Kinsus Interconnect Technology Corp. (3189.TW): The Silicon Chokepoint & The AI Substrate Supercycle

I. Introduction & Episode Roadmap [00:00 - 09:00]

Take a Blackwell-class AI accelerator out of its packaging tray and look at it edge-on. The part everyone photographs is the silicon: the compute dies, the stacks of high-bandwidth memory, the mirror-flat lid. Underneath all of it sits something far less glamorous — a slab of resin and glass cloth, perhaps sixty millimetres on a side and barely a millimetre thick, threaded with tens of thousands of copper wires finer than a red blood cell.

That slab is the IC substrate. It serves as the adapter between two incompatible physical domains: a silicon die with connection pads spaced a few tens of microns apart, and a printed circuit board with solder balls spaced a thousand microns apart. Without it, even the most advanced semiconductor chip is an expensive piece of silicon with nowhere to plug in.

For most of the past two decades, substrates were a commodity backwater — low margin, cyclical, and invisible to anyone outside Taoyuan and Gifu Prefecture. Then AI accelerators arrived, and the industry discovered that the bottleneck in shipping GPUs was not always fab capacity. Sometimes it was resin.

Enter 景碩科技 Kinsus Interconnect Technology Corp. — ticker 3189 on the Taiwan Stock Exchange, headquartered in Taoyuan City, and majority-influenced by the 和碩聯合科技 Pegatron Corporation group.¹ In FY2025, the company generated consolidated revenue of NT$39.3 billion (roughly US$1.2 billion) with a gross margin of 21.1% and net profit of NT$1.596 billion — a figure that rose more than thirty-fold from the prior year's near-breakeven result.²³

That thirty-fold expansion is a statistic that both flatters and misleads. The growth rate appears dramatic largely because the comparison base was near zero. Kinsus spent 2023 and 2024 operating close to breakeven while absorbing heavy depreciation from a capacity expansion committed to in a very different demand environment. The financial recovery is real, but the percentage growth is an artifact of low-base arithmetic.

By the first half of 2026, operational performance had measurably improved. First-half revenue reached NT$23.6 billion, up 29.8% year on year, with a gross margin of 23.8%. In the second quarter, gross margin reached 26.1% — the company's highest quarterly print since the previous industry supercycle.⁴⁵ Equity markets responded aggressively: after closing 2025 at NT$159, the stock traded above NT$800 during the summer of 2026.⁶

What this episode is actually about

This is not a story of breakthrough invention; Kinsus invented very little. Instead, it is the story of a company sitting at a physical manufacturing chokepoint. After entering the most critical segment of that market late, Kinsus spent five years and over NT$50 billion in capital expenditure attempting to catch up. After nearly exhausting its patience in 2023, the company found itself positioned to capture demand just as the AI advanced packaging boom took off.

Key analytical questions include:

• How did a company built on Bismaleimide Triazine (BT) substrates for smartphones and memory chips transition into Ajinomoto Build-up Film (ABF) substrates for CPUs, GPUs, and AI accelerators — and is it a genuine peer to industry leaders or a capable follower capitalizing on a supply shortage?
• What happens inside the reported financials when a semiconductor substrate manufacturer consolidates a contact lens business, and how does that single consolidation affect consolidated margin metrics?
• Why does substrate manufacturing behave less like electronics assembly and more like a high-precision yield business, where a microscopic misalignment on the eighteenth layer destroys the entire unit?
• What are the operational dynamics of competing against 欣興電子 Unimicron Technology — more than three times Kinsus's size — and イビデン Ibiden, while sourcing critical raw materials from Japanese suppliers that hold dominant market shares?
• When a company raises equity at a 31% discount and requests authorization for a private placement of another hundred million shares, what risk profile is a minority investor being asked to underwrite?

Answering those questions requires starting where the company began — as part of another corporate empire.

---

↑ Back to Top

II. Founding, ASUSTeK Spin-Off & The Pegatron Umbrella (2000–2010) [09:00 - 22:00]

In September 2000, Taiwan's electronics industry was in the middle of a land grab. The island had secured dominance in motherboards, notebooks, and optical drives. What it had not yet won was the layer between chip and board — a domain that still belonged, overwhelmingly, to Japan.

Kinsus was incorporated on 11 September 2000 as a subsidiary venture of 華碩電腦 ASUSTeK Computer, then Taiwan's dominant motherboard maker.⁷ The founding logic was straightforward: ASUSTeK understood high-layer-count board manufacturing better than almost anyone in Taiwan, and the substrate business appeared to be the same discipline scaled down by an order of magnitude. The core processes — drilling, plating, and lamination — were identical in concept, just executed at far tighter tolerances.

That assumption turned out to be only half right, and the half that was wrong required fifteen years to resolve.

The company listed on the Taiwan Stock Exchange in November 2004 under ticker 3189.⁷ Its early product line was built around the workhorse of the mobile era: BT substrates, named for the bismaleimide-triazine resin forming their core. BT substrates act as stiff, thin, thermally stable circuit boards — typically containing two to eight layers — designed for applications where thinness and cost take precedence over raw interconnect density.

What BT actually does, in plain language

A modern smartphone application processor features roughly five hundred to a thousand external connections, while a DRAM die requires even fewer. A standard BT substrate handles those connection counts comfortably. The central engineering challenge is not pin density, but managing warpage, package thickness, and manufacturing volume under intense cost pressure from handset makers shipping tens of millions of units annually.

Kinsus established a substantial presence in this segment: supplying substrates for application processors, radio-frequency front-end and Wi-Fi modules, power management ICs, and memory packaging across flip-chip and wire-bond chip-scale packages.¹ While these components carried modest unit margins, high-volume production generated the empirical yield data required for advanced manufacturing.

In substrate manufacturing, operational refinement is empirical rather than theoretical. Yield improvements stem from processing millions of panels, logging defect patterns, and systematically eliminating failure modes. High-volume manufacturing of simpler boards creates a broad defect library that lower-volume specialty production cannot match. Kinsus's focus on mobile components served as an extended operational apprenticeship in process control prior to tackling complex multi-layer packaging.

The company also pursued lateral expansion during this period, acquiring Piotek Computer in 2010 to extend its circuit board and component footprint.⁷ As product strategy evolved over the subsequent decade, the group progressively pruned non-core acquisitions.

The 2008 demerger and the arrival of Pegatron

The company's corporate structure shifted in 2008 when ASUSTeK restructured its operations, separating its branded business from contract manufacturing and spinning off the manufacturing arm as Pegatron. Kinsus was allocated to the manufacturing organization.⁷

Pegatron holds its equity stake in Kinsus through wholly owned investment vehicles, maintaining an aggregate position in the high thirties as a percentage of outstanding shares.⁷ Two of those entities — Hua Yu Investment and Hua Xu Investment — hold dedicated seats on the board of directors.⁸

This ownership structure fundamentally defines Kinsus's corporate governance. The company operates as a group subsidiary whose chairman serves simultaneously as Pegatron's chief operating officer, while its honorary chairman is Pegatron founder 童子賢 Tzu-Hsien Tung.⁸

The conglomerate balancing act

This ownership structure presents distinct analytical trade-offs.

On one hand, Pegatron's majority backing provided capital stability through cyclical downturns. Substrate manufacturing requires years of front-loaded capital expenditure before generating net returns, making a parent company with a strong balance sheet a practical asset. Parent oversight also enforced working capital discipline and a manufacturing culture focused on yield metrics.

