-planes connecting the trays. Multiply by dozens of compute modules per cabinet, and PCB content per rack moves from a rounding error to a line item that hyperscale buyers negotiate directly. Market analysis has put the PCB value of a single Blackwell-generation accelerator board in the region of ¥12,000.¹²

This is why the AI buildout was not merely "good for PCB companies." It was good for a very specific subset: those able to hold micron-level registration across many sequential build-up cycles on large-format panels. Every other PCB manufacturer in the world experienced the AI boom as a story about someone else's order book.

The timeline of the relationship, as reported, ran roughly as follows: Victory Giant entered Nvidia's H-series AI accelerator supply chain in 2023; passed the qualification associated with the GB-generation platform in the fourth quarter of 2024; and operated as a Tier-1 supplier through 2025.¹² The company has also been reported as co-developing dynamic impedance techniques with the customer to optimise power delivery and thermal performance on accelerator boards.¹² Readers should note that these operational details come from market reporting rather than company disclosure — Victory Giant, like most component suppliers, does not name its largest customers in filings.

The financial trajectory tells the story without needing the customer's name:

2024 was the year the inflection became visible in the P&L. Revenue reached ¥10.73 billion, up 35.3%, with net profit of ¥1.15 billion, up 72.0% — profit growing at twice the rate of revenue, which is the signature of mix shift rather than volume growth.¹¹²⁶

2025 was the year it became extraordinary. First-half revenue of ¥9.03 billion grew 86%, while net profit of ¥2.14 billion grew 367%.¹² Full-year revenue of ¥19.29 billion and net profit of ¥4.31 billion followed, with basic EPS of ¥5.01 and R&D spending of ¥778 million across 87 development projects.¹ Operating cash flow reached ¥4.60 billion, up 238.9% — an important detail, because it means the profit was converting to cash rather than accumulating as receivables.³

But the single most revealing line in the 2025 annual report is one that most summaries skipped. Physical volume went down. Sales volume was 8.664 million square metres, a decline of 2.72%, and production volume was 8.090 million square metres, down 9.63%.³ Revenue nearly doubled on less physical output.

That is the entire thesis in one data point. Victory Giant did not grow by making more circuit board; it grew by making dramatically more valuable circuit board on the same physical base. It is the cleanest available evidence that the mix shift to high-stage HDI and ultra-high-layer products is real and not a story. It also implies enormous revenue per square metre expansion, and it explains how gross margin moved from the commodity teens to 35.2%.³

Two other disclosures from that report deserve attention. Direct export revenue was ¥14.82 billion, or 76.8% of total revenue, up 126.9% year over year, versus 60.9% of revenue the prior year — while domestic sales grew only 6.5%.³ This company's growth is now almost entirely an overseas-demand phenomenon, which is both its strength and its geopolitical exposure. And the top five customers accounted for ¥8.10 billion, or 41.98% of revenue, with the largest at 14.97% and the second at 12.56%.³

That customer concentration figure requires careful interpretation, and this is where an investor should slow down. A 42% top-five concentration sounds moderate. But AI accelerator boards do not ship to the chip designer — they ship to contract manufacturers and ODMs who build the systems. If several of the disclosed top-five "customers" are different assemblers building racks for the same end platform, then the disclosed concentration materially understates the true dependency. Market reporting through 2025 suggested the Nvidia-linked share of the company's AI board business exceeded 50%, and some estimates put it far higher.¹⁶ The company does not disclose end-platform revenue. This gap between disclosed customer concentration and true end-demand concentration is the single most important unmeasurable in the story.

The market has already priced that fragility twice. On November 17, 2025, a report that Nvidia had shifted roughly 20% of previously sole-sourced HDI board orders to competitor 兴森科技 Xingsen Technology sent the stock down over 5% intraday, with the board secretary responding publicly to dispute the account.²⁴ And in July 2026, unsubstantiated reports of order losses triggered an 8.32% single-day decline.¹⁶ A company whose share price moves 8% on a rumour about one customer is telling you exactly what its risk profile is, regardless of what the concentration table says.

