1303 / 1303

Nan Ya Plastics Corporation (南亞塑膠工業股份有限公司, 1303.TW, Taiwan Stock Exchange)

WIKI tickertaiwanmaterialsfiberglassglass-clothai-infracclpetrochemdram-proxy updated 2026-07-07

#stock #deepdive #fiberglass #glass-cloth #ai-ccl #petrochem #dram-proxy #taiwan #investment #analysis

🔴 Live NT$226 (yf:1303.TW) · 2026-09-05 · research written 2026-07-01 Consensus NT$26–NT$350 (mean NT$202, 12mo, Yahoo consensus (7 analysts)) · 2026-09-05 your re-engage <NT$106 (sell-side, on-pullback) — +112% away

The world's largest glass-fabric weaver by volume, wrapped inside a commodity-petrochem balance sheet and a roughly 29% DRAM stake. The AI glass-cloth "bottleneck" the market is re-rating is about 3% of revenue, and even that sliver runs on Nittobo's (3110.T) yarn. You are paying an AI-materials multiple for what is, at its core, a DRAM-cycle proxy.

Sector context: the industry-wide material (the glass grade ladder, the copper-clad-laminate chain, the Dk/Df/CTE physics, the full player map) lives on the sector page advanced-packaging. The listed glass-cloth siblings are 3110 (Nittobo, the T-glass and NER-yarn monopoly, the rung Nan Ya rents), 1815 (Fulltech, the NE-class M7 second source), 5475 (Glotech, the Q-glass M9 frontier) and, one layer over, 5706 (Mitsui Kinzoku, HVLP copper foil). This page is Nan Ya-specific and links up rather than re-hosting the primer. The single most important cross-name fact is altitude. Nan Ya owns the weaving step and the CCL chain, not the scarce yarn chemistry; the bottleneck sits one rung above it and is owned by others.

Disambiguation (read first). Three distinct "Nanya / Nan Ya" entities recur in the AI supply chain and must not be conflated. Nan Ya Plastics (1303.TW) is this company, the glass-fabric weaver, CCL producer and petrochem conglomerate. Nan Ya PCB (8046.TW) is its separately listed PCB and ABF/IC-substrate affiliate (about US$2.5B cap), downstream of 1303. Nanya Technology (2408.TW) is the DRAM maker in which 1303 holds about 29.3%; that stake, not glass cloth, is the engine of 1303's reported profit swing. SemiAnalysis's glass-fabric coverage centers on Nittobo (3110.T); its "Nanya" hits refer to the PCB and DRAM entities, not to 1303's glass business.

Price NT$182.5 (2026-07-01 TWSE close, live-verified via yfinance 1303.TW). Market cap about NT$1,447B (roughly US$45 to 47B) on 7.931B shares. Every multiple below scales with the live price; reconcile price, share count and cost basis live in IBKR (book of record) before acting. Directional verdict: PASS as a glass-cloth bottleneck play; at best a leveraged DRAM-cycle beta, and cheaper, cleaner ways exist to own either underlying. Management DD: YELLOW.

PART I. THE BUSINESS

1. Executive summary

Nan Ya Plastics is a Formosa-group commodity plastics and petrochemicals conglomerate that the market has re-rated roughly sevenfold in twelve months (NT$26.25 low to NT$182.5) on an "AI materials plus DRAM recovery" narrative. The narrative is real but mislabeled. The earnings inflection driving the stock is a DRAM up-cycle flowing through Nan Ya's roughly 29% equity stake in Nanya Technology (2408.TW), not organic glass-cloth economics. In Q1 2026, operating profit was NT$3.74B while pre-tax income was NT$16.08B; the roughly NT$12.3B gap (about 77% of pre-tax) is equity-method and non-operating income, overwhelmingly the Nanya Technology DRAM turn plus Formosa Petrochemical (FPCC).

The glass-cloth exposure that ties this name to the AI PCB bottleneck is roughly 3% of company revenue and low-single-digit percent of profit. On that sliver, Nan Ya is a standard E-glass volume leader plus a contract weaver of Nittobo's second-generation low-Dk (NER) yarn, not an independent owner of the scarce high-end grades. It is a genuine beneficiary of the industry-wide glass-fabric shortage as a volume producer and a repeat E-glass price-raiser through 2026. It is not a holder of the low-Dk / low-CTE / quartz bottleneck.

Conviction is low as a glass-cloth bottleneck vehicle. This is a diluted, cycle-levered conglomerate proxy, not a bottleneck pure-play. If the thesis is the glass-fabric bottleneck, the direct owners are the cleaner expressions: Nittobo (3110.T) for T-glass and NER yarn, Fulltech (1815.TWO) for NE/M7, Glotech (5475.TWO) for quartz/M9. If the thesis is actually DRAM, own Nanya Technology (2408.TW) directly rather than through a roughly 29% look-through wrapped in commodity plastics.

2. Corporate overview

Nan Ya Plastics is one of the four core pillars of the Formosa Plastics Group (FPG, 台塑集團), alongside Formosa Plastics (1301.TW), Formosa Chemicals and Fibre (1326.TW), and Formosa Petrochemical (6505.TW). It is a diversified downstream processor spanning plastics, polyester fiber, commodity chemicals, and electronic materials, and it sits at the center of an intricate web of Formosa cross-holdings whose mark-to-market rides the semiconductor and petrochemical cycles.

Revenue by segment (Q1 2026, confirmed against Nan Ya's reported breakdown):

Segment % of revenue Character
Electronic materials 52.8% The "AI" bucket, but mostly PCB, CCL, copper foil and epoxy, not glass cloth
Chemicals 16.3% Commodity petrochem (EG, phenol); price-taker
Polyester 15.6% Commodity fiber and PET; price-taker
Plastics 13.3% Commodity PVC, film, pipe, PU, synthetic leather; price-taker
Other 2.0%

Electronics has genuinely mix-shifted, from roughly 42.8% of revenue a year earlier and about 46% across FY2025 to above 50% in Q4 2025 and about 52.8% by Q1 2026. That shift is driven by PCB, CCL, copper foil and epoxy demand, not by glass cloth.

How small the AI-glass slice actually is. Decomposing the electronics bucket (MoneyDJ / CMoney, approximately FY2025):

Sub-product Approx % of electronics Approx % of total company revenue
PCB ~30% ~15.8%
CCL ~20% plus ~11%
Epoxy resin ~17% ~9%
Copper foil ~14 to 15% ~7.5%
Glass fabric (玻纖布) ~5 to 6% ~2.9%

The scarce-grade AI glass-cloth exposure is therefore roughly 3% of company revenue and, given its small volume even at high specialty margins, plausibly only about 1 to 3% of total earnings. Everything else in the electronics segment (PCB, CCL, copper foil, epoxy) is a broad, competitive pool in which Nan Ya is one of many credible suppliers, not a bottleneck holder. Put plainly, you are buying roughly 95%-plus commodity petrochem plus a DRAM proxy in order to get about 3% glass-cloth beta.

3. First principles: glass and CCL, as it applies here

An AI server printed circuit board is a stack of copper-clad laminate (CCL) layers. Each laminate layer is woven glass fabric impregnated with resin (prepreg) and bonded to copper foil. As per-lane SerDes rates on AI accelerator fabrics move toward 224 Gbps, two glass properties become gating.

The first is dielectric performance, the dielectric constant (Dk) and loss (Df). Lower Dk and Df mean faster, cleaner high-frequency signal transmission with less loss, which is why AI boards must upgrade CCL grade from conventional M7 to M8 and M9. Achieving low Dk/Df requires a special glass chemistry (NE-glass and its low-Dk NER derivative), and that chemistry is set at the melt-and-spin (yarn) step, not at weaving.