On the other hand, operating within an electronics manufacturing services group imposes strategic constraints. Substrate suppliers must convince fabless chip designers — who frequently compete with the parent group's other manufacturing clients — that proprietary designs remain secure. While Kinsus built a broader merchant customer base among independent chip designers, related-party structures require ongoing scrutiny regarding capacity allocation and pricing transparency for minority shareholders.

The broader constraint involved risk tolerance. Group companies typically mirror their parent's capital allocation style. Pegatron's operational model developed in contract assembly — characterized by thin operating margins, strict capital control, and incremental expansion. While effective for plant management, this conservative orientation delayed the aggressive, long-term capital commitments needed to compete against Japanese incumbents with multi-decade head starts in high-end substrates. Throughout the 2010s, Kinsus maintained its core BT position while taking a measured approach to advanced packaging investment.

A third structural element emerged in August 2009, when Kinsus and Pegatron jointly established contact lens manufacturer 晶碩光學 Pegavision, with Kinsus taking the larger equity stake.⁷ Initially a minor diversification, Pegavision grew into a consolidated subsidiary with a distinct margin profile — an arrangement that materially influences Kinsus's reported group financials.

By 2010, Kinsus operated as a mid-sized BT substrate maker supported by a major manufacturing parent. However, the company remained heavily weighted toward standard mobile substrates just as the industry's broader technological transition was gathering momentum.

---

↑ Back to Top

III. The Materials Physics Divide: BT vs. ABF & The High-Performance Computing Pivot (2011–2020) [22:00 - 40:00]

Understanding why Kinsus spent the 2010s playing catch-up requires looking at a technological divide rooted less in corporate strategy and more in fundamental chemistry.

Why leadframes died

Historically, integrated circuits connected to printed circuit boards through metal leadframes — the visible pins lining the perimeter of a package. Leadframes functioned effectively for chips with dozens of interconnects, but physical geometry created a hard ceiling when connection counts escalated into the thousands. Edge space simply ran out.

The semiconductor industry solved this perimeter constraint with the ball grid array, utilizing the entire underside of the chip package. However, routing thousands of solder balls requires an adapter to bridge the spatial gap between micron-scale silicon die pads and millimetre-scale board contacts. That role falls to the package substrate — a high-density miniature circuit board engineered to tolerances far tighter than standard printed circuit board shops can achieve.

Two materials, two universes

BT substrates are built around a rigid bismaleimide-triazine resin core that provides mechanical stability, flanked by a few build-up layers. Typically ranging from two to eight layers, BT substrates are thin, relatively inexpensive, and produced in vast quantities for mobile applications.

ABF substrates belong to a distinctly higher class of complexity. Based on Ajinomoto Build-up Film, this specialized resin film undergoes iterative cycles of lamination, laser drilling, plating, and circuit patterning. Modern high-performance computing components feature ten, sixteen, twenty layers or more, with circuit line and space geometries down to single-digit microns. Advanced artificial intelligence packaging configurations have scaled from roughly three build-up layers per side to eleven per side, with thirteen per side anticipated later in the decade.⁹

The fundamental driver for ABF resin is matching thermal expansion properties. Silicon expands minimally under thermal load, whereas standard organic resins expand significantly. Bonding a large silicon die to a substrate with a high coefficient of thermal expansion causes joint cracking under repeated operating temperature cycles. ABF chemistry — reinforced with low-thermal-expansion glass cloth — closes this thermal expansion gap, preserving structural integrity in large multi-die packages.

The structural contrast can be conceptualized by comparing layout densities: if a BT substrate resembles a suburban street grid with a few layers and wide avenues, an ABF substrate for an AI accelerator functions like a twenty-storey underground transport interchange. Every vertical layer must align across microscopic tolerances, where a single misaligned laser drill hole on the eighteenth layer ruins the entire component.

Critically, manufacturers cannot inspect their way out of defect accumulation. Once the eighteenth layer is laminated, layers one through seventeen are permanently sealed inside. A yield loss on a final layer forfeits all cumulative labor and material costs embedded in the earlier processing steps. Consequently, high-layer-count substrate manufacturing bears the economics of semiconductor front-end fabrication, where yield control dictates financial survival.

This operational reality explains why industry capacity fails to respond quickly during demand surges. Substrate makers cannot simply buy their way to commercial production. Equipment acquisition and installation are merely prerequisites; establishing profitable yields requires twelve to twenty-four months of continuous process optimization on live customer production lines. As a result, theoretical paper capacity routinely lags usable commercial output by a substantial margin.

The seasoning company that owns the resin

A striking feature of the semiconductor substrate supply chain is that its core dielectric material is produced by a Japanese food conglomerate.

味の素 Ajinomoto, widely recognized for seasoning products, developed the build-up film through its Fine-Techno division. The material became the universal standard for high-performance package substrates, maintaining a market share exceeding 95% without a viable secondary supplier.⁹ This market dominance gives raw material suppliers significant pricing leverage. When Ajinomoto announced a 30% price increase on its core ABF materials effective in the third quarter of 2026 — following similar 30% price hikes from Resonac and 三菱ガス化学 Mitsubishi Gas Chemical on copper-clad laminates in April 2026 — substrate manufacturers across the industry were forced to absorb the higher input costs while attempting to pass them down to customers.⁹

Mitsubishi Gas Chemical occupies a similarly dominant position in BT resin raw materials. Consequently, substrate suppliers operate in a market squeezed from both sides: purchasing essential chemistry from two highly concentrated Japanese suppliers and selling finished substrates to a consolidated group of major semiconductor design firms.

Kinsus's late pivot

Throughout the 2010s, the high-end ABF substrate market remained dominated by Japanese suppliers — specifically Ibiden and 新光電気工業 Shinko Electric Industries, which served as primary vendors for Intel's processor sockets — while Unimicron and 南亞電路板 Nan Ya PCB established high-volume manufacturing scale in Taiwan. Kinsus operated at the fringe of the ABF market, supplying substrates with lower layer counts, smaller surface areas, and less demanding specifications.

This secondary market positioning resulted primarily from commercial barriers rather than raw technical capability. The principal ABF volume drivers during the 2010s centered on desktop and server x86 platforms. Incumbent chipmakers relied on long-qualified suppliers and had little incentive to introduce new vendors. Entering high-spec ABF production required a platform shift, an industry-wide capacity shortfall, or a customer actively seeking dual-sourcing options. For most of the decade, none of these conditions materialized in sufficient force.

Simultaneously, the mobile handset segment that historically financed Kinsus's operations entered structural stagnation. Consolidated revenue remained essentially flat across five consecutive years: NT$24.9 billion in 2014, NT$23.1 billion in 2015, NT$23.2 billion in 2016, NT$22.3 billion in 2017, and NT$23.7 billion in 2018.² As fixed depreciation charges escalated, operating leverage worked in reverse. Group net profit collapsed over the same interval, contracting from NT$3.62 billion in 2014 to NT$349 million in 2018.²

The financial trough arrived in FY2019. Revenue dropped to NT$22.3 billion, yielding an operating loss of NT$1.65 billion and a net loss of NT$2.03 billion.² For an enterprise burdened by over NT$4.6 billion in annual depreciation, such losses signalled that the legacy mobile product mix was no longer financially viable.