The capital markets response to all this was to raise as much money as possible while the window was open. A domestic private placement of roughly ¥1.9 billion was completed in late October 2025 at ¥248.02 per share, funding the Vietnam AI HDI project and the Thailand high-multilayer project.²⁵²⁴ Then came the Hong Kong listing: 37 to 38 cornerstone investors subscribing close to US$1 billion, including 云锋资本 Yunfeng Capital, CPE 源峰 CPE, Morgan Stanley and 高瓴 Hillhouse Capital, with proceeds earmarked roughly 74% for mainland capacity and intelligent manufacturing equipment, 7% for mSAP equipment and product diversification, 9% for R&D, and 10% for working capital.²⁵ The stabilisation period ended in May 2026.²⁷

Per Prismark, the company ranked sixth globally among PCB suppliers and third among mainland Chinese manufacturers in 2025.¹²⁸ Per Frost & Sullivan data cited in the Hong Kong offering, it held 13.8% of the global AI and high-performance computing PCB market by first-half 2025 revenue — first worldwide in that niche.²

The dual-listing decision deserves a moment of investor scrutiny, because a ChiNext-listed company with a soaring share price did not need Hong Kong for capital alone. Three motives are visible in the structure. First, currency: an H-share listing produces Hong Kong dollars that can fund Thai, Vietnamese, and Malaysian construction without navigating outbound capital controls — a practical point that matters enormously to a company committing to an overseas build. Second, shareholder base: the cornerstone roster was deliberately international and institutional, which changes who the company answers to and creates pressure toward better disclosure over time. Third, and least discussed, credibility with customers: a Western hyperscaler evaluating a Chinese supplier for multi-year sole-source content finds an internationally listed, internationally audited counterparty easier to approve than a purely domestic one. The listing was, in part, a sales instrument.

The cost of that instrument was dilution at a moment when insiders had recently been sellers — a tension examined in the governance section below.

So what for investors: the 2024–2026 results are not an accounting illusion — declining physical volume alongside doubling revenue, plus strong cash conversion, is hard to fake. What remains genuinely open is whether this is a position or a franchise. The next section examines what the company actually sells, and what it declines to tell you about it.

↑ Back to Top

VII. Segment Breakdown, Product Mix, & Industrial Anatomy (20 min)

A walk through Victory Giant's Huizhou campus reveals four distinct business lines operating side by side, ordered by manufacturing complexity:

The crown jewel: high-stage HDI. These boards stack micro-vias across multiple build-up cycles. Victory Giant states it was among the first globally to mass-produce six-stage 24-layer HDI, holds eight-stage 28-layer and 16-layer any-layer capability, and reached ten-stage 30-layer capability.¹⁷ It has also disclosed development work toward 14-stage 36-layer HDI.³ Targeted at AI accelerator modules, high-speed optical transceivers, and flagship smartphone mainboards, these high-density products drive the company's incremental margin expansion.

Ultra-high-layer-count rigid boards. The company has stated manufacturing capability above 100 layers.³ These large, mechanically complex panels serve as switch trays, backplanes, and baseboards in AI server racks—carrying substantial value per panel because a single manufacturing defect at this scale ruins an entire board's accumulated processing value.

Flexible and rigid-flex circuits (via MFS). Acquired in 2023, the MFS product line produces bendable circuits for automotive battery management systems, display modules, and industrial and medical devices, carrying structurally lower gross margins than high-end rigid AI boards.

Legacy multilayer boards. Standard four-to-eight-layer circuit boards for consumer electronics, personal computers, and general automotive controls. This segment serves as baseload production to maintain factory utilization across legacy lines.

One allocation in the 2026 Hong Kong offering prospectus highlights management's next technical target: 7% of net proceeds were designated for modified semi-additive process (mSAP) equipment.² Unlike conventional subtractive manufacturing—which etches away unwanted copper from a solid sheet and risks undercutting traces—mSAP builds copper traces upward, producing narrower lines with straight, precise sidewalls. Originally developed for high-end smartphone mainboards, mSAP forms the technological bridge between printed circuit board manufacturing and advanced semiconductor substrate packaging. Reinvesting in mSAP represents a deliberate effort to migrate toward higher-margin semiconductor-adjacent processes rather than defend legacy turf. However, this strategy carries significant execution risk and capital expense, placing Victory Giant in direct competition with established Japanese and Taiwanese substrate leaders.

Investors examining the business must navigate a key transparency constraint: Victory Giant does not publish a clean revenue split by product stage or AI end-market in its public filings. Widely cited metrics—such as claims that AI accounts for 45% of total revenue or that HDI margins reach 35% to 45%—represent analyst estimates and media projections rather than audited disclosures. The company discloses only composite figures: a 35.2% group gross margin in 2025, a 76.8% export revenue share, and the previously noted divergence between falling physical volume and surging revenue.³ Third-party market reporting indicates that HDI grew to roughly 40% of total revenue by early 2025 and was expected to remain the primary driver of top-line growth.¹¹