The second is thermal expansion, the coefficient of thermal expansion (CTE). Very large AI packages need low-CTE glass (T-glass) so the board stays dimensionally stable through thermal cycling and does not warp against the silicon. The extreme rung is quartz (Q-glass), used where both Dk and CTE must be pushed furthest, on the most demanding next-generation platforms.

The first-principles point for Nan Ya is that the difficulty, and therefore the scarcity, lives in the yarn: the glass melt and chemistry that produce low-Dk NER, low-CTE T-glass, and quartz fiber. Weaving that yarn into fabric is the more replicable, capacity-bound step. Nan Ya's scale and skill sit predominantly at the weaving step and downstream in CCL, prepreg, copper foil and PCB. It owns the full stack only for commodity E-glass, where it makes its own yarn. For the AI-grade rungs, it accesses the scarce chemistry by buying Nittobo's NER yarn under contract. That is the whole distinction between owning a bottleneck and renting access to one.

4. Product and segment detail

Within electronic materials, Nan Ya is vertically integrated across the laminate stack. It produces epoxy resins (running near 100% utilization), copper foil (about 80%), CCL and prepreg, glass fabric, and, through affiliate Nan Ya PCB (8046.TW), finished PCBs and IC substrates. In late 2025 it pushed an 8% price increase across its entire CCL and prepreg line to pass through raw-material costs, and glass fabric has been its fastest-rising product, raised repeatedly into the 2026 shortage.

The glass-fabric line itself splits into industrial-grade and electronic-grade cloth. The bulk is standard E-glass, where Nan Ya is the global volume leader. The high-end entry is narrow and contracted: a reciprocal arrangement with Nittobo (3110.T) in which Nittobo supplies Nan Ya long-term NER (second-generation low-Dk) yarn, and Nan Ya weaves about 20% of Nittobo's global specialty-fabric supply by 2027. Nan Ya is simultaneously Nittobo's customer for yarn and its junior contract weaver for finished specialty cloth. It is not an independently certified low-CTE or low-Dk competitor. (The specific "no independent low-CTE certification" claim is consistent across Taiwan press and the sibling coverage but is not confirmed to a primary spec sheet; flag as unverified.)

5. Value-chain position and upstream bottleneck check

The scarce input for AI-grade CCL is specialty yarn, and Nan Ya does not own it. Its high-value-add on the specialty side is downstream of the choke point.

UPSTREAM  (the bottleneck: melt + glass chemistry)          NAN YA'S TURF                         DOWNSTREAM
silica sand / boron / batch
        |
        v
  glass melt + yarn spinning
   |            |               |                  |
   v            v               v                  v
 E-glass    low-Dk NER/M8   low-CTE T-glass    quartz Q/M9
 yarn       yarn            yarn               yarn
 (Nan Ya    (Nittobo        (Nittobo ~90%)     (Glotech 5475,
  owns)      ~60-70%)                            AGY)
   |            \
   |             \  (Nan Ya BUYS NER yarn under contract)
   v              v
 weaving / glass fabric  <----------------------------------- Nan Ya weaves ~20% of
 (Nan Ya = largest weaver by volume)                          Nittobo specialty cloth by 2027
        |
        v
 prepreg  -->  CCL  -->  PCB / IC substrate  -->  AI server board
 (Nan Ya integrated: epoxy, copper foil, CCL; Nan Ya PCB 8046.TW)

Upstream bottleneck check. For the AI grades, the bottleneck sits above Nan Ya and is owned by others. Low-Dk NER yarn and low-CTE T-glass yarn are Nittobo's (about 90% T-glass, 60 to 70% NER, and the sole source of M8 NER yarn); quartz yarn is Glotech (5475.TWO) at scale plus AGY. Nan Ya's exposure to the scarce rung is external and singular: it depends on Nittobo's yarn. For E-glass, Nan Ya is fully integrated and owns its melt, so there is no upstream dependency, but E-glass is not the scarce chemistry; the E-glass "shortage" is an aggregate melt-and-weave capacity constraint, and Nan Ya benefits from it as a volume price-raiser, not as a technology gatekeeper. The net read is that Nan Ya passes the bottleneck check only in the weakest sense. It sits downstream of the choke point on the specialty side and holds no proprietary upstream position at the tip of the ladder.

Grade-ladder positioning (honoring precision versus the siblings):

Rung Grade Gating property Independent owner Nan Ya's position
Base E-glass none (commodity) many Global volume leader; fully integrated, owns yarn
2 T-glass low CTE Nittobo (3110.T) ~90% No independent cert; Fulltech (1815.TWO) is the TW second source
3 NE / NER (M7/M8) low Dk/Df Nittobo NER (sole M8); Fulltech at M7/NE Access only as Nittobo's contract weaver on Nittobo's NER yarn
4 Q-glass (M9) lowest Dk plus CTE Glotech (5475.TWO), AGY No position

For contrast, Fulltech (1815.TWO) is the cleanest Taiwan NE/M7 pure-play plus a T-glass second source, Glotech (5475.TWO) is the sole Taiwan quartz/M9 maker at scale, Nittobo (3110.T) owns the T-glass and NER yarn (the melt), and Mitsui Kinzoku (5706.T) sits on a different layer entirely (HVLP copper foil). Nan Ya is the E-glass volume king and a downstream CCL/PCB integrator, which is a real and defensible position, just not the one the AI-bottleneck thesis is pricing.

6. Key customers

Nan Ya's glass fabric is consumed primarily inside its own stack, feeding its captive CCL and prepreg lines and, downstream, affiliate Nan Ya PCB (8046.TW). Beyond captive use, it sells merchant glass fabric and CCL into the broader Taiwan and China PCB supply chain. On the specialty side, the roughly 20% of Nittobo cloth it weaves by 2027 is effectively contracted into Nittobo's order book rather than sold on Nan Ya's own account, so Nittobo (3110.T) is both a supplier and, functionally, the offtaker for that capacity. Nan Ya also appears as one of Nittobo's own CCL customers, alongside Doosan, Panasonic, Shengyi, ITEQ and Elite, which underscores that on the high-end laminate it is a buyer of Nittobo material, not a peer at the yarn tier. The ultimate end-demand is AI accelerator and switch boards for hyperscale servers; Nan Ya's participation there is broad and indirect through commodity-competitive CCL, not privileged through a scarce input.

7. TAM

The relevant markets sit at three different altitudes, and the size falls sharply as you climb toward the bottleneck.

The broadest is the copper-clad laminate market, roughly US$16.0B in 2025 and forecast toward US$21.5B in 2026 on AI-driven specification upgrades, growth of up to about 34% year on year. This is where Nan Ya has wide participation, but as one of many CCL makers with no pricing moat.

One rung up in scarcity is electronic-grade glass fabric, and higher still the low-Dk/Df specialty cloth that AI boards actually gate on. Estimates for the low-Dk/Df electronic glass cloth market vary widely by source, roughly US$0.47B to US$1.06B in 2025, with forecasts ranging from about US$0.7B to US$1.6B by 2032 at mid-single-digit CAGRs, and one aggressive outlier projecting about US$0.53B in 2026 rising to about US$4.1B by 2035 (about 25.6% CAGR). The dispersion itself signals an early, small, and imperfectly measured market. Supply is concentrated: only five to seven producers globally make electronic glass fabric at all (Nittobo, Nan Ya, Asahi Kasei, Taiwan Glass, AGY, PFG, Fulltech and a few China names), and the specialty weave styles for thin cores are production-constrained, with NE-glass and T-glass having few qualified sources.

