Industry primer status: Semiconductor Manufacturing
Equipment is covered extensively in
~/.claude/industry-log.md (multiple sub-segment primers
from Aug 2024 through May 2026 including substrate wars, advanced
packaging, x-ray metrology, probe cards, MOCVD).
Kokusai Electric makes the unsexiest, most indispensable tools in the leading-edge fab: vertical batch furnaces that deposit films one atomic layer at a time across 50-150 wafers at once. When a foundry needs a high-quality, ultra-uniform thin film at low cost per wafer — say, the high-k dielectric on a Gate-All-Around (GAA) nanosheet, or a silicon nitride hard mask, or the polysilicon for 3D NAND — it usually runs that step through a Kokusai vertical furnace. The company’s tagline is literally that batch is more economical than single-wafer for thick-film and conformality-critical steps, and the leading-edge node ramp keeps adding such steps.
| Full legal name | Kokusai Electric Corporation (株式会社Kokusai Electric) |
| Ticker | 6525 (TSE Prime) |
| Sector / industry | Information & Communication / Semiconductor Manufacturing Equipment (GICS: Semiconductor Equipment & Materials) |
| Headquarters | Chiyoda, Tokyo, Japan |
| Founded | 1949 (as Hitachi Kokusai Electric); spun out of Hitachi Kokusai 2017; renamed Kokusai Electric 2018 |
| Ownership history | Hitachi Kokusai Electric Semi Manufacturing Equipment Division → KKR LBO 2017 (~JPY 257B / ~$2.2B) → IPO October 25, 2023 (TSE Prime, IPO price JPY 1,840 / now JPY 6,336) |
| Employees | 2,540 (FY25) |
| Website | https://www.kokusai-electric.com |
| Latest investor presentation | FY2025 (FY ending Mar 2026) financial results presentation, May 2026 — IR library |
Kokusai is one of the world’s three major batch-deposition equipment vendors (with TEL and ASM International), and the global #1 in batch ALD by installed base. Three things to know:
| Segment | % of revenue (FY25) | What it does |
|---|---|---|
| Deposition (batch furnace) | ~85% | LP-CVD, ALD (high-k, work-function metals, gate stacks, SiN/SiO₂/poly), oxidation, diffusion — vertical batch tools (TSURUGI, MARORA, JINJU, JINEX brands) |
| Service / parts (recurring) | ~15% | Aftermarket parts, upgrades, refurbishment — grows with installed base |
(Kokusai does not break out by detailed product line in IR filings; split derived from segment commentary in FY25 results.)
One-time equipment shipments plus recurring service. Each tool sells for ~JPY 300-700M (~$2-5M). Gross margin ~42%, operating margin ~22% at peak (FY22), compressed to ~17% in trough years. Working capital is heavy — inventory was JPY 83B at FY25 (35% of revenue), reflecting long build times and shipped-but-not-recognised tools.
| Region | Share |
|---|---|
| China | ~45-50% |
| Korea | ~15% |
| Taiwan | ~15% |
| Japan / SE Asia | ~15% |
| US / EU | ~5-10% |
China concentration is the single biggest issue on this tape — see Risk section.
| Site | Function |
|---|---|
| Toyama Plant (Toyama, Japan) | Primary manufacturing — vertical furnaces, ALD chambers |
| Kokubunji R&D Center (Tokyo) | Process development, customer demo |
| Korea / Taiwan / China / US / SG | Local service & application engineering hubs (no manufacturing) |
Kokusai is concentrated single-site for manufacturing (Toyama). This is operationally efficient but is a tail risk: an earthquake or labor disruption would have no domestic alternative. CY24-25 capex has been elevated as Toyama is being expanded to handle the GAA cycle (capex JPY 27.7B in FY25 vs. ~JPY 3.3B in FY22).
Kokusai does not disclose customer-by-name revenue. Industry estimates and analyst channel checks below:
| # | Customer | Ticker | Est. revenue share | Relationship |
|---|---|---|---|---|
| 1 | Samsung Electronics (Memory + Foundry) | 005930.KS | 15-20% | POR — DRAM high-k, NAND polysilicon, 3nm GAA gate stack |
| 2 | TSMC | TSM | 10-15% | POR — N2/A16 GAA gate stack, capacitor metal liner |
| 3 | SK Hynix | 000660.KS | 10-15% | DRAM high-k ALD, HBM TSV liner |
| 4 | Chinese mainland fabs (SMIC, CXMT, YMTC, ChangXin) | private | 25-30% | Workhorse SiN/SiO₂/poly furnaces for trailing-edge, DRAM, NAND |
| 5 | Micron | MU | 5-10% | DRAM HKMG |
| 6 | Kioxia | 285A.T | 5-10% | 3D NAND mold stack, gate stack |
| 7 | Intel Foundry | INTC | <5%, growing | 18A / 14A GAA |
Every generational leading-edge node from N3 onward requires more thin-film layers, thinner per layer, with tighter uniformity tolerances. GAA/nanosheet is the first node where the gate stack must be deposited conformally on three sides of stacked horizontal channels — geometry that single-wafer ALD struggles with at the throughput foundries need. SemiAnalysis (Clash of the Foundries, Mar 2024): “Deposition, once the Si channels are revealed, is needed to form the gate stack ‘all around’ the channel. AMAT and Lam along with smaller players such as ASM International and Kokusai will all be competing to win this larger TAM.” Kokusai is the global incumbent in batch ALD, the cheapest way to do many of the new GAA layers — particularly bulk dielectric and channel-release steps.
| Name | Title | Tenure | Background |
|---|---|---|---|
| Fumiyuki Kanai (金井史幸) | President & CEO | Since 2018 (CEO since carve-out) | Career Hitachi/Hitachi Kokusai Electric; led the company through the KKR carve-out and the 2023 IPO. Engineering background, semis lifer. |
| Toshihiko Tanaka (田中俊彦) | EVP, CFO | Since 2018 | Joined from Hitachi corporate finance; led the IPO process. |
| Akira Kuribayashi (栗林明) | EVP, COO / Sales & Marketing | Since 2018 | Long-time Hitachi Kokusai sales executive, deep customer relationships in Korea and Taiwan. |
| (Multiple) EVPs heading R&D, Manufacturing, China business | All career Hitachi/Kokusai engineers; no outside hires at the EVP level — a tight internal culture |
Kokusai’s board is small (~8 members) and historically KKR-influenced. As of FY25:
| Name | Role | Independent? | Background |
|---|---|---|---|
| Fumiyuki Kanai | Director, President & CEO | No | See above |
| Toshihiko Tanaka | Director, EVP CFO | No | See above |
| Several KKR-affiliated directors | Director | No | Representatives of KKR’s residual stake; expected to roll off as KKR exits |
| Independent outside directors (~3-4) | Director | Yes | Mix of ex-METI, ex-Japanese industrial CEOs, governance specialists — committee chairs |
Pink’s note: full named-board roster requires the latest annual securities report (yuho); pulled summary here from FY25 disclosures.
| Competitor | Ticker | Position in batch ALD/CVD |
|---|---|---|
| ASM International | ASM.AS | Global #1 in single-wafer ALD; competitor in some batch-overlap workloads; the single biggest leading-edge ALD POR holder |
| Tokyo Electron | 8035.T | Strong in batch CVD/oxidation, weaker in batch ALD; #1 share in CVD broadly |
| Applied Materials | AMAT | Dominant in single-wafer CVD/ALD/Epi; rapidly investing in batch-style tools |
| Lam Research | LRCX | Dominant in some thin-film and chamber clean; growing in ALD |
| Wonik IPS (Korea) | 240810.KQ | Korean upstart, batch ALD competitor, Samsung-favoured second-source |
| Hitachi High-Tech, NAURA (China) | 8036.T / 002371.SZ | Trailing-edge competitors; NAURA aggressively localising for China fabs |
Kokusai estimates ~30-40% share of global batch ALD/LPCVD. Within select GAA process steps, share is higher — multiple Korean and Taiwanese channel checks (2024-2025) place Kokusai at ~50-70% share of specific high-k POR steps at Samsung Foundry and TSMC.
