Scrutica
| Technology | Category | Top Supplier | HHI | Substitutes | Demand Growth | Trajectory | Severity | Gov. Leverage |
|---|---|---|---|---|---|---|---|---|
| EUV Lithography Systems | Lithography | ASML (Netherlands) | 0% | 15% | Stable | 100 NL/EUUSJP | ||
| ABF Substrates (Ajinomoto Build-up Film) | Substrates | Ajinomoto (Japan) | 5% | 20% | Diversifying | 54 JP | ||
| Gallium & Germanium (Compound Semiconductors) | Materials | China (aggregated producers) (China) | 10% | 15% | Concentrating | 0 | ||
| EML Laser Chips (800G+ Optical Interconnect) | Optics | Lumentum (United States) | 15% | 160% | Concentrating | 49 US | ||
| High Bandwidth Memory (HBM) | Memory | SK Hynix (South Korea) | 5% | 55% | Stable | 82 US | ||
| Advanced 2.5D Packaging (CoWoS-class) | Packaging | TSMC (Taiwan) | 15% | 40% | Diversifying | 57 US | ||
| Chiplet Interconnects (UCIe) | Packaging | TSMC (CoWoS/InFO) (Taiwan) | 10% | 50% | Concentrating | 49 US | ||
| Wafer Inspection & Metrology Equipment | Equipment | KLA Corporation (United States) | 10% | 15% | Concentrating | 92 USNL/EUJP | ||
| Semiconductor-Grade Neon Gas | Materials | Ukraine (Ingas, Cryoin) (Ukraine) | 15% | 10% | Stable | 6 | ||
| US Data Center Grid Interconnection | Infrastructure | PJM Interconnection (United States) | 20% | 35% | Concentrating | 57 US | ||
| EDA Tools (Advanced Node) | EDA | Synopsys (United States) | 5% | 12% | Stable | 89 USNL/EU | ||
| EUV Photomasks & Pellicles | Materials | Toppan (Japan) | 5% | 20% | Concentrating | 58 JPUS | ||
| Photoresist (EUV-grade) | Materials | JSR Corporation (Japan) | 10% | 18% | Stable | 55 JP | ||
| Silicon Wafers (300mm) | Materials | Shin-Etsu Chemical (Japan) | 20% | 8% | Stable | 67 JPUSNL/EU |
No competitor has demonstrated EUV capability; ASML order backlog extends through 2027 with High-NA EUV (0.55 NA) shipping to Intel and TSMC.
The Netherlands controls the sole node for sub-5nm fabrication globally. Dutch export licensing decisions (via the Wassenaar Arrangement) determine which countries can manufacture frontier chips.
Ajinomoto investing ¥25B for 50% capacity expansion by 2030. Intel and SK Hynix (Absolics) developing glass substrates as alternative, but production timeline is 2028-2030.
A single Japanese company supplies the dielectric material for every advanced AI chip package. No US or European producer exists. Glass substrates offer an eventual escape path, but not before 2028.
China export suspension expires November 27, 2026 — a known deadline. USGS MCS 2025 confirms China produces 99% of primary gallium and ~68% of germanium. If restrictions are not renewed, the primary global source of gallium (GaN power electronics, GaAs optical components) and germanium (fiber optics, IR optics) is removed from the market.
The November 2026 deadline is a binary event embedded in a broader US-China trade framework renegotiation. Concurrent with rare earth deadline, creating a single high-stakes negotiation window. Risk profiles differ: gallium (99% China) is a near-monopoly; germanium (~68% China) has more non-Chinese supply but no rapid scale-up path.
70% supply-demand gap for EML laser chips. 800G+ transceiver demand growing 2.6x YoY (24M to 63M units). NVIDIA has pre-allocated supplier capacity, pushing non-NVIDIA lead times past 2027. Silicon photonics is an emerging substitute but requires different packaging. Note: market shares here reflect EML laser chip production (the binding constraint), not transceiver module assembly. Module-level concentration is lower because module integrators source chips from multiple suppliers.
Optical interconnect capacity is the binding constraint on AI cluster scale-out that GPU procurement timelines do not capture. A facility with GPUs but insufficient 800G+ transceivers operates at degraded throughput. Lumentum controls ~55% of EML laser chip production (the component inside the transceiver that cannot be substituted at the module level). US companies collectively control ~85% of EML chip production, but 2.6x YoY demand growth outpaces all announced capacity expansion.
Samsung staged a recovery from 17% (Q2 2025) to ~25% (Q3 2025) after HBM3E qualification with NVIDIA. SK Hynix share declined from 62% to ~56% but retains clear market leadership. Cross-qualification achieved (Samsung/Micron both NVIDIA-qualified for HBM3E), but ~81% production remains in South Korea.
South Korea controls ~81% of HBM production (SK Hynix + Samsung). A peninsula-wide event (grid failure, seismic, geopolitical) would remove the majority of AI accelerator memory supply. Micron (US, ~19%) is the only non-Korean producer. No allied coordination mechanism exists for HBM allocation during a crisis.
TSMC expanding from ~80K to ~130K wafers/month by end-2026; OSAT partners adding ~240-270K wafers/year capacity. Allocation concentration (NVIDIA >60%) is the binding constraint, not gross capacity.
Taiwan concentration in advanced packaging parallels the fabrication chokepoint. Even if logic chip fabrication diversifies (TSMC Arizona, Samsung Taylor), packaging remains Taiwan-centric through at least 2028.