The company's cost disclosures reveal mounting margin sensitivity to raw materials. Direct material inputs rose to 65.9% of cost of goods sold in 2025, up from 62.7% in 2024.³ Copper-clad laminate represents the primary raw input, with the high-frequency, low-loss grades required for AI acceleration supplied by a concentrated group of vendors, including Shengyi Technology (生益科技), Kingboard Group (建滔集团), and specialized Western manufacturers. Rising laminate prices squeeze gross margins unless Victory Giant can pass those increases downstream—a difficult task when negotiating with powerful hyperscale buyers. This friction surfaced in the first quarter of 2026: gross margin slipped to roughly 34.5%, down from 35.2% for full-year 2025, with management and sell-side analysts attributing the compression to upstream material inflation.¹⁶

Two additional balance-sheet items from 2025 warrant careful observation. Finished inventory expanded 12.6% to 857,800 square meters during a period when total physical production fell, creating an ambiguous signal: it reflects either inventory pre-positioning for scheduled 2026 deliveries or an accumulation of unabsorbed output.³ Meanwhile, financial expenses surged 467.1% in 2025, driven by currency fluctuations on a business that generates over three-quarters of its revenue overseas.³ Foreign exchange volatility has evolved into a material earnings driver rather than a secondary accounting detail; in the first quarter of 2026, financial expenses rose another 265.7% year-over-year.¹⁴

Management's capital deployment plan underscores its aggressive growth targets. Victory Giant outlined a 2026 capital investment program of approximately ¥20 billion, including roughly ¥18 billion dedicated to fixed asset expansion.³ In practical terms, the company plans to spend more than four times its total 2025 net profit on new facilities and equipment in a single twelve-month period. The geographic layout reflects a multi-region supply-chain strategy: Thailand handles high-end rigid board upgrades via a $250 million capital injection routed through its Singapore subsidiary; Vietnam adds high-stage HDI capacity aimed at North American server and AI clients, with production scheduled for 2026; and Malaysia manages flexible circuit operations through MFS.¹⁵²⁵ Balance-sheet work-in-progress construction reached ¥5.17 billion by the first quarter of 2026, up 43.2% from year-end 2025.¹⁵

So what for investors: Victory Giant's product anatomy confirms the reality of its high-value mix shift, but its corporate disclosures have not kept pace with the business's growing complexity. Shareholders are effectively being asked to underwrite a ¥20 billion capital expenditure expansion without access to audited segment margin tables or dedicated AI revenue reporting. This disclosure gap presents a clear governance challenge, increasing the cost of capital at the precise moment Victory Giant is seeking sustained backing from global institutional investors.

↑ Back to Top

VIII. Helmer's 7 Powers & Porter's 5 Forces Analysis (20 min)

Strip away the narrative and ask the war-game question: if a competitor had unlimited capital and wanted to capture this market position, could it?

Myth vs Reality

Before applying strategic frameworks, four consensus claims about Victory Giant deserve critical evaluation, as each is frequently cited and each is partially inaccurate.

Myth: "Victory Giant foresaw the AI boom in 2019." Reality: the company invested in high-density interconnect (HDI) to escape the commodity multilayer ceiling and serve high-end smartphone, automotive, and networking applications. The strategic rationale documented at the time was general premium positioning, not AI-specific foresight.⁵ This distinction matters because it clarifies what investors are underwriting: not a management team with a repeatable gift for predicting end-market shifts, but one with a repeatable willingness to invest in demanding manufacturing capabilities and absorb long ramp cycles. The evidence supports the latter attribute—a valuable trait, but a distinct one.

Myth: "Victory Giant is the sole supplier of Nvidia's AI accelerator boards." Reality: while the company has been a leading supplier with reported dominant allocations in specific programs, the November 2025 report of order reallocation to 兴森科技 Xingsen Technology—and Victory Giant's public response to it—demonstrates that customer multi-sourcing is active policy rather than a theoretical risk.²⁴

Myth: "The AI margin expansion represents a structural re-rating unique to Victory Giant." Reality: the company's 35.2% group gross margin reflects a product blend and is not materially higher than 沪电股份 Wus Printed Circuit's core PCB gross margin of roughly 36.9% achieved in the same period.³²² High profitability was characteristic of AI-exposed, high-end PCB capacity across 2025 rather than unique to Victory Giant's proprietary process. That margin represents a scarcity rent shared among a small cohort of qualified vendors, and scarcity rents are inevitably competed down over time.

Myth: "The MFS acquisition de-risked Victory Giant's manufacturing footprint away from mainland China." Reality: the MFS acquisition brought assets located in Malaysia and mainland China.⁸ Genuinely non-China expansion in Thailand and Vietnam was initiated later through separately raised equity capital.¹⁵²⁵ The acquisition provided the cross-border corporate framework that facilitated those subsequent projects—a meaningful second-order benefit, but distinct from immediate geographic diversification.