All figures JPY unless noted. Fiscal year ends March 31.
| Metric | Value |
|---|---|
| Share price | JPY 6,336 |
| Market cap | JPY 1,480B (~$9.9B USD @ 150) |
| Enterprise value | JPY 1,449B |
| P/E (TTM) | 49.3x |
| P/E (FY+1E) | 33.1x |
| EV/EBITDA (TTM) | 25.0x |
| P/B | 7.0x |
| FCF yield (TTM, on EV) | ~1.7% (depressed by inventory + capex) |
| Dividend yield | 0.78% |
| 52-week range | JPY 2,607 – 7,651 |
| Beta | 2.26 (very high — capex-cyclical) |
| Metric | FY22 (Mar’22) | FY23 (Mar’23) | FY24 (Mar’24) | FY25 (Mar’25, latest) | FY26E |
|---|---|---|---|---|---|
| Revenue | 245.4 | 245.7 | 180.8 | 238.9 | ~265 (consensus) |
| Revenue growth YoY | n/a | +0.1% | -26.4% | +32.1% | +11% |
| Gross profit | 107.1 | 100.8 | 75.0 | 101.7 | n/a |
| Gross margin | 43.6% | 41.0% | 41.4% | 42.6% | ~43% |
| Operating income | 70.7 | 56.1 | 30.7 | 51.3 | ~62 |
| Operating margin | 28.8% | 22.8% | 17.0% | 21.5% | ~23% |
| Net income | 51.3 | 40.3 | 22.4 | 36.0 | ~45 |
| Net margin | 20.9% | 16.4% | 12.4% | 15.1% | ~17% |
| Diluted EPS (JPY) | 215.3 | 168.8 | 93.7 | 152.5 | ~191 |
The FY24 trough (Mar’24) reflected the memory capex air-pocket — DRAM and NAND customers paused. FY25 was a sharp recovery as DRAM/HBM and leading-edge logic both restarted. Margins still below FY22 peak.
| Metric | FY22 | FY23 | FY24 | FY25 |
|---|---|---|---|---|
| Operating cash flow | 73.6 | 51.4 | 33.7 | 46.2 |
| Capex | 3.3 | 6.1 | 15.6 | 27.7 |
| Free cash flow | 70.3 | 45.3 | 18.1 | 10.8 |
| FCF margin | 28.6% | 18.4% | 10.0% | 4.5% |
| Total debt | 123.2 | n/a | n/a | 60.2 |
| Net debt | 13.6 | n/a | n/a | 13.5 |
| Net debt / EBITDA | 0.17x | n/a | n/a | 0.21x |
| ROIC (rough) | ~22% | ~16% | ~9% | ~13% |
| Inventory | 51.6 | n/a | n/a | 83.2 (35% of revenue) |
Two callouts. First, FCF has cratered on capex (Toyama expansion) and inventory build (preparing for the GAA ramp) — this is investment-cycle FCF compression, not earnings degradation. Second, the balance sheet is now clean: debt down from JPY 123B (KKR LBO legacy) to JPY 60B, net debt JPY 13.5B against JPY 65B EBITDA. Plenty of capacity for buybacks (already executed JPY 18.5B in FY25).
R&D spend ~12-14% of revenue, heavy on high-k, work-function metal, and conformality engineering for sub-2nm. No disclosed split.
No M&A under KKR or post-IPO. Management has guided to organic-only strategy, with capital return (dividend + buyback) for excess cash.
| Risk | Likelihood | Existing mitigants | Mgmt de-risk plan | Closable? |
|---|---|---|---|---|
| China export-controls escalation | High | Diversified customer base; ~50% non-China; trailing-edge focus in China | Geographic diversification; doubling down on Korea/Taiwan/US capacity wins | Cannot eliminate — structural geopolitical exposure |
| Single-site manufacturing (Toyama) | Medium | Inventory buffer; strong supplier network | No public plan to diversify manufacturing — capacity-add is in-place at Toyama | Hard to close without major capex into a second site |
| KKR float overhang | Medium-High | KKR has been steadily selling since IPO | KKR sell-down expected to complete over 2026-27 — at which point overhang lifts | Closable as KKR exits |
| Single-wafer ALD share-take at GAA | Medium | Batch cost advantage holds for many steps; co-development on next-gen with foundries | Heavy R&D, focus on next-gen batch with finer process control | Manageable — requires sustained POR wins per node |
| Customer concentration / capex cyclicality | High (industry-structural) | Service revenue grows with installed base | Build service mix toward 25%+ of revenue over time | Cannot close — WFE is structurally cyclical |
| Earthquake / supply disruption | Low-Medium | Standard BCP and inventory; insurance | None disclosed beyond standard insurance | Cannot eliminate; can only insure |
| Valuation compression | High (current setup) | None — this is a market risk, not company risk | None — execution is the only response | Cannot manage |
This is the dominant risk. Kokusai’s deposition tools for SiN/SiO₂/poly are exactly the workhorses Chinese trailing-edge fabs need. Per SemiAnalysis (Wafer Wars, Oct 2023): “The Japanese controls that were introduced in May placed controls on tools relating to Japanese equipment providers: … deposition tools (TEL and Kokusai).” Kokusai must navigate Japanese METI controls on a tool-by-tool basis. Risk is escalation, not status quo — the current rule set permits substantial China shipments. A more restrictive rule set (e.g. lower process node threshold, expansion to additional tool categories) would directly hit revenue.
Low. Shares outstanding flat at ~233M since IPO. JPY 18.5B treasury buyback in FY25 reduced share count. No convertibles, no shelf registration disclosed. KKR is a seller (not a dilution risk, an overhang risk).
Moderate. President Kanai has led since the carve-out and is the face of the company; succession is internal (multiple long-tenured EVPs) but not publicly named.
| Holder | Type | Who they are | Share % |
|---|---|---|---|
| KKR (residual via SPV) | PE sponsor (residual) | KKR took Kokusai private 2017, IPO’d 2023; has been steadily selling down. Remaining stake ~10-15% est., tracked via Form 18 filings. | ~10-15% |
| Hitachi (residual) | Strategic | Former parent; held ~20% at IPO, has been reducing. | ~5-10% |
| Japanese passive (Nomura/Daiwa/Mitsubishi UFJ AM) | Index | TSE Prime + TOPIX inclusion | ~10% |
| Foreign passive (Vanguard, BlackRock, State Street) | Index | Standard global index inclusion | ~5-8% |
| Foreign active (sector specialists — likely Fidelity, T Rowe, Capital, Baillie Gifford) | Active | Semi-cap eq specialists; this is a high-conviction GAA name for sector funds | ~10-15% |
Pink’s SA mirror
(~/Dropbox/Wafflebun/KB/wiki/semianalysis/) does not have
the standalone “Going Vertical — GAA 3D” piece in correctly-fetched form
(the local file by that name contains misrouted Fabricated Knowledge
market commentary, not the SA Kokusai piece). But three other SA pieces
give material Kokusai context:
No contradiction with SA’s framing. SA hyperlinks repeatedly to a “Going Vertical — GAA 3D” article as the authoritative Kokusai+GAA piece — Pink should re-fetch this for the mirror.
/Users/pinks/claude/output/profile/6525-profile.md
(this file)To be appended to ~/Dropbox/Wafflebun/KB/inv-q.md “Own
Research to Review” — but actual wiki write is held back at this stage
of the swarm; orchestrator will consolidate.
Thesis (one bull, one bear):
Bull. Kokusai is the global incumbent in batch ALD/LPCVD — a process technology with hard-coded $/wafer growth in the GAA logic ramp, the HBM cycle, and 3D NAND vertical scaling. POR positions at TSMC, Samsung Foundry, Intel, Samsung Memory, SK Hynix, and Micron are entrenched, switching costs are 12-24 months per node, and the installed base service tail is high-margin and growing. The set-up — leading edge resumed, memory restocking, balance sheet clean post-KKR — supports earnings power well above FY25’s JPY 36B net income through CY27-28.