In response, Kinsus initiated a capital reallocation strategy to pivot away from legacy low-margin board manufacturing toward high-layer-count ABF substrate capacity. To support this operational shift, the company acquired the Younglion facility in Yangmei, Taoyuan, in 2021 — designated as Plant 6 — dedicated exclusively to ABF expansion.⁷

What this tells an investor: the transition into advanced packaging was reactive rather than visionary. Kinsus did not foresee the high-performance computing supercycle; it was displaced from a maturing mobile market and gravitated toward the only viable growth vector. While this context tempers management claims of strategic foresight, the central investment question hinges on operational execution. Capital expenditure can acquire manufacturing space and machinery, but achieving commercial yields on complex substrates requires technical qualification.

The timing of this expansion was about to deliver eighteen months of extraordinary market gains, followed by a sharp two-year cyclical downturn.

↑ Back to Top

IV. The Pandemic Supercycle, PC Hangover & The AI Catalyst (2020–2024) [40:00 - 58:00]

When component demand outstrips global manufacturing capacity and physical production cannot scale quickly, standard market pricing dynamics break down. Package substrate manufacturers experienced that operational squeeze in 2021.

The supercycle

The sudden shift toward remote work and home entertainment expanded hardware demand across notebooks, servers, gaming consoles, and cryptocurrency mining rigs simultaneously. Because substrate manufacturing capacity had expanded conservatively over the preceding decade, supply lead times stretched dramatically. Customers who had spent years negotiating aggressive price reductions began offering upfront cash prepayments to secure factory capacity allocation.

This shift appeared directly on the corporate balance sheet: non-current deferred revenue — representing long-term customer prepayments — expanded from NT$55 million at the end of 2021 to NT$2.44 billion by the end of 2022.² These prepayments functioned as direct customer funding for plant expansion.

FY2022 marked the peak of that cycle. Annual revenue reached NT$42.4 billion, gross margin hit 37%, operating income reached NT$9.6 billion, net profit touched NT$6.98 billion, and earnings per share rose to NT$15.47.² Kinsus converted roughly twenty-five cents of every revenue dollar into operating profit — an extraordinary margin level for an organic substrate manufacturer.

Management responded to peak profitability by accelerating capital expenditure. The company committed NT$13.4 billion to capital investments in 2021 and NT$17.2 billion in 2022 — the latter representing roughly 40% of full-year revenue.² Consequently, free cash flow remained negative across both peak years as operational cash was reinvested directly into equipment and plant expansion.

The hangover

When personal computer demand contracted sharply in late 2022, channel inventory accumulated rapidly. As an upstream component vendor, substrate manufacturers absorbed the full impact of the order drop.

In FY2023, consolidated revenue fell 37% to NT$26.8 billion, while operating income dropped from NT$9.6 billion to NT$1.04 billion.² Factory utilization across legacy BT and lower-layer ABF lines declined significantly. Meanwhile, annual depreciation charges stemming from the 2021–2022 expansion reached NT$5.4 billion, eroding most of the remaining gross profit.

Customer prepayments reflected the unwinding of the cycle. Non-current deferred revenue peaked at NT$3.91 billion at the end of 2023 before declining to NT$3.13 billion at end-2024 and NT$2.10 billion at end-2025 as long-term allocation agreements were fulfilled or concluded.²

Dividend allocations adjusted accordingly. After distributing NT$3.44 billion in 2023 from record FY2022 profits, Kinsus reduced total dividend payouts to approximately NT$450 million in both 2024 and 2025.² For FY2025, the board approved a dividend of NT$1.75 per share, representing approximately a 50% payout ratio.³ This payout pattern illustrates a capital management policy tied directly to annual earnings variability rather than fixed dividend targets.

The Pegavision adjustment — the single most important disclosure nuance

Evaluating Kinsus's underlying substrate economics requires adjusting for the financial consolidation of contact lens maker Pegavision. In FY2025, Pegavision generated revenue of NT$7.04 billion — up 3.25% year-on-year to a seventh consecutive annual record — while delivering a gross margin of approximately 52%.¹⁰ Combining high-margin optical consumer products with industrial substrate manufacturing elevates headline group margins.

Excluding Pegavision provides a clearer picture of core substrate performance. In FY2025, subtracting Pegavision's NT$7.04 billion from group revenue of NT$39.35 billion leaves approximately NT$32.3 billion in substrate and circuit board revenue. Similarly, deducting an estimated NT$3.66 billion in Pegavision gross profit from group gross profit of NT$8.31 billion leaves roughly NT$4.65 billion — yielding a standalone substrate gross margin in the mid-teens rather than the reported 21.1% group figure.²¹⁰ A similar adjustment for FY2023 demonstrates that substrate operations earned mid-teens gross margins during the trough, compared to reported group margins in the mid-twenties.

A parallel distortion affects bottom-line profit attribution. Because Kinsus holds a non-majority economic stake in Pegavision while consolidating its financial statements, a significant portion of reported earnings is allocated to non-controlling interests. Non-controlling interests represented NT$8.64 billion of the group's NT$41.37 billion total equity at the end of FY2025.² In FY2024, consolidated net income reached NT$1.33 billion, but net profit attributable to Kinsus shareholders was NT$48.9 million, or NT$0.11 per share.² Consequently, headline group figures overestimate both the gross margin and net earnings generated by core substrate manufacturing.

China minus one

During the industry downturn, Kinsus restructured its production footprint. In January 2025, the company announced the sale of its two Suzhou subsidiaries — comprising a 100% stake in its PCB operation and an 80.5% interest in its substrate unit — for approximately NT$8.2 billion, consolidating manufacturing facilities in Taiwan.¹¹ The divestment responded to customer preferences for non-China supply chains while exiting low-end market segments where domestic Chinese competition had compressed margins.

This geographic consolidation streamlined operations and reduced trade policy exposure with Western customers, though it concentrated physical assets on a single island — a structural trade-off addressed later in the risk analysis.

The AI catalyst

The primary catalyst for operational recovery was not a rebound in standard personal computers, but a structural transition in semiconductor packaging architecture.

Leading semiconductor designs transitioned from monolithic silicon dies to multi-chiplet configurations assembled on advanced packaging platforms such as TSMC's CoWoS. Because AI accelerator substrates require significantly larger surface areas than traditional server CPUs, each AI package absorbs multiple units of equivalent manufacturing throughput. This surface-area multiplier allowed modest AI accelerator unit volumes to absorb substantial factory capacity.

Kinsus's market communications during this transition highlighted the rapid shift in supply-demand dynamics. In August 2025, the company stated to DIGITIMES that overall ABF and BT substrate supply exceeded demand, while projecting that high-end ABF capacity would tighten during 2026.¹² Market tightening materialized faster than anticipated; by April 2026, high-end ABF production capacity across Unimicron, Kinsus, and Nan Ya PCB was reported as fully booked.¹³

Financial results reflected this utilization recovery. Full-year FY2025 revenue reached NT$39.35 billion with a gross margin of 21.1% — improving to 22.1% in the fourth quarter — while net profit reached NT$1.596 billion, or NT$3.51 per share.²³ By the second quarter of 2026, quarterly revenue rose to NT$12.5 billion — up 12.5% sequentially and 30.7% year-on-year — with gross margin reaching 26.1%, operating margin reaching 13.4%, and parent-attributable net profit reaching NT$1.311 billion, more than triple the prior-year period.⁴ Monthly sales disclosures confirmed continuous expansion, with monthly revenue throughout the first half of 2026 posting year-on-year growth between 10% in February and nearly 55% in January.⁵

Equity markets revalued the company rapidly during this operational turn. Kinsus shares traded up from NT$159 at the end of 2025 to NT$252 in January, NT$314.5 in February, and NT$528 in April, before advancing further during the summer of 2026.⁶ By late April 2026, market analysis noted that the equity had exceeded prior target valuations and was trading at approximately 67 times projected 2026 earnings.¹⁹ This valuation shift represented a market reclassification from a cyclical board manufacturer to an advanced AI packaging supplier — embedding high expectations for sustained capacity utilization and margin execution.