Hamilton Helmer's 7 Powers

Process Power — the central claim, and the critical variable to evaluate. The bullish argument asserts that mass-producing ten-stage, 30-layer HDI boards at commercial yields represents accumulated organizational know-how that competitors cannot simply purchase. Supporting evidence is tangible: industry analysis indicates Victory Giant achieved high-stage HDI yields above 80%, compared with an industry average nearer 50%.¹² If accurate, that gap is decisive: at a 50% yield, high-stage boards are unprofitable at accessible price points; at 80%, they generate exceptional margins. This process advantage aligns with the company's operational pattern of expanding revenue while physical volume declines, indicating value capture beyond standard industry capabilities.

However, process power has a recognized vulnerability: it erodes when competitors are given sufficient capital and time. Competitors now possess both. 深南电路 Shennan Circuits committed approximately ¥17 billion toward AI computing capacity.²⁹ 鹏鼎控股 Avary Holding, the largest Chinese PCB manufacturer by revenue, announced roughly ¥23.3 billion in expansion plans.²³ Meanwhile, 沪电股份 Wus Printed Circuit—whose 2025 profitability matched Victory Giant's—is pursuing a Hong Kong listing to fund its own high-end capacity. A process advantage spanning one to two product generations is real, but it does not constitute a permanent structural moat. The verdict: strong today, but unproven beyond roughly 2028.

Scale Economies — genuine, but not unique. Victory Giant's Huizhou campus is among the largest single-location automated PCB facilities globally, providing raw-material purchasing leverage with laminate suppliers. However, major peers are constructing facilities at comparable scale, while upstream suppliers remain consolidated. Scale in this context lowers unit costs, but it does not grant pricing power.

Switching Costs — real, and the most durable short-term power. Once a circuit board is qualified inside a computing platform's reference design, replacing the supplier requires repeating signal-integrity validation, thermal testing, and reliability certification—a multi-month process funded by the customer that risks delaying hardware generations worth billions of dollars. This creates significant incumbency advantage within a given platform cycle. The key constraint is that AI hardware generations turn over in under two years, reopening the qualification process with each cycle.¹² Switching costs that reset every eighteen months offer a rolling tactical advantage rather than permanent customer lock-in—explaining why market reports of order reallocation triggered severe share-price volatility in November 2025.²⁴

Counter-Positioning — weak, and often mischaracterized. Bullish accounts sometimes claim legacy PCB makers could not pivot into high-stage HDI without cannibalizing existing operations. Evidence contradicts this view: established peers such as Wus, Shennan, and Avary are pivoting aggressively without structural barriers. Victory Giant secured an early lead, not a position protected by incumbent self-cannibalization.

Cornered Resource, Branding, Network Economies — not applicable. The company possesses no proprietary raw inputs, enjoys no brand premium at the component layer, and benefits from no network effects.

Porter's Five Forces

Rivalry: high, and intensifying. The 2025 financial record is instructive. Wus Printed Circuit generated revenue of ¥18.95 billion, up 42%, with net profit rising 47.7% to ¥3.82 billion and a core PCB gross margin of roughly 36.9%—slightly above Victory Giant.²² By contrast, Avary Holding posted ¥39.15 billion in revenue, up 11.4%, with net profit growing 3.3% and a 21.5% gross margin, reflecting its heavy exposure to lower-margin communications boards.²²²³ This comparison underscores that Victory Giant's high margins are not unique across the sector; they are characteristic of the AI-exposed board segment. Industry competition centers not on general manufacturing capability, but on securing allocations within next-generation AI platforms.

Supplier Power: moderate and rising. Direct material inputs reaching 65.9% of cost of goods sold, with the ratio trending upward, demonstrate active supplier pricing power.³ Specialized low-loss copper-clad laminate remains genuinely scarce during hyperscale expansion cycles.

Buyer Power: high. Hyperscale cloud providers and chip designers hold structural advantages: detailed cost models, explicit multi-sourcing mandates, generational design reviews, and the ability to shift volume across suppliers as a discipline mechanism. The November 2025 order-reallocation episode demonstrated that primary buyers actively exercise this leverage.²⁴

Threat of Substitutes: low in the near term, relevant long term. Glass-core substrates represent the primary technology capable of replacing organic high-stage HDI, with commercial adoption generally anticipated in the late 2020s. For current and immediate next-generation platforms, organic laminate remains the standard. While not an immediate threat, material technology shifts have historically disrupted established component vendors.