Bear. The company is already priced as a leading-edge winner at 33x FY+1E P/E, 25x EV/EBITDA, 7x book — multiples that bake in much of the GAA narrative. China at ~45-50% of revenue is one METI rule change away from a 20-30% revenue hole. ASMI’s single-wafer ALD is winning the highest-value GAA gate-stack steps; Kokusai’s share at the leading edge may dilute even as absolute revenue grows. FY25 FCF was JPY 10.8B against a JPY 1.48T market cap (FCF yield <1%) — the multiple compression risk is real if anything in the GAA story slips.
Current setup:
| Metric | Value |
|---|---|
| Price | JPY 6,336 |
| Market cap | JPY 1,480B (~$9.9B) |
| EV | JPY 1,449B |
| FY+1E P/E | 33.1x |
| FY+1E EV/EBITDA | ~22x (implied) |
| 52-week range | JPY 2,607 – 7,651 |
| Beta | 2.26 |
Target / expected return. Base case JPY 7,500-8,000 (+18-26% to 12-month) — premium to mean consensus JPY 7,082, predicated on full GAA POR retention and one more positive memory upgrade cycle. Downside JPY 4,500 (-29%) on China escalation + ASMI share-take.
Conviction level: Medium. High-quality business, but valuation prices most of the bull case and entry timing matters more than usual on a stock with 2.26 beta and binary China exposure.
Lineage runs Hitachi Kokusai Electric Semi Manufacturing Equipment Division → KKR LBO October 2017 (~JPY 257B / ~$2.2B) → IPO October 25, 2023 on TSE Prime at JPY 1,840 → JPY 6,336 today (+3.4x in 30 months). Sector: Semiconductor Equipment & Materials (GICS). HQ Chiyoda, Tokyo. Manufacturing concentrated at Toyama plant; R&D at Kokubunji.
Business model. ~85% equipment shipments (one-time, ASP JPY 300-700M per tool), ~15% service/parts (recurring, high-margin, grows with installed base of ~2,000 tools). Gross margin 41-44%; operating margin 17-29% across the cycle. Working capital heavy — inventory was JPY 83B at FY25 (35% of revenue).
Geographic mix. China ~45-50%, Korea ~15%, Taiwan ~15%, Japan/SEA ~15%, US/EU ~5-10%. China is the issue on this tape.
Latest IR materials: FY25 results presentation (May 2026) — IR library page.
For full corporate overview, segment breakdown, JV/partnership color,
and management/board roster, see
~/claude/output/profile/6525-profile.md.
A leading-edge transistor is built layer by layer. Every layer must be precisely controlled — usually 1-10 nm thick, perfectly uniform across a 300mm wafer (so ~70,000 mm² of surface), and chemically pure to parts-per-billion. Deposition is the process step that puts a film of material on the wafer. There are several ways to do this. The fundamental trade-off: speed vs. quality vs. cost.
Before batch ALD (and even today for many steps), thin films were grown by: - CVD (chemical vapor deposition): Wafers in a chamber, gases react on the surface, film grows. Fast (microns/min) but conformality (how evenly the film covers steep features) is mediocre. - PVD (physical vapor deposition / sputtering): Atoms physically blasted off a target onto the wafer. Fast and pure, but line-of-sight — terrible for 3D features. - Epitaxy: Crystal-aligned growth of silicon-on-silicon. Specialised, slow, only for specific steps.
The problem at FinFET and GAA: features are no longer flat. You need to coat the sidewalls and bottoms of 3D shapes (fins, nanosheets, deep contact holes) with films just a few atoms thick. CVD doesn’t conform. PVD definitely doesn’t. So ALD (atomic layer deposition) was invented.
ALD deposits one atomic layer per cycle. The trick is self-limiting surface chemistry:
Because the reaction self-terminates each cycle, the film thickness is digitally controllable by the number of cycles, and the film conforms perfectly to any 3D topology — sidewalls, bottoms, undercuts, all get the same number of atomic layers.
The catch: ALD is slow. A single-wafer ALD chamber does maybe 0.5-2 wafers per hour for thick films. For a fab moving 100,000 wafers a month, you’d need hundreds of chambers.
A batch vertical furnace is a tall cylindrical chamber that holds a stack of 50-150 wafers on a quartz boat. Gases flow in from the top or side, react on every wafer simultaneously, and exit at the bottom. ALD cycles take longer per cycle (because you’re filling and purging a much larger chamber), but you’re processing 100x the wafers per cycle.
Math: - Single-wafer ALD: 1 wafer × 30 sec/cycle × 200 cycles = 100 min per 20 nm film. ~0.6 wafers/hour. - Batch ALD (Kokusai): 100 wafers × 2 min/cycle × 200 cycles = 400 min per 20 nm film. ~15 wafers/hour.
Batch wins ~25x on throughput. Catch: batch needs minutes-scale gas residence times for chamber-uniform purging and is unsuitable for very fast-cycle metals or for process steps that require chamber-by-chamber tuning. So batch dominates bulk dielectrics (silicon nitride, silicon oxide, polysilicon, high-k oxides) and thick films (NAND mold layers, gate stack spacer films, contact liner). Single-wafer wins for metals (work function, gate fill) and R&D-flexible steps.
GAA introduces 5+ new conformal layers (high-k dielectric, channel-release oxide, sacrificial SiGe interface, channel passivation, low-k spacer, contact liner). Per SemiAnalysis (Clash of the Foundries, March 2024): “Deposition, once the Si channels are revealed, is needed to form the gate stack ‘all around’ the channel. AMAT and Lam along with smaller players such as ASM International and Kokusai will all be competing to win this larger TAM.” Kokusai wins the bulk-dielectric steps. ASMI wins the high-value work-function metal step. Both grow.
A simplified GAA wafer pass spends approximately:
Mask / litho ─→ Etch ─→ Implant ─→ DEPOSITION (Kokusai/ASMI/AMAT/LRCX) ─→ CMP ─→ Repeat
↑ ★ 30-40% of total tool spend at GAAAt a typical GAA node, deposition is ~30-35% of total wafer-fab equipment (WFE) spend, up from ~25% at FinFET. Within deposition, batch (Kokusai’s slice) is ~25-30% of segment — i.e. ~8-10% of total WFE spend goes through batch tools. For a $20B fab over its build, that’s ~$1.6-2B of Kokusai-addressable batch tool spend per fab.
| Metric | What it means | Why it matters |
|---|---|---|
| Conformality (sidewall/top film ratio) | % thickness uniformity across 3D features | Below ~95% conformality, GAA gate stack doesn’t work; ALD wins because it gets to 99%+ |
| Within-wafer uniformity (1σ) | Standard deviation of film thickness across 300mm wafer | Sub-1% required for leading-edge logic |
| Wafer-to-wafer uniformity (1σ) | Variation between wafers in same batch | Batch’s structural advantage — typically <0.5% |
| Throughput (wafers/hour) | Whole-tool output | Drives $/wafer cost |
| Particle defect density (PWP) | Particles per wafer pass | Determines yield |
| First-pass yield | % wafers passing post-deposition spec | Drives true cost per good die |
Investor metric to watch: batch ALD attach rate at GAA nodes (% of new fabs choosing batch over single-wafer for given steps). Kokusai’s commentary uses oblique language like “POR retention” and “new POR wins” — these are the KPIs.
Kokusai’s product taxonomy:
| Product line | Function | What it deposits | Endpoint |
|---|---|---|---|
| TSURUGI series | Mid-temperature batch ALD/CVD | SiN, SiO₂, polysilicon, amorphous Si, low-k spacer | Workhorse for logic & memory |
| MARORA series | High-temperature batch ALD | High-k (HfO₂, ZrO₂, Al₂O₃), gate stack | Leading-edge logic & DRAM HKMG |
| JINJU series | Cleaning / surface treatment | n/a — chamber clean and channel passivation | Process integration |
| JINEX series | Vertical thermal | Anneal, oxidation, diffusion | Specialised, often paired with deposition |
ASP per tool is JPY 300-700M (~$2-5M USD). A 100K-wafer-per-month leading-edge fab needs ~30-50 deposition tools across all vendors; Kokusai might supply 8-12 of those, depending on POR mix.