The load-bearing implication for Nan Ya is a matter of arithmetic. The AI-grade bottleneck TAM is small in absolute dollars, well under US$1B to US$2B today, against Nan Ya's roughly US$8B revenue base. Even a maximal capture of the scarce glass-cloth prize would move Nan Ya's consolidated top line only modestly, which is exactly why a roughly 3%-of-revenue exposure cannot, on its own, justify a sevenfold re-rate. The re-rate is being paid for by the DRAM cycle, not by glass.

8. Sector inflection: why now

Two inflections are running at once, and the market has fused them into a single "AI materials" story.

The first is the glass-fabric shortage. As AI server fabrics push to 224 Gbps per lane, CCL grade must upgrade from M7 to M8 and M9, and AI motherboard layer counts are climbing to 24 to 40 for 2026 to 2027. That drives demand for low-Dk2/Dk3 and low-CTE cloth into a supply base of only five to seven qualified weavers with multi-year yarn-qualification cycles. The scarcity has spilled down the ladder: even standard E-glass fabric is short, and prices are rising repeatedly through 2026. Nan Ya has raised E-glass fabric prices multiple times this year and continues price discussions with clients, while rival Kingboard raised prices 15% again in one week of June 2026. Nittobo is second-sourcing its specialty supply by design toward mid-2027, and Nan Ya is the largest such second source, pulled in as contract weaver. Nan Ya plays this inflection genuinely, but as a volume beneficiary of the E-glass squeeze and as Nittobo's junior weaver, not as an owner of the scarce chemistry.

The second, and the one actually moving 1303's earnings, is the DRAM up-cycle. Nanya Technology (2408.TW) posted Q1 2026 revenue up 582.9% year on year to NT$49.1B, roughly NT$26B of net profit, about 68% margins, on DRAM price increases above 70%. That flows into Nan Ya Plastics as equity income and explains why Q1 2026 pre-tax income (NT$16.08B) so vastly exceeds operating profit (NT$3.74B). The "why now" for the stock is therefore predominantly a DRAM-cycle "why now," with the glass-fabric shortage supplying the narrative color that lets the market file a commodity petrochem conglomerate under "AI materials."

The honest framing to carry forward is that Nan Ya is the largest glass-fabric maker by volume, but that leadership is in standard E-glass; Nittobo owns the high-end AI-grade rungs. "Largest by volume" must not be misread as leadership in the bottleneck. It is not.

PART II. MANAGEMENT AND GOVERNANCE

Nan Ya is not run as a glass-cloth company or even as a standalone company. It is one of the four listed pillars of Formosa Plastics Group (FPG, 台塑集團), governed collectively through the group's Management Center, controlled through an interlocking web of cross-holdings and a hospital foundation rather than through large family equity stakes, and steered by an aged, professionalized leadership whose mandate is group perpetuation, not maximizing the per-share economics an outside minority holder cares about. The books are clean and the operators are competent. The alignment is the problem.

1. Leadership

Chairman Wu Chia-chau (吳嘉昭) runs Nan Ya, and as of 15 August 2025 he was elevated to President (總裁) of the entire FPG Administrative Center (行政中心), the first non-Wang-surname person ever to lead the group. Wang Wen-yuan (王文淵), the founder's nephew and the group's leader since 2006, stepped back to chair the oversight-level Management Center (管理中心). The elevation is genuinely a professionalization milestone and reads as a positive on paper. The caveat is age and concentration: this was, in the Taiwanese press's own framing, a 78-year-old handing to an 81-year-old, and Wu now simultaneously chairs Nan Ya and presides over the whole conglomerate. This is entrenched, late-career stewardship, not a fresh bench.

President Ming-Jen Tzou (鄒明仁) runs day-to-day operations and doubles as the Nan Ya-appointed chairman of Nanya Technology (2408.TW), the DRAM affiliate that drives most of Nan Ya's reported profit. That single overlap tells you how the org actually works: the people running Nan Ya are the same people running the affiliates whose equity income is the earnings engine, so the "operating company" and the "holding company" are staffed by one team.

The governance architecture sits above all of this. FPG formalized a split of ownership and management in June 2017, moving the four senior Wang family members (Wang Wen-yuan, Wang Rui-hua, Wang Wen-chao, Wang Rui-yu) up into the Management Center and leaving the five operating-company chairmen to run the Administrative Center. The doctrine binding it is "永不分家" (never divide the house), an explicit commitment to keep the group intact and cross-held in perpetuity. For a minority buyer, that doctrine is the single most important governance fact: the controlling objective is the survival and cohesion of the group, and per-share value at any one pillar is subordinate to it.

2. Ownership and skin in the game

Control and economic ownership are deliberately decoupled here. The Wang family's direct personal stakes in Nan Ya are small: individual family directors hold well under 1% each (Wang Wen-yuan roughly 0.47%, Wang Rui-yu roughly 0.24%, Wang Gui-yun roughly 0.14%). There is no founder or family member with a large personal economic stake whose interests track an outside shareholder's. Skin in the game, in the ordinary sense of insiders eating their own cooking on a per-share basis, is thin.

Instead, control runs through corporate and foundation blocs. On Nan Ya's register (most recent detailed vintage August 2023, structure stable since):

Holder Shares % of 7.931B
Chang Gung Medical Foundation (長庚醫療財團法人) ~876.7M ~11.1%
Formosa Plastics (1301.TW) ~783.4M ~9.9%
Formosa Chemicals and Fibre (1326.TW) ~413.3M ~5.2%
A further Formosa group affiliate (label ambiguous, likely Formosa Petrochemical (6505.TW) or a private group entity) ~179.2M ~2.3%

The identifiable FPG-plus-foundation bloc is therefore roughly 28 to 30% before counting smaller undisclosed affiliates, comfortably enough to control the board of a NT$1.4T company. Registered director / insider holding is cited around 18.2%, which is the corporate-director seats plus family fractions; the Chang Gung Foundation's roughly 11% sits on top as the effective family control vehicle. Note that the pledge ratio is essentially zero across directors and corporate holders (only a minor 1.1M-share pledge by one family director). Near-zero pledging is a real positive and removes one of the classic Taiwanese family-group red flags.

The honest read: alignment exists, but it is alignment to the group, achieved through cross-holdings and a foundation, not through concentrated personal ownership that would make insiders feel a low share price the way a minority holder does.

This is where the governance profile earns its Yellow. Nan Ya is simultaneously controlled by its FPG siblings and the controller of its own listed affiliates, sitting in the middle of a circular ownership web. It also owns the two listed "golden geese" the market actually cares about:

  • Nanya Technology (2408.TW), DRAM: about 29.29% (FPG collectively over 50%, with Formosa Chemicals and Fibre and Formosa Plastics each about 10.81% alongside Nan Ya).
  • Nan Ya PCB (8046.TW), ABF/IC substrate: about 60.97% (a consolidated majority subsidiary).
        Chang Gung Medical Foundation
                   | ~11.1%
                   v
  Formosa Plastics (1301) --~9.9%--> NAN YA (1303) <--~5.2%-- Formosa Chemicals (1326)
       ^                                |   |                        ^
       |                                |   |                        |
       +----- Nan Ya reciprocal --------+   |   +-- Nan Ya reciprocal +
              cross-holdings                |   |       cross-holdings
                                            |   |
                          ~29.3% (assoc.)   |   | ~61.0% (subsidiary)
                                            v   v
                            Nanya Technology (2408)   Nan Ya PCB (8046)
                               DRAM engine              ABF substrate

Red-flag scan:

  • Circular cross-holding. The four FPG pillars own each other reciprocally. This inflates group book equity, dilutes accountability (no single external party can discipline the board), and means "book value" carries a layer of intra-group double-counting. It is the defining structural weakness for a minority holder and it is by design, not oversight.
  • Foundation as control vehicle. Chang Gung Medical Foundation, a non-profit, is the largest single holder of Nan Ya at about 11% and holds comparable blocs across the group. Using a tax-advantaged hospital foundation as a perpetual control anchor is legal and long-standing in Taiwan, but it entrenches insiders and is not a shareholder whose incentives align with market returns.
  • Related-party dependence is pervasive, not incidental. Formosa Petrochemical (6505.TW) supplies the group's feedstock; intra-group sales and purchases across the four pillars are enormous and recurring. Board seats are held by corporate representatives of Formosa Petrochemical, Formosa Chemicals and Fibre and Formosa Plastics, so the counterparties to Nan Ya's largest related-party flows sit on its board. Transfer pricing and capacity allocation across the group are therefore governed internally, and an outside holder must simply trust that Nan Ya is not the pillar being asked to absorb group-level costs.
  • Founder-inheritance overhang. Wang Yung-ching died in 2008 with no will, three wives and nine children, having moved large blocks of group shares into offshore trusts in Bermuda, the British Virgin Islands and the Cayman Islands. Inheritance litigation has run for years. Operating control has been insulated so far by the trust-and-foundation structure, but the ultimate beneficial ownership of a meaningful slice of the group sits in offshore vehicles subject to unresolved family dispute. That is a latent control-stability risk that has not, and may never, crystallize, but it is real and it is opaque.

4. Capital allocation

The tell of the last two years is that Nan Ya is funding itself by selling down its crown-jewel listed stakes rather than out of operating cash flow, while cutting the dividend hard.

On the dividend, the payout was slashed from NT$7.5 per share for 2022 to NT$0.7 for both 2024 and 2025, with NT$0.8 declared for 2026 (ex-date 30 June 2026, yield roughly 0.5% at the current price near NT$182). On trough earnings that NT$0.7 to NT$0.8 is above 100% of EPS, but in absolute terms the company is conserving cash, which is defensible in a petrochem trough.

The more revealing move is serial monetization of the affiliates:

Asset Action (2026) Proceeds Accounting
Nanya Technology (2408.TW) Board approved trimming about 29.3% toward about 28.3%; roughly 31M shares in the March tranche ~NT$7.9B tranche; ~NT$24.9B cumulative across three disposals Booked as disposal gain (associate under 50%); the March tranche gain alone about NT$6.4B
Nan Ya PCB (8046.TW) Selling about 19.4M shares May to Dec 2026; stake about 64% down to about 61% ~NT$17.4B Above 50% ownership, so NOT a P&L gain; credited to retained earnings and cash

Stated rationale is "to replenish operating capital and improve share liquidity." Read plainly: the core operating business is not self-funding through the trough. FY2025 operating cash flow collapsed to about NT$8.6B while capex held near NT$11B to NT$12B, so free cash flow turned negative (around negative NT$2.7B), net debt sits around NT$126.5B and is rising. The company is plugging that gap by harvesting the very listed assets (the DRAM proxy and the substrate maker) that give the equity its narrative premium, while keeping commodity-segment capex running. That is not reckless, and the affiliate stakes remain large, but it is the opposite of a management team compounding a scarce franchise. It is a conglomerate selling growth optionality to fund a cyclical trough.

5. Compensation

I could not retrieve Nan Ya's specific aggregate board and executive remuneration figures from public sources in this pass; the numbers are disclosed in the annual report (director and supervisor remuneration table, individual bands) filed to MOPS (mops.twse.com.tw) and should be pulled from there before relying on any figure. Flag this as a disclosure gap not closed, not as a clean bill. What can be said with confidence is structural: FPG's culture is famously frugal and reinvestment-oriented (the founder's ethos), executive pay has historically been modest relative to the group's scale, and there is no evidence of the outsized option-grant or self-dealing compensation pattern that would independently trip a red flag. The compensation risk here is not excess; it is that pay and incentives are set within a group culture optimized for cohesion and cost discipline rather than for per-share value creation at any one pillar.

6. Board and governance

The board is 12 members: eight non-independent (the chairman, several Wang family directors, and corporate representatives of Formosa Petrochemical, Formosa Chemicals and Fibre and Formosa Plastics) and four independent directors (Yi-Fu Lin, Yun-Peng Chu, Shuh Chen, Jonq-Min Liu). Independents are one-third of the board, which meets the letter of Taiwanese governance norms but leaves the group firmly in control of every vote that matters. Nan Ya runs the audit-committee model (no separate supervisors), with an Audit Committee composed of the independents, plus a Remuneration Committee and a Sustainable Development Committee. Gender diversity is low at 2 of 12 (16.7%).

There are no accounting scandals, no going-concern issues, and no pledging problem, which is why this is not a Red. But the committee scaffolding is best read as compliant rather than empowered: with three sibling companies holding board seats and the chairman also leading the group, the independent directors are structurally incapable of acting against group interests, and the entire premise of "永不分家" is that they will not be asked to.

7. Management DD verdict: YELLOW

Nan Ya is a competent, honest, deeply capable industrial operator with clean books, near-zero share pledging, real independent directors, functioning committees, and a genuine, if late, step toward professional (non-family) group leadership. On those axes it is better-governed than most Asian family conglomerates and clearly not a Red.

It is not a Green, and for this specific thesis the gap between the two is the whole point. Control rests on interlocking cross-holdings, a hospital foundation, and offshore family trusts rather than on aligned per-share ownership; related-party dependence on the FPG siblings is pervasive and self-governed; leadership is aged and concentrated; an unresolved founder-inheritance dispute sits offshore as a latent overhang; and capital allocation is currently harvesting the crown-jewel affiliate stakes to fund a commodity trough. Above all, the controlling doctrine is group perpetuation, not maximizing the value of the roughly 3%-of-revenue glass-cloth sliver that a bottleneck buyer is actually reaching for. There is no governance mechanism by which an outside holder can push this conglomerate to prioritize that sliver, and every incentive says it will not. This corroborates the sector reads (Nittobo (3110.T) management YELLOW; the cross-holding governance flag already noted on Fulltech (1815.TWO)): for the glass-cloth franchise you want direct, aligned ownership, which is precisely what Nan Ya's structure denies you. Buy 1303 only as a deliberate FPG-cycle and DRAM-proxy conglomerate exposure, never as a governed route to the AI glass-cloth bottleneck.

PART III. COMPETITIVE DYNAMICS

1. Where Nan Ya actually sits in the chain

The single most important fact about Nan Ya's competitive position is that it operates at two rungs of the glass-cloth value chain, and is a follower at both. It weaves fabric (the shallow, replicable step) and it laminates that fabric into CCL (a broad, competitive pool). It does not own the deep step, the glass melt and low-Dk / low-CTE yarn chemistry, where the bottleneck and the pricing power actually live. The full value-chain map, with competitor names per rung, is drawn in Part I section 5; this section reads it competitively rather than re-drawing it. Value and scarcity concentrate at the yarn rung and dissipate downward. Nan Ya's celebrated "AI glass" exposure is entirely at the weaving rung and, crucially, runs on yarn chemistry it licenses from the incumbent. Even the loudest second-source story in the market is a Nittobo-fed weaver, not an independent glass maker.

2. Competitive landscape

The table below is the honest map of who owns what rung. Valuations are yfinance live as of 2026-07-01 (Nan Ya reconciles to the NT$182.5 close; IBKR remains source of truth for any position sizing). Caps are converted at approximate spot FX for comparability only.