↑ Back to Top

V. Deep-Dive into Core Economics & Industry Structure [58:00 - 78:00]

The clearest snapshot of Kinsus's underlying sales mix came during its investor conference on 10 March 2026, when management outlined the fourth-quarter product breakdown.¹⁴

The anatomy of the revenue line

A detailed breakdown from the fourth quarter of 2025 shows smartphones generating 35% of total revenue, followed by ABF and high-performance computing at 24%, consumer electronics and memory at 18%, contact lenses at 18%, and base stations at 5%. Viewed by substrate technology, ABF accounted for 46% of substrate sales, while BT substrates represented 36%.¹⁴

These figures present a notable contrast to public market commentary. While financial media frequently categorizes Kinsus as a pure-play AI substrate producer, high-performance computing represented under a quarter of sales in its late-2025 disclosure, while smartphones remained its largest individual segment. In fact, contact lens manufacturing generated as much revenue as consumer electronics and memory combined.

This distribution does not invalidate the AI growth thesis, but it highlights the company's ongoing structural transition. A substantial portion of Kinsus's prospective margin expansion depends on a product segment that still constitutes a minority of its total revenue base.

The ABF business carries the highest average selling prices, the highest layer counts, the longest customer qualification cycles, and the steepest yield risks. It represents the primary engine of incremental gross profit and receives the bulk of corporate capital expenditure.

The BT business provides a steady baseline of cash flow. Demand for memory and mobile substrates exhibits greater stability than high-performance computing. In the current cycle, BT operations have benefited from an unusual dynamic: substrate prices have risen alongside ABF rates, driven primarily by raw material supply shortages in glass fabric rather than end-market demand alone.¹⁵ Management noted at the March 2026 investor conference that memory customers had demonstrated the highest acceptance of price adjustments, with average quarterly price increases of approximately 5.5% anticipated.¹⁴

Legacy board and other operations constitute a deprioritized remnant, having contracted significantly following the Suzhou facility divestments.

The competitive board

Comparing Kinsus against its industry peers using FY2025 operational metrics highlights a substantial scale disparity.

Unimicron generated revenue of NT$131.2 billion and net profit of NT$6.67 billion — more than three times Kinsus's total top-line scale.¹⁶ Nan Ya PCB, backed by the Formosa Plastics group, reported revenue of NT$40.2 billion and net profit of NT$1.95 billion, with a gross margin of roughly 9%.¹⁷ Ibiden and Shinko Electric Industries remain the premier Japanese benchmark suppliers for high-end applications, with Ibiden historically dominating complex processor packaging. Austria's AT&S serves as the primary European challenger, supported by expanding manufacturing capacity in Southeast Asia.

Two analytical conclusions emerge from these competitive comparisons:

First, Kinsus's reported group gross margin of 21.1% for FY2025 appeared superior to Unimicron's 13.9% and Nan Ya PCB's 9.1% — until the financial contribution of its contact lens subsidiary is removed, after which the core substrate margins appear broadly comparable.²¹⁶¹⁷ The headline margin premium is a corporate consolidation artifact rather than evidence of superior substrate manufacturing efficiency.

Second, scale dictates commercial advantage in substrate manufacturing. Operating at roughly one-third the scale of the market leader puts Kinsus at a structural disadvantage when negotiating tool delivery schedules, raw material allocations, research amortization, and customer allocations for flagship chip platforms.

Economic moats: the 7 Powers test

Evaluating Kinsus through Hamilton Helmer's 7 Powers framework yields a constrained set of competitive advantages.

Process Power — present, representing the company's strongest claim. The primary advantage for a substrate manufacturer lies in process control. Fabricating a twenty-layer ABF substrate requires executing dozens of sequential lamination, laser-drilling, plating, and circuit-imaging steps, with each layer meticulously aligned to the preceding pattern. Overall yield is the compounded product of individual layer yields; minor yield gains at each stage compound into substantial profit improvements, whereas small process errors destroy batch profitability.

This process knowledge is tacit, empirical, and difficult to replicate quickly. It resides in specialized processing recipes, machine calibration parameters, and engineering experience. However, every established competitor possesses similar operational expertise. Process Power creates barriers against new market entrants, but it does not confer a distinct advantage over direct peers like Unimicron.

Switching Costs — present, but moderate. Substrate qualification involves an extended co-design process among the chip designer, semiconductor foundry, assembly and test providers — such as ASE Technology and Amkor — and the substrate manufacturer. Qualification cycles typically span twelve to eighteen months. Re-qualifying a component mid-production is costly and disruptive, granting high incumbency protection for the duration of a specific product design. However, this stickiness applies to individual chip generations rather than long-term customer relationships, requiring suppliers to compete anew for subsequent product platforms.

This qualification mechanism directly influences operational strategy. The chip designer defines package specifications, the foundry sets thermal reflow tolerances, and the assembly house tests for structural warpage. Each party approves a specific manufacturing facility running a validated process line. Shifting production between plants requires full re-qualification. Consequently, commercial success depends on scaling capacity within qualified facilities, making Plant 6's ramp-up timeline a critical operational variable.

Scale Economies — present, but disadvantageous to Kinsus. The company operates at a scale deficit relative to Unimicron and Ibiden.

Cornered Resource — absent for Kinsus, but heavily concentrated among suppliers. Ajinomoto maintains a dominant position over ABF resin film, while Nittobo controls the vast majority of global T-glass cloth production. Kinsus holds no equivalent proprietary material assets.

Counter-Positioning, Branding, and Network Economies — not applicable to substrate manufacturing.

In summary, Kinsus possesses one strong operational power and one moderate structural switching cost, operating in an industry where critical inputs are controlled by upstream monopolists and scale advantages belong to larger rivals. Such positioning permits elevated returns during industry supply shortages, but yields modest profitability during normal operating conditions.

Porter's five forces, applied to a real shortage

Supplier power: very high, forming the binding operational constraint. On the first-quarter 2026 earnings call, company spokesperson Michael Mu stated that revenue could have been roughly 10% higher were it not for raw material supply limits, noting that material bottlenecks suppressed overall revenue expansion by 10% to 15%.¹⁸ T-glass represented the primary bottleneck, with a single supplier controlling roughly 90% of global output and meaningful capacity additions delayed until mid-2027.¹⁵ In addition, essential inputs like copper foil and drill bits faced supply diversion toward server circuit board manufacturers.¹⁸

Mu also clarified the impact on corporate margins, stating that "the extent of our price increases has been eaten up by rising raw material prices."¹⁸ Management emphasized that gross margin gains stemmed from higher factory utilization rather than expansion of pricing power.

This disclosure clarifies the nature of Kinsus's financial recovery. The company is experiencing classic operating leverage driven by volume recovery, rather than expanding structural pricing power.

Buyer power: high. Customer concentration is substantial, consisting of major global technology firms. In April 2026, reports indicated Kinsus had secured substrate orders for Google TPU and Meta AI accelerator programs, with Amazon Web Services supply slated for the second half of 2026.¹⁹ While securing these tier-one clients validates technical execution, these sophisticated buyers maintain aggressive dual-sourcing strategies that exercise pricing leverage whenever market capacity permits.