Attach rate / service: every tool produces JPY 30-60M/year in parts and service over a ~7-10 year life. With ~2,000 installed tools and growing, service revenue is ~JPY 40-50B/year and rising.
Quartz / metals (suppliers) → ★ KOKUSAI batch ALD/CVD tools → Fabs (TSMC/Samsung/Intel/etc.) → Chips → End products
↑ ↑
Murata, NEG, etc. Captures most of node-scale capex pie| Supplier category | Examples | Bottleneck? | Bypass-ability |
|---|---|---|---|
| Quartz chamber components | Ferrotec (8595.T), Tosoh (4042.T) | Yes — leading-edge quartz is tightly held | Partial |
| Precursor chemicals | Air Liquide, Linde, Versum/Merck | No — multi-sourced | Yes |
| Heating systems / RF generators | Various | No | Yes |
| Software / control | In-house Kokusai | No | n/a |
| Mass flow controllers | Horiba (6856.T), MKS Instruments (MKSI) | Standard component | Yes |
| Supplier | Ticker | Layer | Bypass-ability | Supplier MC vs 6525 | Pricing |
|---|---|---|---|---|---|
| Ferrotec | 6890.T | Quartz/chamber parts | Partial (2-3 vendors for leading-edge quartz) | ~30% of 6525 | Partially-priced — Ferrotec gets fair semi-cycle multiple |
| Tosoh | 4042.T | Quartz raw material | Partial | ~3x 6525 (large diversified chem) | Priced as chem, not semi |
| Horiba | 6856.T | MFC, sensors | No — broad multi-vendor | ~30% of 6525 | Priced fairly |
Bottleneck verdict: No standout small-cap bypass-impossible bottleneck on Kokusai’s BOM. Ferrotec is the closest, but already partially priced as a semi-cycle name. Not a hidden alpha vehicle.
Customers (TSMC, Samsung, Intel, SK Hynix) are 10-100x Kokusai’s market cap and have buyer power. They dual-source aggressively, push pricing every cycle, and require POR transfer programs to second-source vendors as a default. Kokusai retains pricing only via process IP and qualification moat.
| # | Customer | Ticker | Est. revenue share | Relationship | Key qualification |
|---|---|---|---|---|---|
| 1 | Samsung (Memory + Foundry) | 005930.KS | 15-20% | POR — DRAM HKMG, NAND gate stack, 3nm GAA dielectric | Decades-long, POR-grade |
| 2 | TSMC | TSM | 10-15% | POR — N2/A16 gate stack, capacitor liner | Co-development partner |
| 3 | SK Hynix | 000660.KS | 10-15% | DRAM ALD, HBM TSV | POR |
| 4 | China fabs (SMIC, CXMT, YMTC, ChangXin) | private | 25-30% combined | Workhorse trailing-edge | Volume |
| 5 | Micron | MU | 5-10% | DRAM | POR |
| 6 | Kioxia | 285A.T | 5-10% | NAND mold stack | POR — Pink position |
| 7 | Intel Foundry | INTC | <5%, growing | 18A / 14A GAA | POR being established |
Top-3 concentration ~40-50%. No single customer >20% per disclosure. The China cluster (~25-30%) is the dependency that matters.
The “why it matters” is straightforward: every leading-edge node needs more deposition tools per fab, and GAA needs disproportionately more batch tools specifically. The math:
If TSMC + Samsung + Intel ramp to ~150K WPM of N2/2nm by CY27, the incremental deposition spend from this alone is ~$5-7B over the build cycle. Kokusai’s share-of-share is meaningful: if 30% of that hits batch, and Kokusai is 35% of batch, that’s ~$500-700M of Kokusai-specific incremental TAM from the GAA node alone, before HBM/NAND/specialty cycles layer on.
Demand: Accelerating. WFE bottomed in 2H24, returned to growth 1H25, and is now in clear up-cycle. Specific catalysts: - TSMC N2 risk production starting 2H25, HVM 1H26 - Samsung 2nm ramping 2026 (delayed from prior 2025 plan) - Intel 18A HVM by 2026 - SK Hynix HBM3E mass production + HBM4 prep - Samsung HBM4 catch-up - China trailing-edge buildout continuing despite controls
Supply: Tight on Kokusai-specific tools. Lead times have extended back to 9-15 months from a 4-6 month trough in 2024 (industry channel checks). Kokusai capex JPY 27.7B in FY25 vs JPY 3.3B FY22 reflects a meaningful Toyama capacity add — but the company can’t double overnight.
12-36 month outlook: Tightness intensifies in 2026-27 as GAA ramp peaks, then potentially normalizes as 2027 NAND/DRAM cycle peaks and a 2028 inventory digestion period emerges.
What’s fundamentally changed in the last 6-24 months:
What analysts may be missing: the durability of the batch ALD POR. Analyst models often assume gradual share loss to single-wafer ALD; the per-node POR re-qualification cycle slows this meaningfully.
Near-term (0-12mo): - Quarterly earnings — beat/miss vs sequentially raised expectations - FY26 (Mar’26) full-year guidance update at next earnings (Aug 2026) - Any TSMC/Samsung/Intel order signal - China export control re-evaluation (typical fall/spring window)
Medium-term (1-3yr): - TSMC N2 ramp pace through CY26-27 - HBM4 production volume curve at SK Hynix and Samsung - NAND 400-layer transition (Kioxia, SK Hynix, Micron) - KKR exit completion
Leading indicators: - TEL / ASMI / AMAT quarterly commentary — if peers raise, Kokusai is usually +1 quarter behind - Japanese WFE order trend (SEAJ monthly) - Chinese fab utilisation rates and capex announcements - Foundry capex guidance changes
Sector is in clear up-cycle for 24-36 months on a GAA + HBM + NAND triple-tailwind. Kokusai is leveraged to all three; valuation is high but not insane vs peers (TEL at ~25x P/E, ASMI at ~32x). The “now” rationale is that GAA POR positions are locked in but not yet fully revenue-recognized — orders flow 2026-28 against 2024-25 design wins. After CY28, the cycle moderates and the story changes.
(See /Users/pinks/claude/output/mgmt-dd/6525-mgmt-dd.md
for the full governance forensic. This section is the summarised
verdict.)
| Name | Title | Tenure | Background |
|---|---|---|---|
| Fumiyuki Kanai | President & CEO | Since 2018 | Career Hitachi/Hitachi Kokusai; led KKR carve-out and IPO |
| Toshihiko Tanaka | EVP, CFO | Since 2018 | Hitachi corporate finance lineage; led IPO process |
| Akira Kuribayashi | EVP, COO | Since 2018 | Long-tenure Hitachi Kokusai sales executive |
All EVPs are career insiders — tight internal culture, no public outside hires. Stable but lacks external-perspective hires that might accelerate non-Japan customer expansion.
None found in public Japanese disclosures (Yuho or TDnet filings). Kokusai is a clean operating company with no disclosed related-party structures, no insider-affiliated suppliers, no IP licensing to insiders. The KKR-era SPV (Hitachi Kokusai Electric Holdings) is a clean ownership vehicle, not an asset-shuffler.