Threat of New Entrants: low for de novo startups, high for adjacent incumbents. Establishing a high-end PCB operation from scratch is precluded by high capital requirements, lengthy qualification cycles, and steep yield learning curves. However, the relevant threat is not startups, but established top-tier manufacturers redeploying tens of billions of renminbi into high-stage HDI—a transition currently underway.

The synthesis: Victory Giant occupies a strong position characterized by attractive current economics but challenging long-term industry dynamics. Its exceptional returns reflect a temporary supply-demand imbalance in high-stage HDI capacity, protected by an estimated process lead of one to two product generations. That is an enviable position, but it is not an unassailable fortress. The central question for investors is whether management can allocate its current cash flows to build enduring advantages before competitive capacity matures.

↑ Back to Top

IX. Management Credibility, Capital Allocation, & Governance Stress Test (15 min)

On the evening of June 8, 2026, Victory Giant’s board announced an employee stock ownership plan (ESOP) of up to ¥700 million, to be funded through secondary market purchases for non-independent directors, senior executives, and subsidiary staff.¹⁸

The timing was striking. Two days earlier, on June 6, a social media account published personal allegations against Chairman Chen Tao, including leaked elevator surveillance footage said to date from April 2025.¹⁷ On June 7, President Zhao Qixiang (赵启祥) stated that Chen had briefed directors, that the claims were "not entirely factual," and that the company had filed a police report.¹⁷ On June 8, Victory Giant issued a statement asserting that the material was unrelated to operations and did not involve corporate governance, internal controls, or disclosable events.¹⁷ Nevertheless, the stock dropped 7.11% that day, wiping out more than ¥50 billion in market capitalization across two sessions.¹⁸ The shares recovered over 6% the next morning following the ESOP announcement.¹⁸

How should investors evaluate this episode? Not through a moral lens—a chairman’s private life is not intrinsically a fundamental business metric, and operational continuity remained intact. The analytical takeaway is narrower and more instructive. First, it highlighted severe key-person concentration: Victory Giant remains a founder-controlled enterprise where markets treat the chairman’s personal standing as proxy for enterprise value. Chen Tao holds effective control over 33.85% of voting rights through Shenghua Xinye (胜华欣业, 18.61%) and Hong Kong Victory Giant (香港胜宏, 15.24%), in which he holds 90% and 70% equity stakes, respectively.²⁰ Second, the rapid crisis playbook—denying allegations, filing a police report, and announcing a buyback-adjacent ESOP within 48 hours—demonstrated aggressive market-capitalization management. While effective in calming the market short term, it leaves an unresolved governance reality: a business so dependent on a single leader lacks sufficient institutional insulation.

The broader capital allocation record warrants a similar audit, revealing a distinctly mixed performance.

On the credit side: The 2019 HDI investment was an ambitious, non-consensus wager that yielded extraordinary returns. Management also executed the Hong Kong listing with precision—raising nearly US$2.95 billion (HK$23.1 billion) at peak sector valuations, pricing at the top of the range, and fully exercising the over-allotment option.²¹³ Furthermore, 2025 operating cash flow reached ¥4.60 billion, confirming that paper profits translated directly into cash conversion rather than unpaid receivables.³

On the debit side, several decisions invite closer scrutiny:

Insider divestments ahead of public capital raises. On May 5, 2025, Shenzhen Shenghua Xinye Investment Co., Ltd. (深圳市胜华欣业投资有限公司)—wholly owned by Chen Tao and his wife Liu Chunlan (刘春兰)—sold a 2.98% stake via block transfer, realizing ¥1.694 billion.²⁰²¹ In August 2025, five executives disclosed plans to reduce their holdings by roughly ¥450 million, pushing cumulative insider sales above ¥2.1 billion.¹² These monetizations directly preceded both the ¥1.9 billion domestic private placement and the HK$23.1 billion Hong Kong IPO. While compliant with regulations, the sequence—controlling insiders monetizing at cycle peaks shortly before soliciting public equity—naturally invites skepticism from minority shareholders and domestic market commentators.¹²

Pre-raise leverage. The company entered 2025 with a debt-to-asset ratio of 53.8% and a modest cash buffer, driven by the all-cash MFS acquisition and heavy HDI capital expenditures.¹² Management ran a stretched balance sheet and relied on a receptive capital market to refinance. While successful, the strategy carried meaningful balance-sheet risk rather than conservative prudence.

