# 3081.TW - LandMark Optoelectronics
## Date: 2026-04-12
## Score: 64/100

### Thesis (2-3 sentences)
LandMark is a Taiwan-listed upstream III-V epi-wafer supplier selling InP and GaAs laser / photodetector epi-wafers into optical-transceiver customers, which makes it directly relevant to the AI photonics stack. The real attraction is not scale but positioning: LandMark sits at the materials layer for 400G/800G/1.6T optical modules and has shown a sharp growth inflection as Taiwan's silicon-photonics ecosystem accelerated. The limitation is that this is still a small-cap supplier in a competitive niche rather than a true monopoly chokepoint.

### Source Trail
| Topic | Source | Date / Period |
|-------|--------|---------------|
| Business model / products | LandMark website + 2024 Annual Report | 2024 / accessed 2026-04-12 |
| Monthly revenue | LandMark investor financial information page | 2025-2026 / accessed 2026-04-12 |
| Q1 2025 performance | LandMark Q1 2025 investor conference presentation | 2025-04-30 |
| 9M 2025 financials | LandMark 9M 2025 financial statements | 2025-09-30 |
| Taiwan growth ranking | DIGITIMES article | 2025-12-12 |

### Key Metrics
| Metric | Value | Notes |
|--------|-------|-------|
| Exchange | 3081.TW | Reported by company |
| 2024 revenue | NT$1.21B | Reported in 2024 annual report |
| 2025 revenue | NT$2.20B | Reported monthly revenue total |
| Jan-Mar 2026 revenue | NT$904.4M | Reported monthly revenue |
| 9M 2025 revenue | NT$1.56B | Reported financial statements |
| 9M 2025 gross margin | 40% | Reported financial statements |
| Core products | InP / GaAs laser and PD epi-wafers | Reported by company |

### Value Chain Position
LandMark is an upstream epitaxy supplier. It manufactures the laser diode epi-wafer for transmitters and the photodetector epi-wafer for receivers, then sells those wafers to midstream chip makers whose parts end up inside downstream optical transceivers. The company explicitly states it has been cooperating with international customers on 40G, 100G, and 400/800G high-speed modules for data centers.

### Moat Analysis
**Rating: Moderate (3/5)**

- **Direct exposure to the right layer**: unlike broader Taiwan electronics names, LandMark is actually selling III-V epi-wafers used in optical communications.
- **Relevant product roadmap**: the company highlights InP high-speed FP / DFB / EML epi-wafers, GaAs / InP high-speed photodetector epi-wafers, and products for 400G / 800G / 1.6T modules.
- **Small but accelerating**: 2025 revenue reached NT$2.20B versus NT$1.21B in 2024, and Jan-Mar 2026 revenue already reached NT$904M.
- **Not a clean chokepoint**: there are credible alternatives in III-V epi and materials, so this is better described as a specialized supplier with leverage to a hot market rather than a sole-source bottleneck.

### Supply Chain Position
**Upstream:** InP / GaAs substrates, MOCVD tools, specialty gases and process materials.  
**Downstream:** Optical chip makers and module vendors serving data center and telecom optics.  
**Competitors / alternatives:** IntelliEPI, IQE, other III-V epi specialists, and partially integrated optical-device vendors.  
**Risk:** Customer concentration is likely meaningful at this scale, though the company does not disclose a public customer list in the materials reviewed.

### Catalysts
1. **Taiwan photonics demand inflection**: DIGITIMES ranked LandMark as Taiwan's fastest-growing semiconductor manufacturer in 2025, with 76.2% cumulative growth through November 2025.
2. **Data-center product fit**: LandMark's own product pages show DFB, VCSEL, high-power CW laser, silicon-photonics epi-wafer bases, and GaAs PD epi-wafers aimed at data-center use cases.
3. **Margin recovery**: 9M 2025 gross margin improved to 40% from 16% in 9M 2024, showing operating leverage if the mix stays favorable.
4. **2026 carry-through**: Jan-Mar 2026 revenue of NT$904M is already roughly 75% of full-year 2024 revenue.

### Risks
1. **Small-cap execution risk**: one program slip or customer delay matters much more here than at larger optics vendors.
2. **China end-market exposure**: the 2024 annual report shows 73.4% of sales into China, which creates geopolitical and demand-concentration risk.
3. **Competitive niche**: III-V epi is important, but LandMark does not appear to control the layer tightly enough to claim monopoly pricing power.
4. **Technology path uncertainty**: if demand shifts toward architectures or vendors where LandMark is not designed in, the growth narrative can reverse quickly.

### CHIPS Act Update (2026-04-12)
Status: `NONE`.

I did not identify a direct US CHIPS incentives award or a CHIPS-linked R&D award for LandMark in the materials reviewed for this note. This remains a commercial Taiwan photonics thesis rather than a subsidy-linked one.

### Verdict: HOLD
LandMark is a real AI-photonics name, not a theme tourist. The company sits in the III-V epi-wafer layer that feeds optical transmitters and receivers, and the growth profile is strong enough to justify coverage. The reason this is not a `BUY` is structural: LandMark looks like a valuable specialist in a good market, but not yet like a proven bottleneck with durable pricing power.

**Final Score: 64/100**

### Update 2026-08-17 — 經濟日報 InP hike

Bloomberg 17 Aug: NT$2,670, mcap NT$272bn (~$8.5bn), TTM P/E 262x, P/B 58x, P/S 84x, GM 42.7%, OM 23.1%, TTM sales NT$2.20bn / NI NT$428m. 1M +80%, YTD +379%, 1Y +550%. 15 analysts, rec 4.47, BEST TP 3,134.

FY27E BEST: sales NT$5.34bn (+387m in 4 weeks), EPS 19.35 → ~138x. FY28E sales NT$10.47bn (+1.09bn in 4 weeks), EPS 41.70 → ~64x. Street is still chasing a name that is already paying for FY28.

Independently confirmed since the April note: MOPS 22 Apr 5-year InP substrate LTA with Sumitomo Electric (2026–2030) and ≤NT$13.15bn equipment capex; Q2 EPS NT$4.62 record; 2027 capacity guided 2.5×. The 17 Aug 經濟日報 “Q4 >10% hike” is still trade press.

China was 73.4% of 2024 sales — unchanged structural risk. MS had already moved to EW in March because the stock had discounted the good news. Verdict unchanged: HOLD. This is a real epi name at a discovery-era price that no longer exists.