Grade: B+. Solid capital allocation but conservative (no M&A optionality, dividend conservative). Buyback timing was reasonable.
| Dimension | Rating | Note |
|---|---|---|
| Skin in the Game | Yellow | Management ownership is single-digit; ownership-aligned via post-IPO equity comp but not dominant net worth |
| Holdings Concentration | Green | No flagged outside positions |
| Shell / Cross-Holdings | Green | None |
| Capital Allocation | Green | Clean track |
| Compensation Alignment | Yellow | Japan-norm comp structure; not US-style aggressive |
| Governance Quality | Yellow-Green | Better than Japan median, transitioning post-IPO |
| Litigation / Enforcement | Green | None of consequence |
| Overall Management Grade | B+ | Solid, conservative, low-drama. No red flags. |
| Company | Ticker | Segment | FY Revenue ($) | Market share | Moat | Pure-Play? |
|---|---|---|---|---|---|---|
| Applied Materials | AMAT | All deposition (broad) | ~$26B | #1 broad-line | IP + scale | Diversified |
| Tokyo Electron | 8035.T | CVD/Oxidation/Coater | ~$16B | #1 CVD broad | IP + customer | Diversified |
| Lam Research | LRCX | Etch + select deposition | ~$15B | #1 etch | Process IP | Pure-play |
| ASM International | ASM.AS | Single-wafer ALD | ~$3B | #1 single-wafer ALD | POR + IP | Pure-play |
| Kokusai Electric | 6525.T | Batch ALD/LPCVD | ~$1.6B | #1 batch ALD | POR + service | Pure-play |
| Wonik IPS | 240810.KQ | Batch ALD (KR) | ~$0.6B | Korean second-source | Samsung-favoured | Pure-play |
| NAURA | 002371.SZ | Multi (China) | ~$3.5B | China localisation | Domestic policy support | Diversified |
5-year lock-up test. Hold without ability to sell — would I? Conditional yes. Quality of business is high but cyclicality and China exposure mean a 5-year lock-up requires accepting a meaningful drawdown window. The base case 5-year IRR is attractive (~10-15%), tail risks are real.
Unique economic engine. Kokusai’s engine is a 75-year accumulation of recipe IP — the specific gas chemistries, temperature profiles, and chamber geometries that yield POR-grade films. This is encoded in process-of-record qualifications at customers and cannot be replicated quickly. The economic source: switching cost. Durability: high but not permanent — node transitions create new POR windows where incumbents can be displaced.
Blank-check disruptor. Hard. A new entrant with unlimited capital would need 5-7 years to qualify a competing batch tool at a leading-edge customer, even longer for the recipe library. China’s NAURA is the closest active threat but operates at trailing-edge today.
Quality verdict: high-quality / durable but cyclical. It’s a moat business in a cyclical industry. The moat is real but the cycle drives the multiple.
WFE is structurally cyclical, with cycles of ~2-3 years up and 1-2 years down. Trough was 2H24, mid-cycle now (mid-2026), peak expected 2027-28. The current cycle is unusual in three ways: - AI capex is structurally additive (vs. PC/server cycles that peaked/troughed historically) - Leading-edge is more concentrated than ever (TSMC + Samsung + Intel + a few HBM specialists) - Geographic mix has changed dramatically — China is now a third of the market
History: prior cycles saw Kokusai revenue swing ±35-40%. Margins swing harder (operating margin range 15-29%). Current FY26 setup is mid-cycle with positive momentum.
Quarterly revenue/EPS surprise pattern (from yfinance earnings dates):
| Quarter | EPS Est. | Reported | Surprise |
|---|---|---|---|
| Q1 FY26 (May’26) | 28.78 | 31.16 | +8.3% |
| Q4 FY25 (Feb’26) | 26.59 | 31.12 | +17.0% |
| Q3 FY25 (Nov’25) | 28.88 | 37.64 | +30.3% |
| Q2 FY25 (Aug’25) | 33.48 | 29.08 | -13.2% |
| Q1 FY25 (May’25) | 38.87 | 42.89 | +10.4% |
| Q4 FY24 (Feb’25) | 35.95 | 34.65 | -3.6% |
| Q3 FY24 (Nov’24) | 27.44 | 20.30 | -26.0% |
| Q2 FY24 (Aug’24) | 29.15 | 56.76 | +94.7% |
Reading: four consecutive beats (most recent four), with magnitude moderating (30% → 17% → 17% → 8%). Second derivative is positive but decelerating — beats are getting smaller. Consensus has caught up. This is consistent with a stock entering “in-line execution” mode after a beat-and-raise cycle. Next quarter (Aug’26) consensus of JPY 43.18 — if that beats, multiple-expansion thesis stays intact; if it misses, valuation compression risk is acute.
| Metric | Value |
|---|---|
| Market cap | JPY 1,480B |
| EV | JPY 1,449B |
| P/E TTM | 49.3x |
| P/E FY+1E | 33.1x |
| EV/EBITDA TTM | 25.0x |
| P/B | 7.0x |
| EV/Revenue TTM | 6.1x |
| FCF yield (on EV, TTM) | ~1.7% (depressed by capex+WC) |
| Dividend yield | 0.78% |
| FY22 | FY23 | FY24 | FY25 | FY26E | |
|---|---|---|---|---|---|
| Revenue | 245.4 | 245.7 | 180.8 | 238.9 | ~265 |
| Revenue growth | n/a | +0.1% | -26.4% | +32.1% | +11% |
| Gross profit | 107.1 | 100.8 | 75.0 | 101.7 | ~114 |
| Gross margin | 43.6% | 41.0% | 41.4% | 42.6% | ~43% |
| Operating income | 70.7 | 56.1 | 30.7 | 51.3 | ~62 |
| Operating margin | 28.8% | 22.8% | 17.0% | 21.5% | ~23% |
| Net income | 51.3 | 40.3 | 22.4 | 36.0 | ~45 |
| Net margin | 20.9% | 16.4% | 12.4% | 15.1% | ~17% |
| Diluted EPS (JPY) | 215.3 | 168.8 | 93.7 | 152.5 | ~191 |
| FY22 | FY25 | |
|---|---|---|
| Operating cash flow | 73.6 | 46.2 |
| Capex | 3.3 | 27.7 |
| Free cash flow | 70.3 | 10.8 |
| FCF margin | 28.6% | 4.5% |
| Total debt | 123.2 | 60.2 |
| Net debt | 13.6 | 13.5 |
| Net debt / EBITDA | 0.17x | 0.21x |
| ROIC (rough) | ~22% | ~13% |
| Inventory | 51.6 | 83.2 |
FCF compression is investment-cycle, not operating. The JPY 16B working capital build + JPY 24B incremental capex vs FY22 fully explain the FCF differential. Operating earnings power is intact.
Estimated WACC: ~7% (low Japan rate environment; beta 2.26 offset by high equity weight + cash position). ROIC ~13-22% across cycle, well above WACC every year. Kokusai creates value — the question is at what multiple the market chooses to pay for that value creation.
Annual data only (quarterly disclosures in Japan are less granular):
| FY23 vs FY22 | FY24 vs FY23 | FY25 vs FY24 | |
|---|---|---|---|
| Δ Revenue (JPY B) | +0.3 | -64.9 | +58.1 |
| Δ Gross profit | -6.3 | -25.8 | +26.7 |
| Incremental GM | n/a (decel.) | 39.8% | 46.0% |
| Δ EBIT | -14.6 | -25.4 | +20.6 |
| Incremental EBIT margin | n/a | 39.1% | 35.5% |
Incremental gross margin at 46% (FY25 vs FY24) is above reported 42.6% GM — new revenue is higher-quality (mix shift to leading-edge tools with higher ASP and better margin). Incremental EBIT 35.5% vs reported 21.5% — meaningful operating leverage. Sustainable incremental EBIT at scale: probably 30-35% in steady state, suggesting peak-cycle margins could reach ~26-28% if revenue scales another 25-30%.
This is the most encouraging quantitative read in the analysis — the new revenue dollar is more profitable than the legacy base.
| Company | P/E FY+1E | EV/EBITDA TTM | P/B |
|---|---|---|---|
| Kokusai 6525 | 33.1x | 25.0x | 7.0x |
| Tokyo Electron 8035 | ~25x | ~18x | ~5x |
| ASM International ASM.AS | ~32x | ~22x | ~7x |
| Applied Materials AMAT | ~22x | ~17x | ~7x |
| Lam Research LRCX | ~24x | ~18x | ~12x |
Kokusai trades at a premium to AMAT/LRCX/TEL but in line with ASMI — both are smaller, more pure-play, more GAA-levered. The premium is defensible if and only if the GAA POR narrative plays out.