Ambitious performance targets as a credibility test. On July 20, 2026, Victory Giant announced a restricted stock incentive plan granting up to 4.86 million shares—roughly 0.49% of total equity—with an initial grant of 4.66 million shares at ¥155.94 per share to 1,545 core managers and technical personnel. The vesting schedule mandates net profit growth relative to the 2025 baseline of at least 40% (trigger) or 50% (target) in 2026, cumulative growth of 96% or 125% by 2027, and cumulative growth of 174.4% or 237.5% by 2028.¹⁹ This represents management’s most explicit forward operational guidance. The upper 2028 target implies net profit exceeding ¥14.5 billion—roughly 3.4 times 2025 levels. This creates a clear benchmark for evaluating management's forecasting discipline and tracking whether targets are subsequently relaxed.

Strategic consistency alongside disclosure gaps. To management's credit, its strategic narrative has remained consistent: scaling high-stage product capabilities, expanding overseas production, and deepening Tier-1 client integration. The 2025 annual report, the Hong Kong prospectus, and the 2026 capital expenditure blueprint all align around expanding Southeast Asian manufacturing capacity.¹¹⁵ The shortcoming lies not in narrative coherence, but in financial granularity.

Material transparency deficits. The company provides no segment-level margin breakdowns, no dedicated AI-platform revenue reporting, and no isolated disclosures on MFS's financial contribution. For a firm executing a ¥20 billion annual capital plan for an international investor base, these omissions fall short of global reporting standards.

The Activist's Brief

If an activist fund built a position and published a campaign letter, its core demands would focus on pre-empting structural vulnerabilities rather than attacking strategy:

• On disclosure: "Victory Giant is asking public markets to fund a capital program four times its annual net profit while withholding the segment-level margin metrics required to evaluate capital returns. The company should immediately disclose revenue and gross margins broken down by product tier and end-market to eliminate the transparency discount."
• On alignment: "The controlling entity realized ¥1.694 billion in May 2025, and executives announced an additional ¥450 million in planned sales in August 2025—months before executing a ¥1.9 billion domestic placement and a HK$23 billion international offering.²⁰¹² To restore investor confidence, management should commit to binding lock-up windows around future primary capital raises."
• On balance sheet discipline: "Victory Giant entered 2025 carrying a 53.8% debt-to-asset ratio, roughly ¥3.9 billion in total borrowings, and under ¥1.2 billion in cash, relying on open equity windows for liquidity.¹² With Hong Kong IPO proceeds securing the balance sheet, management must formalize a target capital structure to clarify whether future downturns will be absorbed by balance-sheet reserves or further equity dilution."
• On concentration risk: "While the largest customer accounts for a reported 14.97% of revenue, true end-platform dependency is far higher because AI accelerator boards route through multiple contract assemblers.³ Management should disclose true end-platform revenue concentration to prevent market panic over unverified rumors, such as the 8.32% single-day stock decline in July 2026.¹⁶"
• On key-person governance: "A two-session, ¥50 billion wipeout in market capitalization triggered by personal allegations against the chairman highlights an acute governance vulnerability.¹⁸ The board must enhance institutional independence, formalize succession plans, and strengthen the oversight role of independent directors."

These points represent standard governance alignment questions rather than allegations of misconduct. Addressing them offers management the most direct path toward narrowing Victory Giant's valuation gap with global peers.

So what for investors: Management deserves substantial credit for anticipating the HDI technology transition and executing capital raises effectively. However, it has not yet established a track record of transparency or institutional alignment. Investors must weigh exceptional operational execution against ongoing governance and disclosure risks.

↑ Back to Top

X. Investment Thesis: The Bull vs. Bear Case & Key KPIs (15 min)

By mid-2026, public markets had already staged their own debate, written directly into the stock's price action. Victory Giant's shares peaked at ¥402.60 on May 27, 2026, before dropping over 40% in less than two months to ¥241.50 on July 17.¹⁶ That sell-off occurred despite solid operational momentum: first-quarter revenue rose 28% year-over-year to ¥5.52 billion, net profit grew 40.0% to ¥1.288 billion, and operating cash flow surged 399.4% to ¥2.12 billion.⁴¹⁴

When a stock tumbles while current earnings continue to compound, the market is signaling that the core debate is not about the latest quarterly results.

Why Victory Giant Wins From Here

The strongest bullish case rests on the physics of the product roadmap rather than current market share. Successive AI platform generations demand higher layer counts, more lamination stages, tighter impedance control, and stricter thermal management. Process difficulty protects the leader by stranding competitors on the yield curve. If the estimated gap between Victory Giant's 80% high-stage yield and the industry's 50% average is accurate, the company captures supernormal returns on complex specifications that rivals cannot reliably mass-produce.¹²

The second pillar is product-mix dynamics. As demonstrated in 2025, Victory Giant can nearly double revenue while physical production volume contracts.³ If high-stage HDI and ultra-high-layer boards expand as a percentage of total output, blended gross margins can remain above 30% even as unit prices drop across hardware generations, because high-value mix gains outweigh price concessions.

