# Geek Trend Research Audit 09

Research date: 2026-07-11 (Asia/Shanghai)

## Decision

The investable trend is not the temporary 40% PCB price spike by itself. The durable change is that AI servers and high-speed switches are increasing board layers and forcing an upgrade from ordinary E-glass to Low-Dk, Low-CTE and eventually Q-glass. Limited glass-yarn, weaving, treatment and customer-qualification capacity is being redirected toward higher-grade products.

Primary research ticket: **Fulltech Fiber Glass (1815.TWO)**

Quality benchmark / bottleneck owner: **Nitto Boseki (3110.T)**

The Fulltech decision is **starter-position research only**, not a full-position conclusion. At the latest verifiable close of NT$95.0, the market already implies roughly NT$2.22bn of annual earnings at a 25x multiple, versus NT$1.83bn from annualizing 2026 Q1. Remaining upside therefore requires Low-Dk2 and Low-CTE to scale; ordinary industry growth is not enough.

## Signal Hygiene

| Signal | Use | Decision |
| --- | --- | --- |
| Reddit / PCB buyers reporting 20-50% quote increases | Early pain sensor | Keep as discovery signal |
| HN motherboard and workstation-cost complaints | Cross-community confirmation | Keep as breadth evidence |
| April high-purity PPE outage and war-related 40% PCB shock | Acute event | Strip out of structural thesis |
| Panasonic CCL and prepreg increases of 15-30% | Industrial price confirmation | Keep, but does not identify the profit winner |
| AI server move to M8/M9, Low-Dk2, Q-glass and HVLP4 | System design change | Core structural evidence |
| Glass-cloth supplier mix shifts and long qualification | Same-capacity mechanism | Core structural evidence |

## Scarce-Layer Ranking

| Rank | Layer | Scarcity | Profit capture | Judgment |
| --- | --- | --- | --- | --- |
| 1 | Low-Dk / Low-CTE glass yarn and cloth | Very high | High; fixed-cost lines plus product mix | Primary layer |
| 2 | HVLP4/5 copper foil | High | High, but larger-company dilution | Secondary layer |
| 3 | High-purity PPE resin | Acute concentration | Event-driven and normalization-prone | Reject as structural main thesis |
| 4 | High-end CCL | Tight | Input inflation can offset price hikes | Watch, not main layer |
| 5 | PCB / substrate fabrication | Strong demand | Multiple bottlenecks and capex intensity | Downstream beneficiary |
| 6 | ABF film | Strong moat | Low exposure inside Ajinomoto | Reject on stock purity |

## Company Universe

| Company | Ticker | Layer | Exposure / control | Audit decision |
| --- | --- | --- | --- | --- |
| Fulltech Fiber Glass | 1815.TWO | Integrated yarn + glass cloth | 69% electronic applications in 2025 Q1-Q3; high-end mix target 60% in 2026 | **Primary ticket**: earnings inflection plus Low-CTE option, but qualification risk |
| Nitto Boseki | 3110.T | T-glass / NE/NER glass | Electronic materials produced 84% of reportable segment profit in FY2026 | **Quality benchmark**: strongest moat; current value already prices much of FY2027 plan |
| Mitsui Kinzoku | 5706.T | VSP HVLP / MicroThin copper foil | Company estimates 60% share in HVLP2+; actively reducing standard foil mix | Passes scarcity; rejected as main ticket due conglomerate dilution and richer market cap |
| Taiwan Glass | 1802.TW | Low-CTE second source | Capacity expansion; broader flat-glass exposure | Optionality only; qualification and purity below Fulltech |
| Grace Fabric Technology | Private / China | Low-CTE cloth | Expanding second-source capacity | Important competitor, not investable here |
| Nan Ya Plastics | 1303.TW | Glass cloth / resin / CCL | Vertical integration, very diversified | Reject on purity |
| Taishan Fiberglass | Private / China | Low-CTE glass | Emerging alternative | Qualification/geopolitical risk |
| Elite Material | 2383.TW | High-end CCL | Market leader in high-speed CCL | Strong beneficiary, but upstream input pass-through risk |
| Taiwan Union Technology | 6274.TW | High-end CCL | AI server material exposure | Secondary beneficiary; less direct supply control |
| ITEQ | 6213.TW | CCL | High-speed material mix | Secondary beneficiary; less scarce than glass cloth |
| Shengyi Technology | 600183.SS | CCL | Large-scale M7/M8 supplier | Strong growth, but already widely recognized and less pure bottleneck control |
| Kingboard Laminates | 1888.HK | CCL | Large integrated laminate supplier | High cycle exposure; upstream cost pass-through matters |
| Panasonic Holdings | 6752.T | MEGTRON CCL | Official 15-30% price increases; doubling capacity | Product is relevant, stock exposure too diluted |
| Mitsubishi Gas Chemical | 4182.T | BT / low-CTE laminate | Price increases and IC-substrate materials | Relevant but diversified and downstream of glass cloth |
| Resonac Holdings | 4004.T | Package substrate materials | Low-CTE laminate exposure | Relevant but diversified |
| Ajinomoto | 2802.T | ABF film | High supplier concentration and long qualification | Excellent material moat, weak stock purity |
| JX Advanced Metals | 5016.T | Copper foil | High-end foil exposure | Secondary candidate; less direct evidence than Mitsui |
| Chang Chun Group | 1765.TW / private units | Copper foil / resin | Integrated materials | Exposure split across entities and products |
| Ibiden | 4062.T | IC package substrates | Large AI package substrate beneficiary | Demand beneficiary, not upstream material controller |
| Shinko Electric | 6967.T | IC package substrates | Advanced package exposure | Lower scarcity ownership |
| Unimicron | 3037.TW | ABF substrates / PCB | AI substrate demand | Downstream, capex and yield risk |
| Kinsus | 3189.TW | IC substrates | AI package demand | Downstream beneficiary |
| Nan Ya PCB | 8046.TW | IC substrates / PCB | AI and network boards | Downstream beneficiary |
| Gold Circuit Electronics | 2368.TW | AI server PCB | Strong AI server content | Board maker; input and customer concentration risk |
| Zhen Ding | 4958.TW | PCB | Broad product and customer base | Too diversified for main thesis |
| Victory Giant | 300476.SZ | High-layer PCB | AI server PCB growth | Downstream and already consensus |
| Meiko Electronics | 6787.T | PCB | High-layer board demand | Downstream and lower material control |
| SABIC | 2010.SR | High-purity PPE resin | Reported concentration around 70% | Acute event winner/loser; reject because thesis depends on plant outage and geopolitics |