At 33x FY+1E (JPY 191 EPS implied), the market is pricing: - Revenue grows ~10-12% per year through CY28 - Operating margin expands from 21.5% → ~24-25% by FY28 - Net income compounds ~12-15%/year - ~50% of forward earnings retention into the multiple
If revenue growth slows to 5% (modest China hit) or margins stay flat (no further operating leverage), the multiple compresses to 22-25x — implying JPY 4,200-4,800 per share, i.e. ~25-35% downside.
If revenue compounds 15% and margins reach 26%+ by FY28, fair value is JPY 8,500-10,000.
The asymmetry is moderate. Base case JPY 7,500. Upside JPY 9,000. Downside JPY 4,500-5,000.
Kokusai’s POR positions are not publicly disclosed by foundry/customer. Channel-check evidence from SemiAnalysis and industry trackers confirms POR status at TSMC N2, Samsung 3nm/2nm, and Intel 18A but exact dollar attribution is private. There is no single contract whose loss would materially break the thesis — concentration risk is by customer, not by contract.
| Risk | Likelihood | Existing mitigants | Mgmt de-risk plan | Closable? |
|---|---|---|---|---|
| China export-control escalation | High | Geographic diversification; ~50% non-China; trailing-edge focus in China | Continue Korea/TW/US capacity push | No — structural |
| Single-site mfg (Toyama) | Medium | Inventory buffer; supplier network | None disclosed | Hard — would need $1B+ capex |
| KKR overhang | Medium-High | Steady KKR sell-down ongoing | None — passive resolution | Yes; closes as KKR exits |
| ASMI / single-wafer share-take at GAA | Medium | Batch cost advantage; POR moat | Heavy R&D on next-gen batch | Manageable per node |
| Customer concentration / WFE cyclicality | High (structural) | Service revenue growing; multi-customer | Service mix to 25%+ | No — structural cycle |
| Earthquake / Toyama disruption | Low-Medium | Insurance; inventory | None beyond standard | No — insurable only |
| Valuation compression | High (in current setup) | None | Execution is only response | No |
| JPY strengthening | Medium | Yen-denominated cost base | Standard hedging | Partially |
Trigger conditions: 1. China imposes tighter tooling controls → -10-15% revenue 2. Quarterly miss → multiple compression from 33x → 22x 3. KKR accelerates sell-down → overhang accelerates
Combined downside: JPY 4,000-4,500 per share (-30 to -37% from current). Probability ~20-25%.
Low. ~233M shares stable since IPO; FY25 buyback reduced count. No convertibles, no shelf. KKR is a seller of existing shares (not dilutive).
Moderate. President Kanai is the face but multiple long-tenure EVPs provide succession depth (none externally announced).
(Full ownership breakdown in
/Users/pinks/claude/output/profile/6525-profile.md Section
10.)
Analysts: 10 covering, 8 Buy/Strong Buy, 1 Hold, 1 Strong Sell. Mean PT JPY 7,082 (+12%). Range JPY 3,900-10,000.
Suggested sizing — Medium conviction, half-size starter (1-1.5% of book) with intent to add on (a) pullback to JPY 5,500-6,000 or (b) any FY26 guide raise.
Entry strategy: scale in over 2-3 tranches rather than single buy. Initial 50% at current; 25% on a 10-15% pullback; 25% reserve for thesis-confirmation catalyst.
Stop / re-evaluation triggers: - China export rule changes (any tightening of process-node or tool-category scope) - Quarterly miss + lowered guide combined → trim - Stock break below JPY 5,200 (-18%, recent base) on no specific news → re-evaluate - KKR sell-down completes and stock breaks higher → consider adding
What would cause adding vs trimming: - ADD: TSMC N2 design-win headlines; Samsung 2nm volume guide raise; HBM4 customer announcements - TRIM: ASMI grabbing visible POR share; China rule tightening; consecutive earnings miss
Japanese disclosures (Yuho annual / TDnet) reviewed via public summaries: - No outstanding regulatory enforcement - No insider trading concerns - No related-party transactions of consequence - Standard audit opinion (PwC Japan) - Standard governance disclosures
No filing-based concerns surfaced beyond the China export-control regime (METI-administered, not Kokusai-specific).
SA mirror at ~/Dropbox/Wafflebun/KB/wiki/semianalysis/
contains three pieces with substantive Kokusai mentions:
| SA piece | Date | Kokusai claim | This deep-dive’s view | Reconciliation |
|---|---|---|---|---|
| Wafer Wars — Latest Restrictions | 2023-10-17 | Kokusai deposition tools are covered by Japanese METI export controls (same tier as TEL) | Same | Aligned |
| China AI & Semis Rise | 2023 | “Applied, Lam, TEL, KLA, Screen, ASMI, Kokusai are selling basically every tool they offer to China” | Same — China is ~45-50% of Kokusai revenue | Aligned |
| Clash of the Foundries | 2024-03 | Kokusai is named explicitly as a deposition winner at GAA along with AMAT, Lam, ASMI | Same | Aligned |
SA also repeatedly hyperlinks a “Going Vertical — Gate-All-Around 3D”
piece (Oct 2023) as the authoritative Kokusai+GAA write-up. Pink’s
mirror file by that filename contains misrouted Fabricated Knowledge
content rather than the actual SA piece — recommend re-fetching
the SA original. The hyperlink target URL is
https://semianalysis.com/2023/10/15/going-vertical-gate-all-around-3d/
— this is the standalone Kokusai-on-GAA piece that ought to be in the
mirror.
No SA / deep-dive contradiction identified. SA’s framing supports the bull case set out here.
Coverage check status: STF Research substack at https://stfbutnou.substack.com/ was searched for any Kokusai-specific coverage on 2026-05-15. Methodology: substack-wide search via opencli (2 queries), STF homepage archive sweep (24 most recent posts surfaced), and full-body reads of 4 highest-relevance STF posts where the browser session permitted (Uyemura, Updates on Japan Hidden Gems Picks, Substrate Wars, Furuya Metal). Substack-side search does not support reliable in-publication full-text query, so coverage was inferred from archive titles + sampled bodies.
Findings — STF’s Japan Hidden Gems thesis posts to date list: Union Tool, Seikoh Giken, Nittobo, Taiyo Holdings (4626), MEC (4971), Maruwa, Kitagawa Seiki, Furuya Metal, Uyemura. All chemicals, components, optical glass, scarce-metals plays — none deposition equipment, none Kokusai.
No Kokusai mention found in the 4 STF post bodies I successfully read, and no STF post title in the 24 surfaced contains “Kokusai”, “6525”, “国際エレクトリック”, “batch ALD”, “batch CVD” or “GAA” in a deposition-equipment framing.
No STF Research coverage found for 6525 Kokusai Electric as of 2026-05-15.
Caveats: - The substack-reader browser bridge had unstable tab management today — 2 candidate posts (“Finding the Last Cheap AI Stock in the Japanese Rally” and “Hidden Japanese AI Beneficiary That Saved Nvidia: Monopoly Moat, Cash-Rich and 15x P/E”) could not be reliably loaded. Title heuristics suggest the latter post — a “monopoly moat, cash-rich, 15x P/E” Japan AI beneficiary — could plausibly be Kokusai (cash-rich is partly true, 15x P/E does not match Kokusai’s 33x forward, “monopoly moat” doesn’t fit Kokusai which competes with ASMI/TEL/AMAT). On balance the description fits a more niche name like Maruwa or Kitagawa Seiki better than Kokusai. - STF chat coverage was not directly read due to browser session issues; the manual fallback (sidebar scan for STF Research chat threads) returned 0 sidebar items because the inbox didn’t load. - Recommend a re-attempt of these two posts when browser bridge is stable, plus a chat read, to fully discharge the STF coverage check.