Company (ticker) Highest rung owned Position vs bottleneck Pricing power Cap (US$) Valuation Vault verdict
Nan Ya (1303.TW) Weaving plus CCL; no independent yarn Largest E-glass fabric maker by volume; junior contract weaver of Nittobo's high end Price-taker on E-glass (riding the shortage); price-follower to Nittobo on high end ~$45B ~79x TTM / ~19x fwd, 3.8x P/B, 5.5x P/S new page (this DD)
Nittobo (3110.T) Yarn: T-glass plus NER, sole M8 The monopoly; ~90% T-glass, 60-70% NER; ~92% of profit from EM at 38.3% OPM Price-maker but reluctant pricer ~$4.5B ~47-49x clean fwd (vault DD; yfinance ~13x / 0.8x P/B is mis-mapped, ignore) WATCH / half-weight ~¥4,580; DD YELLOW
Fulltech (1815.TWO) Weaving, NE/M7 plus T second-src Cleanest TW low-Dk pure-play; best TW gross margin (Q1'26 GM 35.97%) Price-taker at peak cycle ~$2.0B ~65x TTM, 4.7x P/B, 9.7x P/S NEUTRAL / Medium NT$101
Glotech (5475.TWO) Weaving, Q-glass / M9 Sole TW quartz-fiber maker at scale; Rubin optionality Price-maker on quartz, but Rubin-binary ~$0.9B ~31x P/B, ~29x P/S (priced as if won) Low-Medium NT$241.5
Taiwan Glass (1802.TW) Weaving, T-glass low-CTE 3rd worldwide to crack low-Dk cloth; certified 2025, volume ramping 2026 Price-taker, unproven at scale ~$6.5B ~21x fwd, 4.2x P/B, 5.0x P/S not in vault
Kingboard Laminates (1888.HK) Integrated: own yarn plus fabric plus CCL plus copper foil HK/China CCL leader; fully vertically integrated Price-maker (four 2026 hikes, cumulative over 40%) ~$40B ~127x TTM / ~37x fwd, 19x P/B, 15x P/S not in vault
China Jushi (600176.SS) Yarn: electronic-grade, low-Dk NVIDIA-certified World's largest electronic-cloth volume; low-Dk yarn passed NVIDIA Cost leader / volume price-setter ~$40B ~75x TTM, 8.8x P/B, 15x P/S not in vault
Mitsui Kinzoku (5706.T) Adjacent layer (HVLP copper foil) Different bottleneck; 90%-plus premium HVLP; pricing exercised Price-maker (12% hike stuck) ~$16B ~32x fwd above own SOP WATCH ~¥41,940

Two things jump off this table. First, every company that owns the yarn rung (Nittobo, Kingboard, China Jushi) is a price-maker; Nan Ya, which owns only weaving and CCL, is not. Second, Nan Ya at about 5.5x sales and 3.8x book is priced richer than pure-play weavers should be, because the multiple is carrying the DRAM stake and the petrochem recovery, not the glass sliver.

3. Moat analysis and the business-quality three-test

The moat, at the point the thesis is being priced, is weak. Nan Ya's genuine advantages are scale, a broad electronic-materials portfolio, and Formosa-group integration into epoxy, copper foil and CCL. None of these is the scarce capability. The scarce capability is a certified low-Dk / low-CTE yarn, and Nan Ya has none of its own. Its entire high-end entry is a reciprocal arrangement in which Nittobo sells it second-generation NER yarn and contracts it to weave roughly 20% of Nittobo's specialty fabric by 2027. Nan Ya is simultaneously Nittobo's customer and Nittobo's subcontractor. That is not a moat; it is a rented position at the incumbent's discretion.

Run the three tests a business-quality investor should apply, and Nan Ya fails the two that matter.

Test 1, pricing power: FAIL. Nan Ya has raised E-glass fabric prices multiple times through 2026, and the market reads this as strength. It is not. E-glass is a commodity; Nan Ya can push price only because the entire market is short, the same tide lifting Kingboard (four hikes, over 40% cumulative) and the RMB-per-metre spot (roughly 100% off the Q3-2025 low). That is a rising-tide windfall, not durable pricing power. Nan Ya cannot set E-glass prices in a normal market, and at the high end it is a price-follower to its own supplier. It holds pricing power at neither end of its range.

Test 2, proprietary durable advantage: FAIL. The deep, defensible step is glass chemistry and yarn certification. Nan Ya has no independent low-CTE / low-Dk certification in the searched record; it weaves on Nittobo's melt. Nittobo is deliberately outsourcing the shallow half of its moat (weaving) while keeping the deep half (yarn). Nan Ya's advantage is therefore, by construction, non-proprietary and revocable: the day the shortage eases, contracted weaving capacity is exactly the swing volume a principal throttles first.

Test 3, reinvestment economics: FAIL. A high-quality compounder turns retained capital into high incremental returns. Nan Ya generates roughly mid-single-digit ROE, ran negative free cash flow in FY2025 (about negative NT$2.7B), and carries a holding-company balance sheet where roughly half of book equity is the associate portfolio marked to the DRAM cycle. Capital here is deployed into commodity petrochem and cyclical equity stakes, not into a compounding franchise.

The honest verdict: real assets, real scale, wrong altitude. The moat exists one rung above where Nan Ya operates, and Nan Ya is renting access to it.

4. Industry structure and cycle position

Structurally this is a barbell. The deep rung (specialty yarn) is a near-monopoly (Nittobo) with two credible flankers building (AGY in the US, China Jushi in China). The shallow rung (weaving) is fragmenting fast as Nittobo licenses out capacity and Taiwan/China entrants certify in. Nan Ya lives on the fragmenting side.

On the cycle, the market is at or near peak tightness. Institutional estimates put the T-fabric supply-demand gap widening to roughly 18% in 2026, with total T-fabric demand more than doubling year on year, and layer counts on AI-server boards climbing from roughly 20 to 28 (2024 to 2025) toward 24 to 40 (2026 to 2027). That is what is driving the price hikes Nan Ya is capturing today.

The turn is dated, and it is 2027. Nittobo announced an August 2025 plan to triple T-glass capacity and to invest more than 50B yen across Japan and Taiwan through 2026 to 2027, but new lines need roughly six months to reach stable yield, so meaningful relief does not arrive before mid-2027. Alongside it: AGY targets H2-2027 fabric volume; China Jushi has already lit its 100,000-tonne Huai'an electronic-fibre plant (390 million metres of cloth); Kingboard's own yarn-and-fabric lines at Shixing come on stream around mid-2026. This is a visible, scheduled 2027 capacity wave.

For Nan Ya specifically, that wave is a headwind aimed squarely at its only real AI beta. Nan Ya's incremental profit from the shortage sits in E-glass fabric pricing (the roughly 3% of revenue that is glass cloth), which is the most commoditised, most easily added, and therefore first-to-compress line in the whole chain. When the wave lands, the high-end yarn scarcity may persist a while longer, but Nan Ya does not own that; it owns the E-glass spread that the wave is built to close.

5. Emerging threats

The 2027 capacity wave compresses Nan Ya's actual exposure first. Nan Ya's glass-cloth windfall is E-glass pricing, and E-glass is precisely what China Jushi, Kingboard, and Nittobo's Taiwan expansion are adding most of. The scarce low-Dk / low-CTE rung Nan Ya does not own will stay tighter longer than the E-glass rung it does.