## Fulltech Fundamental Check

- 2025 revenue: NT$5.975bn, +40.7% YoY.
- 2025 gross margin: 28%, up about 14 percentage points.
- 2025 net income: NT$856.5m; EPS NT$1.64.
- 2026 Q1 revenue: NT$1.88bn, +36% YoY.
- 2026 Q1 gross margin: 36%; net income NT$458m; EPS NT$0.79.
- 2025 Q1-Q3 electronic application share: 69% versus 53% a year earlier.
- Low-Dk plus Low-CTE share: 45% at the end of 2025; company target 60% in 2026.
- Low-CTE status: small shipments began in 2025 Q4; qualification expansion and learning-curve work remain in progress.
- Latest verified close: NT$95.0 on 2026-07-09.
- Approximate market cap: NT$55.5bn; TTM P/E around 45x; Q1 annualized P/E around 30x.

## Valuation Model

Share count used: approximately 585m.

| Case | Revenue | Net margin | Net income | EPS | Multiple | Implied value | Meaning |
| --- | ---: | ---: | ---: | ---: | ---: | ---: | --- |
| Failure | NT$7.0bn | 15% | NT$1.05bn | NT$1.80 | 20x | NT$36 | Low-CTE fails and mix/margins reverse |
| Current run-rate | NT$7.52bn | 24.4% | NT$1.83bn | NT$3.16 | 25x | NT$79 | Annualized Q1, before assuming further mix gains |
| Trend | NT$11.0bn | 25% | NT$2.75bn | NT$4.70 | 25x | NT$118 | Low-Dk2 scales and Low-CTE becomes meaningful |
| Bull | NT$14.0bn | 28% | NT$3.92bn | NT$6.70 | 25x | NT$168 | Low-CTE and Q-glass qualification both succeed |

At NT$95, the market is not pricing only the current run-rate; it already assumes more earnings. The ticket works only if the Trend or Bull case becomes visible.

## Kill Conditions

1. Low-CTE remains immaterial for two more reporting periods or customer certification is delayed.
2. Gross margin falls below 30% while revenue continues to grow, suggesting commodity volume rather than high-end mix.
3. Electronic/high-end mix stops rising despite industry shortages.
4. New Taiwan/China capacity clears qualification faster than expected and removes pricing power.
5. Capex, working capital or equity issuance grows faster than operating cash generation.

## Primary Evidence Set

1. Fulltech 2025 investor presentation: https://www.ffg.com.tw/q/paper/114Q3.pdf
2. Fulltech official IR: https://www.ffg.com.tw/en/finance-1.asp
3. Nittobo FY2026 results: https://www2.jpx.co.jp/disc/31100/140120260512525575.pdf
4. Nittobo integrated report 2025: https://www.nittobo.co.jp/eng/ir/pdf/Nittobo_Integrated-Report_2025.pdf
5. Nittobo medium-term plan: https://www.nittobo.co.jp/eng/ir/keiei/chukei.htm
6. Mitsui Kinzoku engineered materials briefing: https://www.mitsui-kinzoku.com/LinkClick.aspx?fileticket=FBZhYz20XWE%3D&mid=1027&tabid=204
7. Mitsui Kinzoku FY2025 Q&A: https://www.mitsui-kinzoku.com/LinkClick.aspx?fileticket=p3auOv8dttw%3D&mid=1027&tabid=249
8. Panasonic price revision: https://industrial.panasonic.com/sa/electronic-materials/news/2026-04-09-em-ccl
9. Panasonic MEGTRON investment: https://news.panasonic.com/global/press/en260304-5
10. TrendForce Rubin PCB architecture: https://www.trendforce.com/presscenter/news/20251120-12790.html
11. Taiwan PCB Association material update: https://www.tpca.org.tw/Message/MessageView?id=28359&mid=112
12. Hacker News hardware pain thread: https://news.ycombinator.com/item?id=48050540
13. Reddit PCB buyer thread: https://www.reddit.com/r/PCB/comments/1usamxh/pcb_prices_really_went_up_so_annoying/
14. Current Fulltech quote cross-check: https://ww2.money-link.com.tw/TWStock/StockTick.aspx?SymId=1815
15. Fulltech market cap cross-check: https://stockanalysis.com/quote/tpex/1815/market-cap/