Rivalry: intensely cyclical. Industry competition vanishes during capacity shortages, allowing all suppliers to operate at full utilization and adjust pricing upward. Conversely, during cyclical downturns, fixed depreciation costs force aggressive price competition to maintain baseline plant utilization.

Threat of substitutes and new entrants: low in the near term. Heavy capital requirements and lengthy qualification timelines deter new corporate entrants. Technological substitution risks center primarily on emerging glass core materials rather than new competitive entrants.

Myth versus reality

Three common market assertions regarding Kinsus warrant examination against corporate disclosures.

Myth: "Kinsus is an AI substrate company." Reality: in detailed segment disclosures, high-performance computing accounted for 24% of revenue, while smartphones represented 35%.¹⁴ Although ABF has become the dominant substrate technology at 46% of sales compared to BT's 36% and serves as the primary growth driver, the majority of total revenue remains anchored in mobile components, consumer electronics, and contact lenses. AI demand represents an important growth driver rather than the foundation of current sales.

Myth: "Margin recovery proves expanding pricing power." Reality: management explicitly attributed margin gains to higher factory capacity utilization, acknowledging that selling price increases were largely offset by rising raw material costs.¹⁸ While Kinsus has implemented quarterly price adjustments averaging roughly 5.5%, these increases represent cost pass-throughs rather than expanding structural margins from product differentiation.¹⁴

Myth: "Kinsus achieves higher margins than Unimicron and Nan Ya PCB." Reality: while group financial statements report higher headline margins, adjusting for the consolidated contact lens subsidiary reveals standalone substrate margins that align closely with industry peers.^2101617 Failing to adjust for non-substrate operations remains a widespread analytical oversight in market coverage.

These disclosures do not undermine Kinsus's investment case, but they reframe its nature. The company represents a cyclical operating-leverage opportunity supported by structural AI demand tailwinds, rather than a wide-moat compounder.

Investor Takeaway: the substrate industry remains defined by pronounced cyclical swings. Kinsus's ongoing earnings recovery is driven primarily by factory utilization and industry-wide supply tightness rather than a fundamental shift in market power. Consequently, its financial trajectory reflects a cyclical earnings stream augmented by structural growth drivers.

↑ Back to Top

VI. Future Material Optionality: Glass Substrates, Quasi-EMIB & CPO [78:00 - 92:00]

Every capital-intensive manufacturer eventually faces the same underlying risk: will current production capacity become obsolete before the underlying capital equipment is fully depreciated? For substrate manufacturers investing heavily in organic build-up facilities, three emerging technologies highlight different trajectories for future material optionality.

Glass substrates: the threat that slowed down

The physical rationale for glass substrates is straightforward. As package dimensions expand past 100 millimetres on a side, organic substrates become increasingly prone to thermal warpage. Glass provides superior dimensional stability, supports finer through-glass vias, and delivers a flatter mounting surface for massive multi-chiplet configurations. If glass becomes the dominant substrate material, manufacturing leadership would naturally shift toward suppliers with specialized glass processing capabilities — a discipline outside the traditional expertise of organic build-up manufacturers.

Kinsus's public position on this transition has evolved. At the March 2026 investor conference, management stated that CoWoS remains the absolute mainstream for advanced packaging over the next three to five years and that glass substrate commercialisation had slowed significantly.¹⁴

While such a perspective aligns with the commercial interests of an organic substrate supplier, it reflects genuine industry hurdles: glass substrates have remained a prospective medium-term technology for several consecutive years due to persistent yield and material handling challenges. Kinsus has allocated research and development resources toward glass core processing to mitigate disintermediation risks, though it has not disclosed specific capital expenditures, development timelines, or customer qualification milestones.

Status: strategic optionality for the second half of the decade, currently deprioritized by management, and lacking sufficient disclosure for independent financial modeling.

The quasi-EMIB collaboration with TSMC

A potentially significant technical development emerged in late July 2026, when reports indicated that 台灣積體電路製造 TSMC is developing an EMIB-like advanced packaging approach internally described as "quasi-EMIB," with Kinsus supporting the effort from development through volume production.²⁰

The underlying engineering concept substitutes large, costly silicon interposers with small silicon bridges embedded directly inside the package to handle high-density die-to-die interconnects, allowing a less expensive organic substrate to manage broader system routing. Intel pioneered this architectural approach with its Embedded Multi-die Interconnect Bridge (EMIB). TSMC's reported initiative serves a defensive purpose, addressing chronic CoWoS capacity shortages at a time when Intel's CEO Lip-Bu Tan has publicly described demand for EMIB-T as very high with a growing backlog and high-volume production targeted for 2027.²⁰

For Kinsus, this architecture represents a potential value shift from foundry-level silicon interposers into the package substrate itself. Embedding silicon bridges inside organic substrates increases component complexity, average selling prices, and product differentiation, raising barriers to supplier substitution.

However, substantial execution caveats remain. Neither TSMC nor Kinsus has confirmed the programme, disclosed the bridge structure, interconnect density, process flow or customers.²⁰ Currently, "quasi-EMIB" represents an unverified development project rather than a commercial platform. Furthermore, peer competitors such as Unimicron maintain established packaging relationships with both TSMC and Intel, indicating Kinsus is unlikely to hold exclusive production rights.

A parallel architectural dynamic could also temper substrate content growth. In the March 2026 Q&A, management discussed how silicon interposers reduce the number of substrate layers required — from roughly thirty down to about sixteen — moderating the growth in substrate complexity.¹⁴ Consequently, increased silicon integration within chiplet packages does not guarantee linear expansion in substrate value per unit.

Status: potentially strategic for core substrate value, but unconfirmed by key parties; represents upside optionality with undefined commercial timing.

Co-packaged optics

The third technological frontier is co-packaged optics (CPO), which integrates optical engines directly alongside switch silicon rather than positioning transceiver modules at the chassis faceplate, significantly reducing electrical signal attenuation across circuit boards.

The primary driver is power efficiency. Transmitting data across multi-hundred-gigabit copper traces over short distances consumes substantial energy. In modern data centers where power availability limits compute density, optical interconnects offer a compelling alternative. However, commercial deployment requires manufacturing, aligning, and testing optical components inside package substrates at high volume without compromising substrate yield.

Management's assessment is that emerging approaches of this kind remain cost-prohibitive and are distant prospects relative to CoWoS and its variants over the next three years.¹⁸

Status: long-term technical optionality, immaterial to near-term earnings, with no active corporate program disclosed.

Strategic conclusion: each of these emerging technologies represents an avenue for long-term monitoring rather than a baseline assumption for near-term valuation. Thin corporate disclosures around next-generation material technology reflect standard industry practices, but they present a structural constraint when attempting to model Kinsus's competitive position beyond 2027.

↑ Back to Top

VII. Management Credibility, Capital Allocation & Governance Stress Test [92:00 - 105:00]

Evaluating a management team in a highly cyclical industry requires examining operational execution across both peak and trough conditions, while assessing whether corporate communications remain consistent as market environments shift.

Who runs the company

The structure of corporate leadership provides clear context regarding parent-group oversight.

Honorary Chairman 童子賢 Tzu-Hsien Tung serves as chairman of Pegatron and is an ASUSTeK co-founder whose career spans Taiwan's electronics evolution from motherboard manufacturing to semiconductor packaging.⁸ His presence on the board reflects strong parent-group alignment.