Chinese vertical integrators are a structural threat at both of Nan Ya's rungs. Kingboard makes its own yarn, fabric, copper foil and epoxy, capturing the full margin stack and pushing prices as a maker rather than a taker. China Jushi has the world's largest electronic-cloth volume and a low-Dk yarn that has passed NVIDIA certification, i.e. it is building exactly the yarn-rung capability Nan Ya lacks. Both compete with Nan Ya on the E-glass and CCL rungs today and threaten to leapfrog it on the yarn rung tomorrow.

Second-sourced by design means throttle-first. Nan Ya's contracted 20% is Nittobo's deliberate outsourcing of replaceable weaving. In a de-tightening market, principals cut contracted swing capacity before their own; Nan Ya's contracted volume is a shock absorber for Nittobo, not a franchise for Nan Ya.

The frontier is moving away from what Nan Ya weaves. Nan Ya weaves second-generation NER (M8-class). The roadmap is already past it: Nittobo's third-generation NEZ glass is slated for 2027, and quartz / Q-glass (M9) is in early adoption for the highest-speed boards. Nan Ya is structurally one rung behind and depends on Nittobo advancing it there. It captures none of the NEZ-versus-Q-glass optionality that Glotech (quartz) or Nittobo (NEZ) hold.

The thesis is diluted at the corporate level, not just the competitive one. This is the threat that overrides the others. Even a flawless glass-cloth outcome barely moves Nan Ya, because glass cloth is roughly 3% of revenue and low-single-digit percent of profit. The share price is a levered proxy for the DRAM cycle (via the roughly 29% Nanya Technology stake and equity-method income) wrapped in a commodity petrochem base. The swing from a negative NT$4.1B net quarter (Q2 2025) to a positive NT$14.25B net quarter (Q1 2026) came almost entirely from equity income, not operations. So the bottleneck exposure a buyer is paying up for is diluted twice over: competitively (rented, replaceable, on the wrong rung) and structurally (a roughly 3% line inside a DRAM-and-plastics conglomerate). Owning 1303 for the glass-cloth bottleneck is owning the dislocation at two removes, at a full price.

PART IV. THE NUMBERS

1. Valuation snapshot (2026-07-01)

Metric Value Note
Price NT$182.5 At the 52-week high (range NT$26.25 to NT$183, roughly a 7x move off the June 2025 low)
Shares out 7.931B
Market cap NT$1,447B (~US$45 to 47B) The dossier's NT$1,201.5B was the ~NT$151.5 print; up about 20% since
Net debt (FY25) NT$126.5B Total debt 173.2B minus cash and ST investments 64.1B
Enterprise value ~NT$1,574B
Trailing P/E (stale FY25 EPS 0.57) ~320x What yfinance and the dossier show; misleading, pre-DRAM-recovery
Trailing P/E (TTM through Q1'26, EPS 2.31) ~79x The honest trailing number (statementdog 近四季 EPS 2.31)
Forward P/E ~17 to 19x yfinance fwd EPS 9.61 gives about 19x; one TW source cites 16.5x [basis unverified, assumes a large FY26 DRAM up-cycle]
P/B 3.77x Book NT$48.40; rich for roughly 5% ROE
P/S ~5.5x Extreme for a commodity plastics maker (norm below 1x); pure narrative premium
EV/EBITDA (FY25) ~47x Trough-distorted; forward far lower
Div yield ~0.4 to 0.5% Payout ratio above 100% on trough EPS; not an income name here

2. Three-year P&L and latest quarters (NT$B)

FY Revenue Gross % Op inc Op % Net inc Net % EPS OCF Capex FCF
2022 (peak) 355.2 15.2 32.0 9.0 32.1 9.0 4.05 75.3 -31.2 44.1
2023 259.8 7.4 2.00 0.8 6.31 2.4 0.80 35.7 -24.0 11.7
2024 259.6 7.2 0.35 0.1 3.34 1.3 0.42 18.8 -12.6 6.1
2025 259.9 8.3 3.68 1.4 4.52 1.7 0.57 8.6 -11.3 -2.7
Quarter Revenue Gross % Op inc Net inc EPS Note
Q1'25 65.6 7.0 0.04 0.45 0.06
Q2'25 65.7 8.6 1.30 -4.12 -0.52 Operating profit positive but net loss; DRAM / equity drag
Q3'25 64.2 8.3 1.04 3.26 0.41 Equity income turns
Q4'25 64.5 9.2 1.30 4.93 0.62
Q1'26 68.6 n/a 3.74 14.25 1.80 Pre-tax NT$16.08B; 15-quarter high, greater than the prior three FY combined

The story is in the two boxed rows. Between Q2 2025 (negative NT$4.1B net) and Q1 2026 (positive NT$14.25B net), operating income barely moved (about plus NT$2 to 3B). The roughly NT$18B swing is almost entirely equity-method income from Nanya Technology's DRAM turn plus FPCC. That is the whole engine.

3. Profit attribution (confirmed against primary reporting)

Q1 2026 pre-tax income was NT$16.08B against operating income of NT$3.74B, so about 77% of pre-tax profit is equity-method and non-operating income, overwhelmingly the Nanya Technology (2408.TW) DRAM stake plus FPCC, Nan Ya PCB (8046.TW) and others, net of roughly NT$1B per quarter of interest. This confirms the earlier working estimate (75 to 80%). Within operating profit, the four commodity segments carry the load, with electronics doing nearly all of it while plastics, polyester and chemicals run near breakeven in the current commodity trough. Glass cloth is a high-margin but tiny-volume line, so it is plausibly only about 1 to 3% of total earnings. By revenue about 3%, by profit about 1 to 3%: you are buying roughly 95%-plus commodity petrochem plus a DRAM proxy to get about 3% glass-cloth beta.

4. Balance sheet and cash

Net debt is NT$126.5B and rising; total debt NT$173B against NT$64B cash. FY2025 free cash flow was negative (about negative NT$2.7B) as operating cash flow collapsed to NT$8.6B (from NT$35.7B in 2023) while capex held near NT$11B. The NT$184B of long-term equity investments on the balance sheet, roughly 30% of total assets and about 52% of book equity, is the associate portfolio (Nanya Tech and others). This is a holding-company balance sheet as much as an operating one; its mark-to-market rides the DRAM cycle.

5. Skeptical pass on the key figures

The headline P/E is a trap. yfinance shows about 320x and the dossier said about 261x, both computed on stale FY2025 EPS of NT$0.57 that predates the DRAM recovery. The honest trailing figure, using statementdog's TTM EPS of NT$2.31 through Q1 2026 (where the Q1 2026 EPS of 1.80 alone exceeded the prior three years combined), is about 79x. Neither trailing number is meaningful at a trough-to-recovery inflection; both should be discarded.

The only multiple that "works," the roughly 17 to 19x forward, is built on yfinance forward EPS of 9.61, which implies FY2026 net income near NT$76B, above the 2022 peak of NT$32.1B. That embeds a full DRAM super-cycle flowing through the Nanya Tech stake. Treat it as analyst optimism, not fact; it is the single most load-bearing and least-verified number in the whole case.

The least-gameable lenses, P/S about 5.5x and P/B 3.77x, both say the market is paying a full AI / DRAM-recovery price. This business historically traded near 0.7x sales and about 1x book. On mid-single-digit ROE and book of NT$48.40, nothing organic supports NT$182.5; the entire premium is narrative plus cycle.

"World's largest glass-fabric maker by volume" is true but narrowly so. The volume is overwhelmingly standard E-glass; Nittobo owns the high-end AI-grade rungs. The superlative must not be read as leadership in the bottleneck. (Note: sources confirm total-volume leadership skewed to E-glass and "one of the world's largest capacities," but an independent ranking naming Nan Ya number one outright was not located, so it is stated as largest E-glass fabric maker by volume.)