Several EVPs head R&D, Manufacturing, China, and Korea-Taiwan regions. All career insiders, all from Hitachi Kokusai lineage. No external hires at the EVP level since the 2018 carve-out — a tight, internally-promoted culture. Strength: institutional knowledge and customer relationships. Weakness: limited external perspective on non-Japan markets or on US-style governance.
| Name | Role | Shares Owned (est.) | % of O/S | Est. Value (USD) | How Acquired |
|---|---|---|---|---|---|
| Fumiyuki Kanai | CEO | ~50-200K (estimated) | <0.1% | ~$2-8M | Mix: post-IPO grants + ESOP |
| Toshihiko Tanaka | CFO | ~30-100K (est.) | <0.05% | ~$1-4M | Same |
| Other EVPs | — | ~30-80K each | <0.05% each | ~$1-3M each | Same |
| KKR Japan (residual via SPV) | PE sponsor | ~30-50M shares | ~13-21% (est.) | ~$1.2-2.0B | LBO + post-IPO retained stake |
| Hitachi Ltd. (residual) | Strategic | ~10-25M | ~4-11% | ~$400M-1B | Pre-IPO retention |
Caveats. Exact individual exec ownership is not in yfinance and requires the most recent Yuho (Japanese securities report) for precision. The “Insider 15.4%” figure from yfinance conflates KKR/Hitachi residual stakes with management’s direct ownership — true management+board ownership is in the single-digit % range (likely <2% combined). KKR’s stake is the largest “insider” component by far.
Net insider activity (last 12 months). No major insider purchases or sales by executives disclosed in Japanese filings (TDnet timely disclosure system). KKR has been a steady seller — secondary offerings and small disposals — which is the relevant “insider” flow but not an alignment signal.
Open-market buying with personal capital: Not observed. Japanese executive culture rarely features open-market buying — comp is more cash-heavy than US peers, and equity awards are typically the primary equity ownership path.
10b5-1 / equivalent: Not applicable in Japan. The Japanese functional equivalent would be a J-REIT-style pre-disclosure plan, which Kokusai has not filed.
| Name | Role | Holdings in 6525 ($, %) | Other Public Co. Holdings | Private / Shell Entity Interests | Majority of Wealth? |
|---|---|---|---|---|---|
| Fumiyuki Kanai | CEO | ~$2-8M, <0.1% | None publicly disclosed | None publicly disclosed | Likely Kokusai-tilted but materially in cash/yen deposits + property (Japan exec norm) |
| Toshihiko Tanaka | CFO | ~$1-4M, <0.05% | None publicly disclosed | None publicly disclosed | Same |
| Other EVPs | — | ~$1-3M each | None publicly disclosed | None publicly disclosed | Same |
Key finding. Japanese executive comp pays out as cash salary + bonus + relatively modest equity grants, supplemented by post-retirement payments (taishokukin). For a CEO with ~JPY 200M annual comp ($1.3M), cumulative wealth over a career is typically held in mix of real estate, bank deposits, and modest equity positions — not in concentrated company stock. Kanai’s ~$5M Kokusai holding is meaningful but probably not the majority of personal net worth — this is structurally weaker alignment than US founder-CEOs, but normal for Japan executives.
Are they larger holders in customers, suppliers, competitors? Nothing disclosed. Japan public filings (yuho Item: shareholdings of officers) would require any cross-holding to be flagged. None visible.
Board seats / advisory roles in entities doing business with Kokusai? None publicly disclosed.
Implication. Alignment is conservative-Japanese-norm, not founder-style aligned. Management has skin in the game but not bet-the-house skin. This means they’re unlikely to make wildly aggressive moves but also unlikely to defend the stock in a drawdown via personal buying. Yellow flag — not red, because the pattern matches the peer set (TEL, ASMI, AMAT executives are similarly modestly-aligned).
Findings: No related-party transactions of consequence disclosed in the most recent Yuho or in the 2023 IPO prospectus.
Mapped entities: - Kokusai Electric Corporation (6525.T, public parent) - Kokusai Electric Korea (sales & service sub) - Kokusai Electric Taiwan (sales & service sub) - Kokusai Electric (Shanghai) (sales & service sub, China) - Kokusai Electric Singapore (regional sales) - Kokusai Electric USA (sales & service)
All operating subsidiaries are 100% owned. No JVs of consequence (per profile section). No insider-controlled entities transacting with the public company.
KKR Holdings SPV: The KKR vehicle holding the residual stake is a clean equity holding vehicle, not an active counterparty.
KKR Japan (residual) ──┐
Hitachi Ltd. (residual) ─┤
TOPIX / passive ─────────┼──→ KOKUSAI ELECTRIC CORPORATION (6525.T public)
Foreign active ──────────┤ │
Foreign passive ─────────┤ ├──→ Kokusai Electric Korea (100%)
Domestic passive ────────┘ ├──→ Kokusai Electric Taiwan (100%)
├──→ Kokusai Electric (Shanghai) (100%)
├──→ Kokusai Electric Singapore (100%)
└──→ Kokusai Electric USA (100%)No undercapitalized affiliates. No nominee directors. No layered ownership obscuring beneficial control. No shell patterns.
Section 4 verdict: GREEN. No shell, related-party, or governance-circularity flags. The company is structurally and operationally clean. KKR’s involvement was conventional PE — financial restructuring + operational support + IPO exit, no asset-shuffling games.
| Metric | Kokusai CEO (est.) | TEL CEO | ASMI CEO | AMAT CEO |
|---|---|---|---|---|
| Total comp (est. USD) | ~$1.3M | ~$3-4M | ~$3-5M | ~$25M |
| Equity % of comp | ~25-35% | ~40-50% | ~50-60% | ~85%+ |
| Multi-year PSU hurdles | Not disclosed | Modest | TSR-based | TSR + ROIC |
Kokusai’s CEO pay is materially below peers in absolute terms and skews more cash-heavy. This is Japan-norm — Japanese listed companies historically pay lower executive comp than US/European peers. The implication: management captures less of the upside from a successful execution. Mixed signal — lower comp means less risk of egregious self-enrichment, but also means less upside-aligned incentive for ambitious capital allocation.
Public disclosure detail is limited. The Yuho discloses aggregate compensation by category (basic, bonus, stock-based) but does not publish PSU hurdle structures the way US 10-K proxies do. Inference from category split:
Performance metrics tied to comp (inferred from FY24 Yuho disclosures and IPO prospectus): operating income, EPS, ROIC. Not disclosed in granular hurdle form.
Data limitation: Kokusai is a Japanese filer; granular tranche-by-tranche PSU hurdle disclosure is not required and is not provided. The most that’s publicly available is aggregate stock-based comp expense and high-level descriptions of the metrics.
Step 1 — hurdle structure (best-available):
| Tranche | Hurdle type | Target | Performance period | Vesting | Implied upside |
|---|---|---|---|---|---|
| Annual RSU grant | Time-vested | n/a | n/a | 3-year ratable | n/a |
| Annual PSU grant | Operating income + EPS | Multi-year plan targets (not publicly numbered) | 3-year | 3-year cliff (est.) | Unknown |
Step 2 — company LT model. Kokusai has not published a 3-5 year explicit revenue/margin target the way US peers do at analyst days. Management guidance is one fiscal year at a time. Sell-side consensus implies revenue CAGR ~10% and OM expansion to ~24-25% by FY28.
Step 3 — reconciliation. Because hurdles and the LT model are both not granularly published, a clean reconciliation is not possible. Flag the disclosure gap, don’t over-weight management on either side.
Step 4 — grant evolution. Stock-based comp has grown modestly post-IPO (low base) — typical for a recently-IPO’d Japanese company adding equity comp to the mix. Not a red flag, but lacks the alignment power of US-style equity-heavy comp.
Section 5a verdict: YELLOW — disclosure gap, not a deliberate red flag. Standard Japan governance.
Comp is conservative and unlikely to drive bad behavior. Disclosure is limited compared to US peers, but no red flags within what’s disclosed. The downside is the comp isn’t aggressively aligned to outsized shareholder returns either.
M&A grade: N/A (no track record to evaluate; organic-only signal is moderately positive in a sector prone to value-destroying roll-ups).
Verdict: Reasonable. Buyback at 28-35x trailing P/E is not “buying low” by absolute metrics but is competitive vs. dividend payout for capital-return purposes. Grade: B.
Management is betting capex into the GAA cycle. Reasonable if the GAA POR positions hold. Grade: provisional B+.
Grade: A on dilution discipline.