Chairman 廖賜政 Tsu-Cheng Liao is Pegatron's chief operating officer.⁸ He joined Kinsus's leadership team in 2021 and assumed the chairmanship in August 2022, immediately following the peak quarter as the cyclical downturn began. His background is operational, having held senior leadership roles across Pegatron group companies rather than a dedicated substrate engineering background. Appointing an operational administrator during a capacity collapse reflected Pegatron's focus on working capital discipline and cost control.

CEO 郭明堂 Ming-Tung Kuo serves as executive president, alongside President 陳河旭 Ho-Hsu Chen as general manager.⁸ Dividing executive president and general manager responsibilities across two roles creates a dual structure, though public disclosures offer limited detail regarding their precise separation of operational authority.

The board consists of nine members: six directors and three independent directors, including one female director.⁸ The independent directors possess strong professional credentials, comprising a certified public accountant, a former Minister of Science and Technology, and a Berkeley engineering professor.⁸ With three independent members out of nine on a board chaired by executives closely tied to the controlling shareholder, investors must assess whether this structure provides sufficient independent governance oversight.

The capital allocation record, unvarnished

Kinsus's capital expenditure sequence highlights a distinctive investment trajectory through the recent cycle.

Capital expenditure ran NT$13.4 billion in 2021 and expanded to NT$17.2 billion in 2022 at the peak of the cycle, when corporate profits reached record levels.² Even as consolidated revenue fell 37% during the 2023 downturn, capital spending remained elevated at NT$10.4 billion, and stayed at NT$10.6 billion in 2024 when parent-attributable net earnings dropped to NT$48.9 million.²

Capital outlays declined meaningfully in 2025 to NT$6.4 billion, allowing Kinsus to generate NT$1.7 billion in positive free cash flow — its first positive annual cash flow result since 2020, following negative free cash flow across 2021, 2022, 2023, and 2024.²

This capital allocation record supports two contrasting analytical interpretations.

The constructive interpretation emphasizes structural manufacturing necessity. Advanced substrate facilities require multi-year lead times to build, equip, and qualify. Restricting capital expenditure exclusively to industry peaks leaves a producer unable to capture subsequent demand surges. Kinsus maintained facility expansion through a severe industry downturn, creating the high-end ABF capacity that became fully booked during the subsequent AI packaging recovery.

The critical interpretation highlights balance sheet risk. Spending NT$21 billion across two years in which net earnings attributable to parent shareholders were negligible, funded largely through debt, represented a high-stakes financial strategy. Total debt expanded to NT$18.8 billion by the end of 2023.² Had the AI-driven demand recovery delayed by two years, corporate priorities would have shifted toward debt restructuring rather than capacity expansion.

Consequently, management's capital record demonstrates both strategic conviction and favorable cyclical timing.

The current programme

The board committed to a three-year capital expenditure programme of NT$23.5 billion (approximately US$744 million) focused primarily on expanding ABF capacity.²¹³ In December 2025, the board approved an incremental allocation that raised the 2026 capital budget toward NT$9.5 billion — compared to roughly NT$5.5 billion spent in 2025 — with approximately NT$2.5 billion earmarked for maintenance capex.²² At its March 2026 investor conference, management budgeted approximately NT$8 billion for full-year 2026 capital expenditure, with nearly all funds directed toward Plant 6.¹⁴

The physical expansion aims to add roughly ten million units of monthly ABF substrate capacity — representing a 25% capacity increase — with construction slated for completion during 2026 and formal operations commencing in 2027.²²⁴ Plant 6 comprises two physical facilities: one operating at full capacity, and the second housing the current expansion alongside undeveloped land for future phases.²²

By the second quarter of 2026, factory utilization stood at approximately 85% on ABF production lines and 75% on BT lines, with management targeting utilization rates of 90% and 80%, respectively.⁴¹⁴

The activist's brief

A governance-focused analysis highlights four primary areas of operational and financial scrutiny.

First, the discounted equity raise. In January 2026, Kinsus priced a cash capital increase of 70 million shares at NT$145 — representing a 31.4% discount to the prevailing market price of NT$211.50 — raising over NT$10.15 billion. The subscription period for existing shareholders and employees ran from 9 February to 10 March 2026, followed by a specific-investor tranche.²³ The company designated the proceeds for bank debt repayment.²³

While the rights offering structure protected participating shareholders from economic dilution, three aspects drew scrutiny: the steep issuance discount, the 15% allocation to employees, and the application of proceeds to debt reduction rather than growth projects.²³ Issuing shares at NT$145 in a stock that subsequently traded above NT$800 within six months represented an expensive debt retirement mechanism, raising questions about whether alternative credit facilities were available.

Second, the private placement authorization. On 9 April 2026, the board authorized a potential private placement of up to 100 million common shares — representing nearly 20% of the enlarged share count — restricted to strategic partners to establish technology cooperation or strategic alliances with industry majors while supplementing working capital.²⁴

Securing an anchor customer or technology partner through an equity stake offers strategic value in the substrate industry. However, Taiwanese private placements can be priced at discounts to market prices and carry transfer restrictions rather than pre-emptive rights for existing shareholders. Granting authorization for substantial equity dilution shortly after a discounted rights issue warrants ongoing monitoring. As of August 2026, no strategic counterparty has been announced.

Third, the Pegavision consolidation. Maintaining a contact lens subsidiary within an industrial semiconductor substrate manufacturer presents a structural anomaly. While Pegavision remains profitable, its inclusion obscures standalone substrate gross margins, demands executive oversight, introduces consumer price competition in China, and allocates a substantial share of consolidated profits to non-controlling interests.¹⁰ Pegavision's gross margin compressed from roughly 56% to 52% in 2025 due to New Taiwan Dollar appreciation and pricing pressure in China, with every 1% currency fluctuation impacting gross margin by 0.4 to 0.5 percentage points.¹⁰ Management has provided limited strategic rationale for retaining the subsidiary beyond its net profit contribution.

Fourth, narrative consistency and disclosure standards. On management transparency, executive communications demonstrate notable consistency. Management's August 2025 projection — that general substrate supply would remain in surplus while high-end ABF capacity would tighten during 2026 — proved accurate.¹² Furthermore, management acknowledged during the first-quarter 2026 call that selling price increases were absorbed by raw material inflation and that gross margin gains derived from utilization rather than pricing power.¹⁸ Quantifying that raw material shortages constrained overall revenue expansion by 10% to 15% provided clear operational detail.¹⁸

Conversely, management offered limited public accountability regarding capex timing during the 2023–2024 trough, and disclosure remains focused on broad market demand rather than granular yield or unit cost metrics. Kinsus does not routinely publish manufacturing yield rates, average selling prices, or standalone substrate segment margins.

A review of corporate conference calls reveals a clear distinction between prepared remarks and Q&A exchanges. Prepared statements during the March and April 2026 sessions emphasized market tailwinds across artificial intelligence, high-performance computing, and memory recovery, projecting sustained growth through 2030 as workloads transition toward inference.¹⁴¹⁸ In contrast, Q&A sessions focused on operational constraints, detailing supply bottlenecks in T-glass cloth, input diversion of copper foil and drill bits to server circuit board makers, and cost inflation across labor, utilities, and solder mask materials.¹⁸ Management addressed these operational challenges directly rather than attributing outcomes to macroeconomic conditions.

Additionally, when asked about alternative packaging architectures, management acknowledged trade-offs that temper the substrate thesis, noting that silicon interposers maintain thermal advantages and reduce required substrate layer counts from roughly thirty to sixteen.¹⁴ Such disclosures reflect a pragmatic executive assessment.