The earnings inflection is real but non-operating, and now confirmed: Q1 2026 operating profit NT$3.74B versus pre-tax NT$16.08B, about 77% non-operating. Q2 2025 booked an operating profit and a net loss in the same quarter, which is the tell that the engine is the equity stake, not operations.

The sell-side does not validate the price. Consensus is Hold, the freshest mean target is about NT$106.5 (roughly 42% below spot), and even the single most bullish target (NT$200) is only about plus 10%. The Street prefers to own the DRAM engine directly (UBS Buy on Nanya Tech 2408 at NT$190) rather than through the wrapper.

6. Flags and unverified

  • The forward EPS of 9.61 and forward P/E near 19x are yfinance-derived and embed an aggressive DRAM up-cycle; the implied FY2026 net income (about NT$76B) would exceed the 2022 peak, a high bar.
  • The "no independent low-CTE certification" claim and the exact Nittobo contract terms are per Taiwan and Japan press and sibling coverage, not a primary contract or spec sheet.
  • The Nanya Tech stake of about 29.3% is confirmed (ifa.ai / annual report; largest holder, with Formosa Plastics and Formosa Chemicals about 10.8% each); Nan Ya announced a 2026 disposal plan trimming toward about 28.3%.
  • Segment mix is from Taiwan sources (MoneyDJ, CMoney); yfinance's US-GAAP-mapped segment fields differ slightly from the audited annual-report definitions.
  • Board and executive remuneration figures were not retrieved this pass; pull from the annual report on MOPS before citing any number.
  • Live price NT$182.5 verified via yfinance; reconcile against IBKR as book of record before any sizing.

PART V. THE DECISION

1. Growth drivers and catalysts

Nan Ya (1303.TW) has three separate engines, and it is essential to keep them apart because the market is pricing all three at once while only one is proprietary and none of the three is the glass-cloth bottleneck itself.

The DRAM equity stake is the real engine. The dominant driver is Nan Ya's roughly 29.3% look-through to Nanya Technology (2408.TW). Nanya Tech posted Q1 2026 net income of NT$26.06B on a 53.1% net margin with revenue up 63.1% quarter on quarter, and management and the sell-side now frame DRAM supply as tight through 2028 as HBM crowds out commodity DDR. UBS upgraded 2408 to Buy at an NT$190 target; consensus targets on 2408 run wide (NT$150 to NT$660). At 29.3%, Nanya Tech alone contributes on the order of NT$7 to 8B of equity income per quarter, roughly half of Nan Ya's entire Q1 2026 net income of NT$14.25B. This is the catalyst that matters, and it is not Nan Ya's to control.

The E-glass shortage is a genuine but shallow tailwind. Nan Ya is the world's largest glass-fabric maker by volume, overwhelmingly standard E-glass, and it is raising electronic-materials prices monthly since March 2026 at mid-to-high single-digit increments. Industry demand for high-end electronic fabric is set to grow above 40% in 2026 and much new capacity is already pre-sold under long-term agreements. Nan Ya captures real volume and price on the low end of the ladder. The catalyst here is broad reflation of a commodity input, not scarcity rent on a proprietary grade.

The move up the grade ladder is optionality, unproven. Two catalysts sit further out. First, the reciprocal Nittobo (3110.T) deal: Nittobo supplies Nan Ya its second-generation low-Dk (NER) yarn as feedstock and Nan Ya contract-weaves about 20% of Nittobo's global specialty fabric by 2027. Second, management guided M8 and M9 high-end CCL to mass production in H2 2026. Both would lift mix, but both are downstream of someone else's chemistry: Nan Ya weaves on Nittobo's yarn and buys into the high end as a junior partner, not a certified independent low-CTE source.

A fourth, softer, driver is petrochemical normalization. Chemicals, polyester, and plastics (about 45% of revenue) are running near breakeven at the bottom of the commodity cycle; any spread recovery is upside to operating income, but it is a cyclical bet, not a secular one.

The profit engine, made explicit (the diagram traces the flow of profit contribution):

1303 Q1'26 net income  NT$14.25B  (EPS 1.80, a 15-quarter high)
   |
   +--> Operating income  ~NT$3.74B  (~23% of pre-tax)
   |        |
   |        +--> Electronics 52.8% of revenue  (PCB, CCL, epoxy, copper foil;
   |        |         glass fabric only ~3% of total company revenue)
   |        +--> Chemicals / Polyester / Plastics  ~45% of revenue, near breakeven
   |
   +--> Equity-method + non-operating  ~NT$12.3B  (~77% of pre-tax)
            |
            +--> Nanya Technology (2408) 29.3% DRAM stake  --> largest single driver
            +--> FPCC (6505), Nan Ya PCB (8046), other Formosa affiliates

The read is unambiguous: about three-quarters of the profit that inflected comes from an equity stake in a DRAM maker Nan Ya does not control, and the glass-cloth line the narrative rests on is roughly 3% of revenue and low-single-digit percent of profit.

2. Risks

"Can it close" asks whether the risk is transient (a path exists to resolve or de-risk it) or structural (it cannot be engineered away and would require a change in the business itself).

Risk Likelihood Mitigant Can it close?
DRAM cycle rolls over and equity income from Nanya Tech (2408) collapses, taking about 75% of profit with it Medium (cycle is early but cyclical; shortage narrative runs to 2028) 2026 partial disposal of the 2408 stake (toward about 28.3%) monetizes some gain; petrochem recovery could partly offset No. Structural. Nan Ya does not control DRAM pricing or capacity; passive, mark-to-cycle beta
Glass-cloth exposure stays a contracted sliver, never becomes proprietary share High (already the base case) Nittobo NER-yarn feedstock deal and H2 2026 M8/M9 CCL ramp lift mix at the margin No. Structural. Weaving on Nittobo's yarn is the shallow half of the moat; independent low-CTE / low-Dk certification is not in hand
Valuation de-rates from AI / DRAM-recovery multiples toward through-cycle norms (P/B about 1x, P/S below 1x) Medium-High (P/S 5.5x and P/B 3.8x are historically extreme for a commodity conglomerate) Sustained DRAM super-cycle plus petrochem recovery could hold the forward multiple for several quarters Only upward. The de-rate risk itself is structural to a commodity name at a narrative premium
Petrochem trough persists; commodity spreads (EG, phenol, PVC, PET) stay compressed Medium (macro / oil and China oversupply dependent) New electronics capacity and glass-fabric price hikes carry operating income while commodities are weak Partly. Cyclical, so it can close on a spread recovery, but timing is exogenous
Negative free cash flow persists (FY25 FCF was negative NT$2.7B; net debt NT$126.5B and rising) Medium Recovering OCF as earnings inflect; asset-heavy but not distressed; access to Formosa-group balance sheet Yes. Cyclical; OCF should recover with earnings if the up-cycle is real
Second-sourcing / new capacity floods E-glass as everyone expands into the shortage Medium-High (much 2026 capacity is pre-sold, but 2027-plus additions are large) Long-term pre-sold agreements and the Nittobo tie-up lock some volume No. Structural to a commodity where Nan Ya is a price-taker at scale, not a spec setter
Liquidity / momentum reversal: stock is up roughly 7x off its NT$26 low and sits at the 52-week high Medium Large-cap, deep TWSE liquidity (unlike the TPEx names in the peer set) This is a price risk, not a business risk; a reversal closes only by falling

3. Ownership and analyst sentiment

Ownership is dominated by the Formosa web. Insiders and affiliated group entities hold about 38.1% (the FPG, FPCC, FCFC cross-holdings that also define the associate portfolio on the other side of the balance sheet), while foreign and domestic institutions hold only about 14.2%. For a company capitalized near NT$1.45T (about US$45 to 47B), that institutional float is modest, which is consistent with a re-rating driven more by domestic momentum and the DRAM / AI narrative than by deep institutional conviction. The low institutional ownership is a flag, not a comfort.