The third advantage is geographic diversification positioned as a commercial feature. Manufacturing bases in Thailand, Vietnam, and Malaysia provide a capability that most mainland Chinese peers lack: satisfying Western buyers' strict supply-chain compliance mandates.¹⁵ Where component sourcing turns on geopolitical risk management rather than raw unit price, overseas capacity operates as a direct sales driver rather than a simple downside hedge.

Fourth, market reporting and management updates indicated that the order book extended into 2027 and 2028, with AI-related backlog up 85% year-over-year in the first half of 2026 and client engagements widening beyond Nvidia and domestic chipmakers to include Google and AMD.¹⁶ If these broader relationships mature into volume contracts, customer concentration risks will diminish noticeably.

Why It May Not

The bearish argument builds on customer concentration and identifies four primary structural vulnerabilities:

Capital intensity in a cyclical sector. Deploying ¥20 billion in capital expenditure in a single year represents an unhedged bet on sustained end-market growth.³ Fixed assets and depreciation overhead remain constant when order volumes drop. If AI infrastructure spending slows due to power grid bottlenecks, high-bandwidth memory shortages, or routine hardware digestion, Victory Giant will carry high fixed depreciation charges against underutilized advanced production lines. Operating leverage amplifies down-cycles just as aggressively as up-cycles.

Parallel industry expansion. Competitors including Shennan Circuits, Avary Holding, and Wus Printed Circuit have committed tens of billions of renminbi toward high-end board capacity scheduled to come online in 2027 and 2028.²⁹²³ Introducing substantial new supply into a market previously defined by scarcity typically converts high-margin specialized components into low-margin commodities. This expansion represents the most predictable medium-term risk, arriving precisely when Victory Giant's internal incentive targets demand peak profit growth.

Cost inflation and currency exposure. Raw material inputs expanded as a share of total manufacturing costs in 2025, financial expenses jumped on foreign-exchange volatility, and first-quarter 2026 gross margins slipped slightly, demonstrating that profitability remains vulnerable to macroeconomic pressures.³¹⁴¹⁶

Rapid product turnover. High-performance board specifications turn over roughly every eighteen to twenty-four months.¹² Each hardware transition reopens customer qualification, resets yield learning curves, and creates opportunities for competitors to gain share. Industry reports indicate Victory Giant's AI revenues have relied heavily on single flagship product platforms in sequence—from the H-series to the GB-generation—leaving the business uniquely vulnerable to design changes or platform delays.¹² Indeed, reported schedule adjustments for next-generation platform ramps contributed to the mid-2026 stock derating.¹⁶

Governance and disclosure headwinds. Substantial insider stock sales ahead of public capital raises, incomplete segment margin reporting, and high key-person reliance do not preclude operational success, but they lead institutional investors to apply a persistent valuation discount.

The Risk Radar That Actually Applies

Standard macro risk assessments are often generic. In Victory Giant's case, four factors possess clear, direct transmission mechanisms into its financial statements, while one widely cited concern is less relevant.

Geopolitical and trade policy. Generating 76.8% of total revenue from direct exports—with high-margin volumes linked to U.S. AI infrastructure—leaves Victory Giant exposed to export regulations, tariffs, and buyer procurement rules.³ The primary risk is not tariff absorption, but customer qualification. If a client mandates non-mainland manufacturing for a platform, Chinese facilities face immediate disqualification regardless of yield or pricing. Production capacity in Thailand and Vietnam must then pass separate, multi-month qualification cycles, making overseas execution timelines a direct earnings variable.

Foreign exchange volatility. Financial expenses surged 467.1% in 2025 and another 265.7% in the first quarter of 2026, driven largely by currency fluctuations on export sales.³¹⁴ Generating over three-fourths of revenue abroad while maintaining debt service and domestic operating costs creates a structural currency exposure large enough to move reported net income independently of core factory operations.

Input material inflation. Raw materials expanded from 62.7% to 65.9% of cost of goods sold in 2025, driven by supply constraints in high-frequency, low-loss copper-clad laminates.³ Because specialized laminates represent an essential input for AI accelerators, price increases directly compress gross margins unless fully passed downstream.