UCIe 1.0 standard finalized but chiplet adoption accelerating faster than interconnect capacity. As monolithic die scaling slows beyond 3nm, chiplet architectures become mandatory — increasing demand for high-bandwidth die-to-die interconnects.
As monolithic die scaling stalls beyond 3nm, chiplet assembly becomes the binding constraint on next-generation accelerator design. UCIe is an open standard, but the packaging capacity to implement it at scale is concentrated: TSMC controls ~55% of advanced 2.5D/3D packaging capacity. The chokepoint is physical packaging capacity, not the standard itself.
KLA share rising from ~55% to ~63% as advanced node defect inspection complexity grows. Onto Innovation gaining share in overlay metrology (from KLA) but not in the e-beam inspection segment where KLA dominance is strongest. No Chinese competitor for sub-3nm inspection tools.
US companies control ~83% of wafer inspection equipment. KLA alone holds 60%. Unlike EDA tools (where the May 2025 restriction episode proved the chokepoint is politically actionable), inspection equipment is already export-controlled under Wassenaar and BIS rules — and is the enforcement mechanism that prevents advanced fab operation.
Post-2022 diversification reduced Ukrainian neon dominance from ~70% to ~55% for semiconductor-grade supply. Air Liquide and Linde expanded non-Ukrainian purification. However, Ukraine remains the single largest source; a renewed disruption (e.g., infrastructure strikes) would tighten DUV laser gas supply within months.
Neon is essential for DUV excimer lasers (248nm KrF, 193nm ArF) used for the majority of chip layers even at leading nodes. The 2022 Russia-Ukraine supply shock demonstrated the vulnerability; diversification has reduced but not eliminated single-country concentration.
PJM interconnection queue has 6,093 entries with ~30% completion rate. Average study time grown from ~2 years (2015) to 4+ years (2025). FERC Order 2023 reforms are in effect but have not yet reduced queue times.
Grid interconnection is the binding constraint on US AI compute expansion independent of GPU supply. Hyperscaler announcements of 1+ GW campuses are not credible without interconnection approvals.
May-July 2025 BIS restriction episode imposed and rescinded within 6 weeks after Chinese rare earth retaliation. No controls currently in force, but the episode proved the chokepoint is technically actionable and politically fragile.
US controls ~62% of advanced EDA tools; combined with Siemens (Germany), allied nations control ~76%. But no burden-sharing framework exists to distribute the costs of Chinese retaliation if EDA restrictions are reimposed.
At advanced nodes (<7nm), EUV photomask defect tolerance tightens to sub-nanometer scale, making mask fabrication yield-critical. EUV pellicles — ultrathin membranes that protect the mask during exposure — remain in limited supply; ASML is the primary pellicle developer and the only company with qualified EUV pellicle production. Japan controls ~70% of photomask blank production (Toppan, DNP, HOYA). As EUV layer counts increase at 3nm and below, photomask demand per wafer start grows proportionally.
Photomask and pellicle concentration creates a chokepoint comparable to EDA tools: invisible in most supply chain analyses but binding at advanced nodes. Japan controls the mask blanks; ASML controls the pellicles. A disruption to either halts EUV lithography regardless of ASML tool availability. Unlike EUV systems (controlled under Wassenaar), photomask materials face no export controls — the chokepoint is structural, not regulatory.
Japan controls ~88% of EUV photoresist production. JSR was acquired by Japan Industrial Partners (JIP) in 2024, keeping production domestic. No geographic diversification underway for EUV-grade materials.
Japan controls the chemical inputs for EUV lithography as completely as ASML controls the hardware. Japan imposed photoresist export controls on South Korea in 2019 (later relaxed), demonstrating willingness to use this lever.
Market structure has been stable for a decade. Japan holds ~54% share (Shin-Etsu + SUMCO, Omdia 2024). The 2011 Tohoku earthquake is the calibration precedent: ~25% global supply reduction, 6-month recovery to full capacity.
Every semiconductor fab on Earth depends on Japanese silicon wafer production. Unlike EUV (a single company), wafers are distributed across multiple Japanese sites, but Japan country concentration remains ~54%. Taiwan (GlobalWafers, 17%) is the second-largest producing country.
All assessments are Tier 3 (analyst judgment + public data). HHI (Herfindahl-Hirschman Index) is computed from estimated market shares: sum of squared market share percentages on a 0–10,000 scale. Severity score = HHI × (1 − substitute availability) × (1 + demand growth rate). Substitute availability is an editorial assessment on a 0–1 scale (0 = no substitute exists at any price; 1 = drop-in replacement available at scale). Demand growth rates are annual estimates from industry sources (TrendForce, 650 Group, Omdia). Trajectory assessments are updated quarterly. The severity formula is deterministic and fully reproducible from the inputs shown in each card. The severity scale is unbounded above 10,000 when demand growth exceeds 0% (the HHI component is capped at 10,000 but the growth multiplier is not). Color thresholds: critical (>7,000), elevated (>4,000), moderate (<4,000). When known market shares do not sum to 100%, the HHI is reported as a lower bound (marked with an asterisk); the true concentration is at least as high as the reported figure. Conversely, where multiple competitors are bundled into an “Others” group, the HHI is an upper bound relative to the actual fragmentation among those competitors.
Sources: ASML, TSMC, SK Hynix annual reports and earnings calls; TrendForce, Omdia, 650 Group / LightCounting, Astute Group market trackers; USGS Mineral Commodity Summaries 2025; MOFCOM Announcement No. 72/2025; PJM Interconnection Queue data; SemiAnalysis; Ajinomoto Co. Annual Report 2025.