So what: Kinsus features an operational management team executing a capital-intensive manufacturing plan within a corporate governance structure that grants its controlling shareholder broad latitude over equity financing. The capital allocation record demonstrates sustained investment through an industry downturn — a strategy that created available capacity for the AI recovery, alongside a willingness to issue discounted equity and seek authorization for additional shares.

---

↑ Back to Top

VIII. Investment Spine, Risk Radar & Core KPIs [105:00 - 118:00]

Stripping away the narrative leaves a central question: is Kinsus a structurally advantaged supplier at a permanent chokepoint, or a capable follower whose earnings are being flattered by a cyclical shortage that will eventually end?

Why Kinsus wins from here

The surface-area and layer-count dynamics are structural. Artificial intelligence packages consume far more substrate area than legacy components, and layer counts are climbing along a clear trajectory — scaling from roughly three build-up layers per side toward eleven and eventually thirteen.⁹ Each added layer requires additional processing steps, increases component value, and absorbs industry line capacity, because a twenty-layer board occupies a production line significantly longer than a ten-layer board. Consequently, demand for substrate surface area is expanding faster than unit demand for silicon dies.

Customer qualifications are recent and established. Reported supplier engagements with Google's TPU and Meta's AI accelerator programs — alongside expected supply for Amazon Web Services in the second half of 2026 — position Kinsus inside the custom application-specific integrated circuit wave rather than solely within the merchant GPU market.¹⁹ Custom silicon programs offer greater incumbency than merchant platforms because chip architectures are co-developed and qualification options remain limited.

The BT substrate line provides structural support. The memory and mobile business is not a legacy drag in this cycle; it functions as a second earnings driver, benefiting from upstream material shortages that lifted BT pricing, with memory clients proving most receptive to price adjustments.¹⁴¹⁵ While a large non-AI segment typically incurs a valuation discount, in this cycle it has provided underlying revenue support.

The financial recovery follows a clear operational sequence. Line utilization recovered first, followed by gross margin, operating margin, and parent-attributable net income — underscored by second-quarter operating margin expanding 5.4 percentage points sequentially to 13.4%.⁴ This progression reflects classic operating leverage against a fixed cost base, characteristic of a manufacturer emerging from an overbuilt capacity expansion.

Why Kinsus may not win

The scale gap remains unbridged. Market leader Unimicron generates more than three times Kinsus's revenue, while Ibiden maintains dominance in high-spec processor packaging.¹⁶ During allocation-driven shortages, scale determines which manufacturer secures scarce raw materials and lands flagship chip sockets. Kinsus is a beneficiary of the current supply tightness rather than its primary architect.

Management has explicitly disclaimed structural pricing power. Executive disclosures confirm that selling price increases have been largely absorbed by rising input costs, with margin expansion driven primarily by higher factory utilization.¹⁸ Because utilization is the most volatile variable in substrate manufacturing, a margin profile dependent on factory loading can reverse quickly during a demand slowdown.

Capital intensity keeps free cash flow constrained. Annual depreciation reached NT$7.24 billion in FY2025 — representing nearly 18% of revenue — and the current expansion program expands that fixed depreciation base before generating incremental revenue.² Even during a strong 2025, free cash flow reached just NT$1.7 billion. If hyperscaler capital outlays pause while Plant 6 depreciation comes online, operating leverage will operate in reverse.

Geographic concentration is high. Following the divestment of its Suzhou facilities, Kinsus's manufacturing operations are concentrated in Taoyuan and Hsinchu.¹¹ While non-China production aligns with Western customer preferences, it creates a single-location operational risk regarding seismic events, regional geopolitical tension, and local power and water supply constraints.

Input material dependency is absolute. Kinsus remains entirely dependent on Japanese suppliers for ABF film and T-glass cloth, lacks viable secondary supply options, and has had to absorb 30% supplier price increases while passing along what the market allows.⁹¹⁵

Risk radar

• AI capital expenditure digestion. This represents the primary downside risk. Cloud provider spending drives end-market demand. On the first-quarter 2026 earnings call, management explicitly identified whether cloud service provider capital expenditure translates into sustainable revenue growth as the key variable to monitor.¹⁸ A spending digestion pause occurring alongside newly commissioned 2027 capacity represents the primary bear case.
• Raw material single-sourcing. Concentrated input supply acts as a production constraint during upturns and a margin compressor during downturns.
• Technological substitution. Risks include glass core substrate adoption or architectural shifts that integrate more interconnect density into silicon interposers rather than organic package substrates. Management's observation that silicon interposers can reduce required substrate layers from thirty down to sixteen highlights this structural trade-off.¹⁴
• Financing and equity dilution. The outstanding board authorization for a private placement of up to 100 million shares remains a potential dilutive overhang.²⁴
• Currency exposure and Pegavision consolidation. New Taiwan Dollar appreciation creates headwinds for substrate export revenues and compresses margins at the consolidated contact lens subsidiary.¹⁰
• Industry-wide capacity oversupply. Substrate competitors are expanding simultaneously, with Unimicron, Ibiden, and Kinsus all bringing high-layer capacity online for 2027 delivery.¹⁸ Historically, the substrate industry has struggled to add capacity without eventually creating supply surpluses.

The three KPIs that actually matter

Evaluating Kinsus's ongoing operational performance focuses on three core metrics:

1. ABF capacity utilization and its share of substrate revenue. Utilization is the primary driver of profitability, given management's confirmation that margin gains stem from factory loading rather than pricing power.¹⁸ The disclosed utilization levels — ABF at roughly 85% targeting 90% to 95%, and BT at 75% targeting 80% — represent the operational baseline.⁴¹⁴¹⁹ Monitoring factory loading as Plant 6's new lines come online in 2027 will determine whether new capacity is absorbed or depresses margins.

2. Standalone substrate gross margin, adjusted for Pegavision. Headline group margins are elevated by Pegavision's 52% gross margin contact lens business. Evaluating core substrate economics requires backing out the consumer optics segment, tracking performance from mid-teens trough levels through 2026 as product mix and utilization improve.

3. Capex-to-revenue efficiency against incremental revenue delivered. Kinsus has committed NT$23.5 billion over three years to expand monthly ABF capacity by roughly 25%.²¹²² The key performance test by 2028 is whether this additional capacity achieves high utilization at profitable margins. Sustainable positive free cash flow — following four consecutive years of cash deficits prior to 2025 — serves as the primary measure of capital efficiency.

Kinsus occupies a functional position within an industry supply shortage, backed by executive communications that clearly detail the operational sources of margin recovery. However, the company lacks pricing power, scale leadership, and input material control. Those structural conditions define the investment profile.

↑ Back to Top

IX. Playbook: Business & Investing Lessons [118:00 - 125:00]

Twenty-six years after an ASUSTeK spin-off set out on the premise that package substrates were simply miniaturized circuit boards, the long-term arithmetic of that bet has resolved — though not entirely as originally envisioned. Kinsus has not emerged as the industry's primary technology pioneer. Instead, it operates as a capable supplier of scarce manufacturing capacity in a market where physical tightness dictates short-term returns.

Three broader business and investing lessons emerge from its trajectory.

Lesson 1: Chokepoints are worth finding — and worth pricing carefully

The most valuable position in a technology supply chain is rarely the most visible one. It is the physical bottleneck through which every finished product must pass and which cannot be expanded quickly. Package substrates represent such a bottleneck. One step upstream, ABF resin film presents an even narrower one. Another step further, T-glass cloth creates a third.

