DEEP RESEARCH · Park Systems
Park Systems (140860) — Q3 2025 Results and the Three-Year Tech Roadmap
From "academic AFM vendor" to "essential metrology partner for advanced semiconductor nodes" — financials, backlog, and R&D milestones.
0. Bottom line first
Park Systems is no longer a "research-tool maker." It sits at the convergence of AI/HBM/EUV/Advanced Packaging and has been elevated into the standard for non-destructive 3D nanoscale metrology. The proof: industrial-tool revenue is 74.6% of 9M25 sales of KRW 148.8B.
- Earnings: 9M25 revenue KRW 148.8B, OP KRW 33.7B (OPM ~22.7%), NI KRW 28.7B.
- Backlog: KRW 64.5B at Q3-end → KRW 68.1B as of 11/14 (industrial KRW 45.1B).
- Utilization: Research 97.06%, Industrial 96.55% — essentially full capacity.
- Roadmap: 2025 DHM-AFM hybrid · 2026 Wafer Bevel / mask defect chemistry · 2027 Gen 8.6 display + IMEC Wafer Bonding.
- Risks: chipmaker CAPEX cycle, US-China friction, FX volatility.
1. Executive summary
Park Systems (KOSDAQ: 140860) has transformed itself from an academic instrument vendor into a global semiconductor process-control solutions provider built on Atomic Force Microscopy (AFM). Its portfolio is perfectly aligned with three secular trends: (1) AI/HBM demand, (2) leading-edge node scaling (EUV/GAA), and (3) advanced packaging. The Lyncee Tec acquisition (March 2025) adds Digital Holographic Microscopy (DHM) — solving AFM's chronic throughput limitation and acting as a game changer.
2. Q3 2025 financials — structural change in revenue mix
2.1 Segment revenue
| Segment | 2023 FY | 2024 FY | 9M25 | Mix (9M25) | Driver |
|---|---|---|---|---|---|
| Research tools | KRW 40.0B | KRW 34.3B | KRW 26.4B | 22.98% | Universities & national labs |
| Industrial tools | KRW 72.4B | KRW 104.7B | KRW 85.8B | 74.61% | HBM, node scaling, packaging |
| Others (consumables, etc.) | KRW 2.5B | KRW 2.8B | KRW 2.8B | 2.41% | Installed base expansion |
| Total | KRW 114.8B | KRW 141.8B | KRW 115.0B (separate) | 100% | — |
Interpretation: A 74.6% industrial mix shows chipmakers replacing/augmenting legacy metrology with AFM. As devices move from 2D to 3D stacks (3D NAND, FinFET, GAA), AFM is the only non-contact 3D measurement option. The probe-consumable line acts as a cash cow tied to installed base.
2.2 Profitability & cost structure
~22.7%
Defended high OPM despite R&D and PMI costs.
14.6%
KRW 16.9B / separate revenue. Up from 11.2% (2023) → 12.1% (2024) → 14.6% (9M25).
NX-Wafer / NX-Mask
Stabilized COGS via richer product mix.
Lyncee Tec integration
R&D resources concentrated on hybrid platforms.
2.3 Backlog — forward visibility
- End of Q3 KRW 64.5B → as of 11/14 KRW 68.1B.
- Industrial KRW 45.1B (~70%) — drives Q4 and H1-26 alongside the semis upturn.
- Research KRW 15.5B — sustained academic demand.
3. Customer dynamics × investment thesis
3.1 HBM & advanced packaging
SK hynix, Samsung, and Micron are racing on HBM capacity. With HBM3E → HBM4 the stack height climbs from 8 → 12 → 16. Stack-die planarity and micro-bump height uniformity are decisive.
WLI + AFM
WLI scans the full field; AFM nails precise points — the HBM yield tool.
TSV sidewall/undercut
Verifies 3D shapes after through-silicon-via for stack integrity.
Interpretation: As HBM scales, metrology intensity per wafer rises super-linearly — Park Systems' tool demand grows faster than chipmaker wafer output.
3.2 Leading-edge nodes (EUV/GAA) and NX-Mask
Foundries and DRAM makers are accelerating EUV adoption for 2nm/1.4nm. EUV reticles cost hundreds of millions of KRW each and admit zero defects. NX-Mask mechanically removes defects via AFM-based repair without damaging the reticle (unlike e-beam approaches) — a structural growth area as advanced-node share rises.