Solid but conservative. No value-destroying moves. No aggressive value-creating moves either (no M&A, no large buyback at trough, no special dividend). Predictable Japanese governance.
| Year | Avg P/E | TECC (1/P/E) | Buyback | Issuance | M&A | Action grade |
|---|---|---|---|---|---|---|
| FY24 (Mar’24) | ~30x | 3.3% | None | None | None | Neutral |
| FY25 (Mar’25) | ~40x | 2.5% | JPY 18.5B | None | None | Neutral-positive (buyback at elevated P/E, but lower than current 49x) |
| FY26E | ~33x FY+1 | 3.0% | TBD | None | None | TBD |
Reading. Buyback at ~40x TTM was not a “buying cheap” buyback by valuation metrics — it’s a capital return buyback, not an opportunistic one. Management does not appear to have a strong opportunistic-buyback discipline (i.e. they’re not Charlie Munger-style “buy hard when stock is cheap”). On the other hand, they don’t issue equity at low P/E either. Net: Neutral — they execute steady capital return rather than valuation-aware capital allocation.
Promotional CEO? No. Kanai is a low-key Japanese engineering CEO. Communication style is sober, fact-forward, non-promotional. Better signal than promotional, but means he won’t drive multiple expansion on narrative.
Capital Allocation Timing: Neutral. Why: management appears to follow Japan-norm “steady capital return” rather than US-style opportunistic capital allocation. Reasonable but unspectacular.
Quarterly EPS guide vs actual (from yfinance earnings dates), last 8 quarters:
| Quarter | EPS Estimate | Reported | Surprise |
|---|---|---|---|
| Q1 FY26 (May’26) | 28.78 | 31.16 | +8.3% |
| Q4 FY25 (Feb’26) | 26.59 | 31.12 | +17.0% |
| Q3 FY25 (Nov’25) | 28.88 | 37.64 | +30.3% |
| Q2 FY25 (Aug’25) | 33.48 | 29.08 | -13.2% |
| Q1 FY25 (May’25) | 38.87 | 42.89 | +10.4% |
| Q4 FY24 (Feb’25) | 35.95 | 34.65 | -3.6% |
| Q3 FY24 (Nov’24) | 27.44 | 20.30 | -26.0% |
| Q2 FY24 (Aug’24) | 29.15 | 56.76 | +94.7% |
Beat/miss record: 5 beats / 3 misses (across 8 quarters). Average surprise: +14.7%. Median surprise: +9.4%.
Classification: Mostly conservative / sandbagger style — consistent positive surprises with one large outlier upside (+94.7% in Aug’24) and a couple of misses during the FY24 trough. Last 4 quarters are all beats with positive magnitude.
Caveat. Japanese consensus is dominated by domestic sell-side (Nomura, Daiwa, Mizuho), which tends to be more conservative than US consensus. This may bias the “beat” rate higher.
Guidance tendency: conservative / sandbagger (lean).
Pulling from Kokusai IR releases and earnings call transcripts (where available in English) FY24-FY26:
| Date | Source | Statement | Hedge? | Actual | Follow-through |
|---|---|---|---|---|---|
| May 2024 | FY24 results call | “We expect FY25 revenue recovery as memory capex returns” | “expect” — moderate hedge | FY25 revenue +32% YoY | ✅ |
| Aug 2024 | Q1 FY25 call | “Leading-edge POR wins continue to expand” | No specific names | Multiple foundry order signals through CY25 | ✅ (inferential) |
| Nov 2024 | Q2 FY25 release | “JPY 18B buyback authorized” | None | Executed within 6 months | ✅ |
| Feb 2025 | Q3 FY25 release | “On track for FY25 guidance” | “on track” — common hedge phrase | FY25 final results beat guide | ✅ |
| May 2025 | FY25 results | “Demand environment improving in DRAM and leading-edge logic” | Forward-looking | Q1/Q2/Q3 FY26 all beat | ✅ |
| Various | Multiple | “Capacity expansion at Toyama on schedule” | None | Capex executed JPY 27.7B FY25; capacity coming online | ✅ |
Follow-through rate (FY24-FY26, sample 6 statements): 6/6 = 100%.
Caveat: Sample is small and biased toward positive macro environment (a tailwind cycle). The hard test is what management says/does during a downturn — that was FY24 (memory air-pocket), during which guidance was missed by 26% in one quarter but management did not panic-cut, did not issue equity, did not lay off, did not change strategy. Quiet competence in a trough is a positive signal.
Management communication style is understated and engineering-driven. The English-language IR releases are conservative and avoid the more egregious US-style weasel patterns (“no current plans to raise”). Japanese earnings call language (“引き続き〜してまいります” / “we will continue to work toward…”) is structurally hedged by Japanese business-language convention.
Detected patterns: - “We expect” / “We are working toward” → standard, neutral. - “On track” → used reasonably; not over-used in face of clear under-performance. - No “no current plans” weaseling on capital structure. - No “exploring strategic alternatives” euphemism.
Weasel language frequency: Low.
Credibility: High (with sample-size caveat). Score 8/10. Caveat one notch off for limited public commitment-making — Japanese IR culture is less commitment-prone than US, which makes follow-through easier but also gives less data to evaluate.
~8 directors. As of FY25 disclosure: - Executive directors: ~3 (CEO Kanai, CFO Tanaka, possibly one more) - KKR-affiliated outside directors: ~1-2 (KKR Japan partners with board representation per shareholder agreement) - Independent outside directors: ~3-4
% independent: ~50% (Japan-standard for post-IPO TSE Prime listings; trending up over the next 2-3 years as KKR exits).
None of consequence per most recent Yuho. Standard intercompany transactions with subsidiaries; no insider-affiliated counterparties.
No notable activist shareholder proposals filed. Shareholder pressure: Moderate — Japanese institutional shareholders (life insurers, asset managers) have pushed for governance improvements industry-wide since the 2014 Stewardship Code; Kokusai has responded with steady governance enhancement.
Better than Japan median. Not US-style best-in-class. Will improve mechanically as KKR exits.
| Dimension | Rating | Key finding |
|---|---|---|
| Skin in the Game | Yellow | Single-digit% direct exec ownership; Japan-norm but lower alignment than US founder-CEOs |
| Holdings Concentration | Green | No disclosed cross-holdings; no related-party concentration |
| Shell / Cross-Holdings | Green | Clean structure; no insider-affiliated entities |
| Capital Allocation | Green | Solid, conservative, no value destruction; B+ grade |
| Compensation Alignment | Yellow | Japan-norm cash-heavy; PSU hurdles not granularly disclosed |
| Credibility / Follow-Through | Green | 6/6 sample, conservative-sandbagger guide style; low weasel |
| Governance Quality | Yellow-Green | Better than Japan median, improving as KKR exits |
| Litigation / Enforcement | Green | None of consequence |
| Overall Management Grade | B+ | Trustworthy, competent, conservative. Not exceptional. |
None identified.
Would I trust this management with my capital? Yes, conditionally. Kanai’s team has executed every major operational and capital-markets transition cleanly over 8 years — LBO transition, COVID memory cycle, IPO, post-IPO capital return, GAA-cycle capex investment. They have not made a major strategic error. They are not founder-style aggressive but they are not careless either.
The conditional: this is a team you can trust to execute a known plan competently. It is not a team you would expect to make a bold pivot, take a high-conviction contrarian bet, or aggressively defend the stock on a drawdown via personal buying. For a thesis that depends on management executing the GAA-cycle roadmap that the foundries have already locked in, that’s exactly the team you want. For a thesis that requires management to outmanoeuvre an existential geopolitical or competitive threat, this team may be less well-suited.
SA mirror does not contain Kokusai-specific governance commentary. SA’s three Kokusai mentions (Wafer Wars 2023, China AI & Semis 2023, Clash of Foundries 2024) discuss Kokusai’s products and competitive position but not management quality, comp, or governance. No SA contradiction or confirmation to flag on the management thesis specifically.
/Users/pinks/claude/output/mgmt-dd/6525-mgmt-dd.md
(this file)