However, the Kinsus story highlights a critical asymmetry within supply chain bottlenecks. Raw material suppliers such as Ajinomoto and Nittobo occupy near-monopoly positions, allowing them to implement 30% price increases and force downstream manufacturers to absorb or pass along the higher input costs.⁹ By contrast, Kinsus sits one step downstream among five or six established substrate competitors. It captures scarcity rents during supply shortages, but faces severe margin compression when market capacity normalizes. Both positions function as bottlenecks, but only the upstream material monopolists possess durable pricing power.

For investors, evaluating a bottleneck requires measuring how many alternative suppliers exist rather than assessing component necessity alone. The relevant question during supply tightness is not whether the industry can function without a given vendor — during a shortage, the answer for almost every supplier is negative. The critical test is whether a vendor retains pricing power once market capacity normalizes. For Ajinomoto and Nittobo, pricing leverage persists across cycles. For substrate manufacturers, pricing power has historically proven temporary, and recent operating disclosures indicate that fundamental dynamic remains unchanged.

Lesson 2: Conglomerate parentage is a floor, not an engine

Pegatron's majority control provided Kinsus with financial stability, offering balance-sheet backing through four consecutive years of negative free cash flow and supplying operational leadership during cyclical downturns.

However, parent-group backing did not confer manufacturing scale, technology leadership, or automatic access to high-value chip platforms. Those achievements required independent commercial execution against established Japanese competitors. Furthermore, conglomerate ownership introduces trade-offs that directly affect minority shareholders: equity financing decisions controlled by the parent entity, a consolidated consumer-optics subsidiary retained without clear operational synergy, and corporate disclosures structured around group reporting preferences.

This dynamic illustrates a broader corporate pattern: a strong parent company provides a safety net against downside risks, but rarely serves as an engine of operational outperformance, while fundamentally altering governance authority.

For investors screening group-affiliated technology suppliers, the key analytical question is not whether the parent offers financial support — parent entities typically do. The vital test is whether the subsidiary has built a commercial customer base that survives independently of parent relationships. On that metric, Kinsus demonstrates genuine commercial independence: its substrate customers consist of independent fabless chip designers operating in competitive open markets rather than captive group affiliates.

Lesson 3: The capex paradox is real, and it is not the same as capex courage

A standard industry maxim dictates that substrate manufacturers must expand production capacity during cyclical troughs to capture demand during subsequent peaks. Kinsus's recent history provides a clear example: the company deployed over NT$50 billion in capital expenditures between 2021 and 2024 despite sharp profit declines, creating the high-end ABF capacity that subsequently became fully booked.²

However, the second half of that capital allocation dynamic is less recognized. Investing through a severe industry downturn is vindicated only if end-market demand recovers before balance-sheet leverage creates financial distress. In Kinsus's case, the AI packaging recovery arrived roughly two years after the cyclical contraction began — a relatively swift turnaround driven by accelerator demand that was unmodeled during initial plant planning. A delayed market recovery would have created severe debt service pressures and potentially forced structural reorganization.

Consequently, the core lesson extends beyond committing capital during downturns. In capital-intensive industries where manufacturing expansion requires three years to construct and qualify, capital allocation decisions must be underwritten against defensible demand projections while maintaining a balance sheet resilient enough to withstand extended cyclical delays. Kinsus navigated the downturn successfully, but doing so required issuing equity at a steep discount to manage debt obligations — illustrating the real costs of balance-sheet survival.

For observers tracking the broader AI infrastructure expansion, this operational reality provides an essential analytical frame. While silicon accelerators capture public attention, the underlying physical constraints bind within resin films, specialized glass fabrics, and multi-layer package substrates — where commercial outcomes are governed by yield progression, capacity allocation discipline, and the financial resilience to sustain capital commitments through cyclical lows.

References

1. Kinsus Interconnect Technology Corp. — Official Corporate & IR Portal ↩↩

2. Kinsus Interconnect Technology Corp. — Financial Information and Annual/Quarterly Reports ↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩↩

3. Kinsus Interconnect Technology's 2025 Profit Soars Over 31-Fold; Company Allocates NT$23.5 Billion for ABF Substrate Capacity Expansion — BigGo Finance ↩↩↩↩

4. 景碩財報／第2季獲利衝上去年同期近四倍 EPS 2.57元、毛利率26.1% — 經濟日報/聯合新聞網, 2026-07-31 ↩↩↩↩↩↩

5. Kinsus Interconnect Technology Corp. — Monthly Revenue Report ↩↩

6. Kinsus Interconnect Technology Corp (3189.TW) Financial Overview — Reuters ↩↩

7. 景碩科技 — 維基百科 ↩↩↩↩↩↩↩

8. 公司治理 — 景碩科技 董事會成員 (Board of Directors) ↩↩↩↩↩↩↩

9. Ajinomoto, Controlling 95% of ABF Film Market, Hikes Prices 30% as AI Chip Packaging Upgrades Spark Substrate Cost Surge — BigGo Finance, 2026-05-14 ↩↩↩↩↩↩

10. 營收連7年創高，獲利卻下滑？晶碩(6491) 2026 年轉強訊號看什麼？ — 鉅亨網, 2026-02-11 ↩↩↩↩↩↩

11. PCB產業「中國減一」 景碩賣蘇州廠吹號角 — 自由財經, 2025-01-12 ↩↩

12. IC substrate market split: Kinsus sees adequate supply, high-end ABF tight by 2026 — DIGITIMES, 2025-08-21 ↩↩

13. AI chip rivalry escalates: ABF substrate sells out for Unimicron, Kinsus, Nan Ya PCB — DIGITIMES, 2026-04-20 ↩

14. 2026/03/10 景碩 (3189.TW) 法說會：AI與記憶體需求強勁，2026年營收與獲利重返成長 — vocus, 2026-03-10 ↩↩↩↩↩↩↩↩↩↩↩↩↩↩

15. Shortages of crucial chip packaging material threatens AI accelerator supply chains — Nittobo's Fukushima plant is tripling capacity — Tom's Hardware ↩↩↩↩

16. Unimicron Technology Corp (3037.TW) — Financial Overview, Reuters ↩↩↩↩

17. Nan Ya Printed Circuit Board Corp (8046.TW) — Financial Overview, Reuters ↩↩↩

18. 【Kinsus FY2026 Q1 Earnings Call】Undercurrents Beneath the AI Demand Frenzy — BigGo Finance, 2026-04-07 ↩↩↩↩↩↩↩↩↩↩↩↩↩

19. 【追蹤報導】ABF漲價潮引爆！景碩(3189)飆漲超越目標 2026獲利翻倍可期 — 理財周刊, 2026-04-24 ↩↩↩↩

20. [News] TSMC Reportedly Develops EMIB-like Packaging with Kinsus to Prevent Potential Customer Shift to Intel — TrendForce, 2026-07-31 ↩↩↩

21. Kinsus approves US$744M ABF expansion on clear three-year demand outlook — DIGITIMES, 2026-02-02 ↩↩

22. 景碩明年資本支出上看95億 加碼六廠ABF擴產 — MoneyDJ理財網, 2025-12-26 ↩↩↩↩

23. 景碩現增籌資逾百億元定價145元 現股價差逾三成 — 經濟日報, 2026-01-19 ↩↩↩

24. 董事會：景碩(3189)決議辦理私募普通股，額度不超過1億股 — 富聯網, 2026-04-09 ↩↩