3.3 Utilization recovery and the yield war
Memory production cuts of 2023–H1-24 ended; utilization is normalizing from H2-25. Leading nodes lead the recovery, amplifying metrology importance. NX-Wafer is used across front-end (CMP roughness) and back-end (defect review). Park Systems' proprietary True Non-Contact Mode minimizes tip wear — a major uptime advantage in 24-hour fab operations.
3.4 Customer concentration & risk
Top customers A 11% and B 10% — much improved from the legacy 50%+ single-customer concentration. Local subsidiaries in the US, Taiwan, China, and Japan support tight customer coverage. In China, despite US export controls, the Beijing office and partner channels preserve meaningful revenue.
4. R&D and the 3-year roadmap
4.1 R&D organization
| Unit | Role / Focus |
|---|---|
| Advanced R&D | Forward research, prototyping, next-gen platforms |
| Digital R&D | SW algorithms, AI defect analysis, control electronics |
| Systems R&D | Mechatronics, system integration, transfer to manufacturing |
| Park Systems GmbH (Germany) | Imaging Spectroscopic Ellipsometry (ISE) and optics |
| Lyncee Tec (Switzerland, new) | Digital Holographic Microscopy (DHM) |
Total R&D headcount: 126. The Lyncee Tec acquisition (March 2025) substantially expands optical metrology capability.
4.2 Roadmap — "speed innovation" × "measurement scope expansion"
- DHM-AFM Hybrid (25): DHM grabs full 3D info instantly; AFM nails precise targets — "Search and Destroy."
- PLP tool (25): Large-area measurement for square panel packaging.
- FX200 IR (25): IR spectroscopy added to the automated FX platform — shape + chemistry.
- Wafer Bevel metrology (26): Dedicated 3D wafer-edge measurement for 3D NAND-style stacks — a new market.
- Mask defect chemistry (26): Adds chemical analysis to NX-Mask to boost repair success rate.
- Dark Factory (26): Zero-touch operation from probe swap to data output.
- Gen 8.6 display / micro-LED (27): Leverages prior Gen 8.6 tool deliveries to capture display-grade hybrid optics.
- IMEC JDP deepening (27): Standardize wafer-bonding process control with IMEC — AFM as the Ångstrom-era reference.
5. Detailed financials & operations — details hidden in the footnotes
5.1 Production capacity & utilization
- Research: 97.06% (198/204 units).
- Industrial: 96.55% (56/58 units).
Effectively full capacity — CAPEX (facility) or outsourcing expansion is timely. Meanwhile high utilization helps OP leverage.
5.2 Raw materials & supply chain
PZT (piezo) price rose from KRW 750k (2023) to KRW 840k (9M25); vibration isolation tables stable at KRW 7,435k. In-housing and supplier diversification absorb cost increases via price pass-through and efficiency.
5.3 FX sensitivity & derivative management
Export ratio 88%. A 10% USD appreciation lifts pre-tax profit by ~KRW 4.8B. Currency options and forwards manage FX risk (derivative liability fair value ~KRW 0.14B).
5.4 IP
- Apparatus/method for target position recognition in AFM (Mar 2023) — core for automation.
- Surface topography measurement via color interferometry (Jan 2023) — likely WLI hybrid related.
- Magnetic-field generator + microscope (Jan 2024) — expands magnetic property applications.
Key assets that block Bruker-class competitor catch-up.
5.5 Talent (RSU)
9M25 RSU grants: 260 shares. Notably, 10,000 shares were granted to key Lyncee Tec staff to prevent PMI-stage attrition.
5.6 Balance sheet
- Shinhan Bank long-term facility loan ~KRW 36.5B, maturing 2027.
- Collateral incl. Gwacheon land payment refund claim (~KRW 11.5B).
- D/E ratio 48.03% (prior year-end 45.98%) — very healthy (<100%).
6. Conclusion & strategy
Investment points
- Decisive tech: Hybrid solution combining native AFM with WLI, DHM, and ISE.
- Structural growth: AI/HBM/EUV/packaging all stimulate tool demand.
- Financial stability: OPM, D/E, and cash flow all healthy — fuels continuous R&D and M&A.
Risks
- Semi CAPEX delay from a global slowdown.
- China revenue volatility from US-China tensions.
- Profit volatility from FX swings.
Expanded AFM adoption in HBM and advanced packaging is a secular trend. Full utilization, healthy backlog, and a concrete roadmap all argue for above-market growth over the next three years.
Sources
- Park Systems Quarterly Report (filed 2025-11-14) — issuer disclosure.
- Original (Naver Blog): link