The sell-side is openly skeptical of the price. On the freshest read (yfinance consensus, six analysts, recommendation Hold), the mean 12-month target is NT$106.5 with a high of NT$200 and a low of NT$26. That mean sits about 42% below the NT$182.5 close, and even the single most bullish target implies only about plus 10% upside. Other aggregators show stale but directionally identical readings (TradingView average NT$79.8, high NT$125; MarketScreener NT$81.67, both quoted against an earlier print near NT$89.6). The message across every source is the same: the stock has run well past where analysts place fair value, and the rating is Hold, not Buy.

Sentiment input Reading Implication
Consensus rating Hold (6 analysts) No conviction Buy despite the earnings inflection
Mean target (freshest) NT$106.5 About 42% below spot
Target range NT$26 to NT$200 Even the high is only about plus 10%; skew is to the downside
Institutional ownership ~14.2% Thin for a US$45B-plus cap; run is momentum-led
Insider / group holding ~38.1% Formosa cross-holdings; low true float

Contrast this with the underlying: the sell-side is bullish on Nanya Technology (2408) directly (UBS Buy, NT$190) while it is neutral-to-bearish on the wrapper (1303). That divergence is itself the thesis. The market prefers to own the DRAM engine directly rather than through a commodity-plastics holding company.

4. Position sizing

Verdict: PASS as a glass-cloth bottleneck play. At best a leveraged DRAM-cycle beta, and there are cheaper, cleaner ways to own both underlyings.

If a bottleneck sleeve is what Pink wants, 1303 does not belong in it. The clean expressions already sit in the vault: Nittobo (3110.T) for the T-glass and NER monopoly, Fulltech (1815.TWO) for the NE/M7 pure-play, Glotech (5475.TWO) for quartz/M9, and Mitsui Kinzoku (5706.T) one layer over in copper foil. Each owns the scarce rung directly. Nan Ya owns roughly 3% of revenue in mostly standard E-glass and weaves on Nittobo's yarn.

If the actual thesis is DRAM-cycle beta, own Nanya Technology (2408) directly. Buying 1303 to get a 29.3% look-through means paying a full commodity-petrochem franchise (P/S 5.5x, P/B 3.8x) as an entry fee, diluting the DRAM exposure with about 45% of revenue in trough petrochemicals and about 3% in glass cloth. That is a worse risk-adjusted way to express the same view.

Practical sizing, if held at all: no more than a starter (quarter-weight), and not at NT$182.5. The stock is roughly 72% above the freshest analyst mean (NT$106.5) and sitting at its 52-week high after a roughly 7x move. Any position should be an explicit, sized DRAM-cycle beta trade with a defined exit, entered nearer analyst fair value or toward about 2x book, never a "own the bottleneck" allocation. Reconcile the live quote against IBKR before sizing; the NT$182.5 figure here is the TWSE close.

5. Bear case and downside

The bear case is not that the earnings inflection is fake. It is real. The bear case is that the market is paying an AI-bottleneck-plus-DRAM-supercycle price for a company that is 95%-plus commodity petrochemicals and a passive DRAM stake, and that the two pillars holding the valuation are both cyclical and neither is controlled by Nan Ya.

The specific failure path: DRAM peaks or merely decelerates, equity income from Nanya Tech (2408) fades, and roughly three-quarters of the profit that inflected reverses. Operating income (1 to 1.4% margins in commodity petrochem) cannot backfill the gap, exactly as Q2 2025 showed when the company booked an operating profit and a net loss in the same quarter. Simultaneously, 2027-plus E-glass capacity additions relieve the shortage, capping the one organic tailwind, and the glass-cloth line stays a contracted sliver on Nittobo's yarn rather than becoming proprietary share. The forward P/E of about 19x, which is the only multiple that looks reasonable, is revealed to have embedded a peak-cycle earnings number; on normalization the stock reverts toward book value and mid-single-digit ROE.

Downside scenarios from NT$182.5:

  • Base bear (de-rate to analyst fair value): NT$100 to 110, about negative 40 to 45%. This is simply the freshest consensus mean (NT$106.5) reasserting itself, and it also lines up with roughly 2.2x book (book NT$48.40). It requires no catastrophe, only the AI / DRAM premium fading toward what the sell-side already models.
  • Deep bear (DRAM and petrochem both roll): NT$60 to 75, about negative 60 to 67%. Through-cycle 1.3 to 1.5x book with equity income normalized and commodity spreads still soft. This is the give-back of most of the 2025 to 2026 re-rate, consistent with the stock having risen 7x off NT$26 in twelve months.

Upside is capped and asymmetric against these. The single most bullish analyst target is NT$200 (about plus 10%), and to justify materially more the market must underwrite a DRAM super-cycle flowing through the stake, a durable petrochem recovery, and a persistent glass premium all at once, three stacked cyclical bets. The skew from NT$182.5 is unfavorable: roughly plus 10% to the loudest bull, negative 40% to consensus fair value, negative 60%-plus if the cycle turns. That asymmetry, not any doubt about the current quarter, is why this is a pass.

Sources

Business, sector and value chain:

Financials, results and sentiment:

Governance and ownership:

Decision log

2026-07-01: initial canonical /deep-dive

Created the canonical 1303 page by consolidating the /deep-dive workstreams (business, management, competitive dynamics, numbers, decision) against the sector siblings 3110, 1815, 5475 and 5706 under advanced-packaging. Live price NT$182.5 (TWSE close, yfinance-verified; IBKR reconciliation pending). Directional verdict: PASS as a glass-cloth bottleneck play, at best a leveraged DRAM-cycle beta with cleaner ways to own either underlying. Management DD: YELLOW (clean operator, structurally misaligned; control via FPG cross-holdings plus Chang Gung Foundation, not per-share family economics). Key skeptical corrections applied: discarded the stale roughly 320x / 261x trailing P/E built on FY2025 EPS of NT$0.57 in favor of about 79x on TTM EPS of NT$2.31; confirmed roughly 77% of Q1 2026 pre-tax profit is non-operating equity income (Nanya Tech DRAM stake); flagged the forward EPS of 9.61 (implied FY2026 net income above the 2022 peak) as unverified analyst optimism; kept "largest glass-fabric maker by volume" qualified as standard E-glass, not bottleneck leadership. Open items: board remuneration (MOPS), identity of the roughly 2.3% affiliate holder, primary confirmation of the "no independent low-CTE certification" claim, and live IBKR price / cost-basis reconciliation before any sizing.

Source updates (auto-maintained)

Drop/z Misc (Dec 9, 24) - JPM_ASEAN 2025 Outlook_ Year of the Snake might rattle nerve...

The "1303" match in this article is analyst Ranjan Sharma's phone number ((65) 6882-1303), not the ticker; Nan Ya Plastics (1303.TW) is not mentioned anywhere in this JPMorgan ASEAN 2025 Outlook, which covers only ASEAN six markets and excludes Taiwan entirely.

Relevant to your thesis: Tangential — flagged for review.

Source: dropfile://z Misc/JPM_ASEAN 2025 Outlook_ Year of the Snake might rattle nerves_28_Nov_2024.pdf

from the vault · open in obsidian ↗