Capital program execution. Management's plan to deploy ¥20 billion across three countries simultaneously in 2026 constitutes a complex operational undertaking for a company building its first overseas high-stage HDI facilities.³¹⁵ Advanced manufacturing plants frequently experience ramp delays, and unabsorbed factory overhead creates immediate depreciation charges before generating offsetting revenue.

The risk that does not apply in conventional form: Direct AI disintermediation. Victory Giant benefits from AI deployment rather than facing disruption by it. The genuine technological threat is narrower and longer-term: the eventual adoption of glass-core substrates to replace organic high-stage HDI boards. While glass substrates represent a late-2020s development rather than an immediate risk, commercial adoption would directly challenge the company's core process advantage.

The Two KPIs That Actually Matter

In evaluating Victory Giant's trajectory, headline growth can obscure underlying performance. Two metrics provide essential signal, alongside a third operational variable:

1. Revenue per square meter. Victory Giant discloses both total revenue and physical sales volume in its annual disclosures.³ Dividing revenue by volume yields the purest measure of product mix and process power. Sustained increases in revenue per square meter indicate that the shift toward higher-stage HDI is outrunning general price erosion. Conversely, if revenue per square meter plateaus while physical output expands, AI board manufacturing is commoditizing—a trend that will surface in unit realization before appearing in total top-line growth.

2. Blended gross margin. Group gross margin stood at 35.2% in 2025 and 34.5% in the first quarter of 2026, reflecting the interplay between high-margin mix expansion, annual price negotiations, and raw-material inflation.³¹⁶ A decline toward the high twenties would signal that scarcity rents are dissipating, whereas margin expansion would confirm a widening technological lead. Wus Printed Circuit's core PCB gross margin serves as the most accurate direct benchmark.²²

3. Overseas capacity utilization. Facilities in Thailand and Vietnam represent the primary destination for ongoing capital expansion.¹⁵²⁵ The speed with which these offshore plants secure customer qualifications and achieve high utilization determines whether the ¥20 billion investment program generates incremental earnings or unabsorbed depreciation.

Conversely, quarterly top-line growth serves as a lagging indicator tied to platform build schedules, while unverified "AI revenue share" estimates lack audited backing and offer limited analytical value.

↑ Back to Top

XI. Outro & Lessons for Investors (5 min)

One narrative version of the Victory Giant story reads as a classic industrial turnaround: a former soldier from Xinjiang self-studies circuit board engineering, builds a factory in Huizhou, bets the balance sheet on a high-difficulty manufacturing process, and eventually secures a seat at the table with Nvidia's leadership. That account is accurate, but it offers limited analytical value for investors.

The more instructive perspective is that between 2019 and 2022, a mid-sized manufacturer accepted three years of stagnant margins, mounting depreciation, and market skepticism in exchange for an option on manufacturing complexity. Management was not specifically predicting artificial intelligence in 2019; rather, it wagered on the broader principle that mastering high-stage manufacturing would yield superior economic returns over commodity output. When hyperscale AI demand arrived, that option paid off precisely because high technical barriers kept the supplier cohort narrow.

Three broader lessons emerge for industrial investors:

First, capital commitments into manufacturing bottlenecks often precede visible demand shifts. The indicator to track is a company whose capital expenditures expand while profit margins remain compressed, and whose management invests in a technological transition that public markets are actively discounting. While many firms fail in this transition, those that succeed experience sharp valuation re-ratings as markets reprice both immediate earnings and long-term competitive positioning.

Second, comparing physical production volume against overall revenue provides a clear measure of product mix shift. Victory Giant’s 2025 financial disclosures revealed physical output contracting while total revenue nearly doubled. That single divergence offered stronger empirical proof of high-value product mix expansion than any executive commentary. When industrial companies report both physical output and financial results, tracking revenue per unit of physical output isolates genuine pricing power from volume growth.

Third, a process advantage operates as a ticking clock rather than an unassailable moat. Victory Giant's technical lead generates substantial current returns, but well-capitalized competitors are deploying tens of billions of renminbi to close the gap. Furthermore, major buyers actively maintain multi-sourcing strategies, and each new hardware generation resets customer qualification. The core analytical question is not whether the company possesses a permanent moat, but how long its process advantage will endure—and how effectively management allocates capital before competitive capacity matures.

That question is now being tested in public markets. Management has committed to incentive targets that imply nearly tripling net profit from a record 2025 baseline by 2028, while committing ¥20 billion annually toward capital expenditures just as major peers execute parallel expansions.¹⁹³ Either Victory Giant's process advantage is far deeper than skeptics estimate, or the industry is set to demonstrate once again how quickly specialized electronics manufacturing can revert to commodity economics.