Kokusai Electric
Sells high-throughput batch deposition systems (vertical furnaces for ALD, LPCVD, oxidation/diffusion) plus a growing recurring stream of parts, service and process support tied to the installed base (~40% of sales).
Earnings, margins, COGS & capex
A focused deposition specialist that punches above its ~$1.5B revenue by dominating batch thermal ALD/LPCVD for 3D NAND and DRAM. FY26 was a mild down year (-1.6% revenue, -17.6% adj. OP) as the memory-capex trough and a China local-market pullback offset a 27% jump in high-margin service. Balance sheet is now net cash; the company guides a sharp +19.1% rebound in FY27 (to JPY 280B) on DRAM/HBM and logic-foundry (GAA) demand. Profitability is solid (41% gross, 20% adj. OP margin, ~13% net) but cyclical.
Income statement — where each revenue dollar goes
% of revenueOf every $1 of revenue, ~59¢ is cost of goods and ~21¢ operating expense, leaving ~20¢ of operating profit (~13¢ net).
Revenue trend
Margins
down ~1.4pt YoY on mix/utilization; guided ~42% for FY27
compressed from prior year (adj. OP JPY 47.6B, -17.6%)
down with the cycle
healthy; capex fell ~20%
COGS structure
COGS is dominated by purchased subsystems and precision components — RF/power delivery, gas/flow control, quartz and heater assemblies, robotics/handlers — plus labor and materials for the furnace platforms. High-value process modules and the service/parts business carry richer margins than new-tool sales, which is why the ~40%-of-sales service mix lifts blended gross margin and cushions equipment-cycle troughs.
Capex
Capex JPY 16.9B in FY26 (~7% of sales), down ~20% YoY. FY27 capex guided up ~60% YoY, driven partly by a US demo/technology center (opening early 2027) to support North American customers and process-of-record wins.
Latest earnings
Full-year results came in above the company's own revised (Q2) forecast, but the stock sold off on the print (rich valuation, cautious near-term read and China local-market softness)
FY ending Mar 2027: +19.1% revenue growth to JPY 280B on DRAM/HBM and logic-foundry demand; net-cash position reached and a buyback of up to JPY 5.3B authorized (May 13 2026); capex up ~60% on the US demo center
- Service revenue mix
- ~40% of sales, +27% YoY
- Adjusted OP margin
- 20.2% (adj. OP JPY 47.6B)
- Free cash flow
- JPY 31.8B (FY26)
- Net cash
- JPY 6.6B (reached a year early)
Growth drivers
- AI-driven DRAM/HBM capacity expansion — FY27 DRAM revenue guided +36% and more deposition steps per wafer
- Gate-all-around (GAA) transistor transition at leading-edge logic/foundry — FY27 Logic/Foundry revenue guided +38%; new ALD PORs (~JPY 10B of GAA-related orders flagged on the call)
- 3D NAND vertical scaling (more layers = more high-aspect-ratio deposition passes), a structural tailwind for batch tools
- Service/parts/upgrade revenue compounding off a large installed base (now ~40% of sales, +27% YoY)
- Geographic diversification beyond China (FY27 non-China revenue guided +30%, China -2% as it normalizes)
Bull & bear
A pure-play on rising deposition intensity: every AI-memory (HBM/DRAM) and GAA-logic node adds deposition steps, and Kokusai owns the high-throughput batch niche with sticky PORs and a compounding service annuity — plus optionality as a long-rumored M&A target.
- AI/HBM-led DRAM capex (FY27 DRAM +36% guided) and 3D NAND layer scaling structurally raise deposition steps per wafer, directly expanding Kokusai's SAM
- GAA ramp brings new ALD process wins (~JPY 10B orders flagged; FY27 Logic/Foundry +38% guided) and pulls the name into leading-edge logic, not just memory
- Service/parts at ~40% of sales (+27% YoY) is a high-margin, less-cyclical annuity that de-risks the equipment cycle
- Net cash, strong FCF (~13.5% margin) and buybacks; management guides +19.1% FY27 revenue rebound
- Strategic scarcity — Applied Materials tried to buy it (2019, raised to $3.5B, terminated 2021 on China antitrust); a specialist deposition asset is a recurring takeover candidate
The stock has already priced the upcycle and then some — ~55x forward / ~88x trailing earnings, roughly double its larger, more diversified peers — on a small, memory-cyclical, China-exposed single-product franchise where consensus targets sit below the current price.
- Valuation is stretched: forward PE ~54.6x (trailing ~87.7x) vs AMAT ~20x, LRCX ~25x, TEL ~25x, ASMI ~30x; analyst consensus target (~JPY 8,700) implies ~18% downside
- FY26 actually declined (-1.6% revenue, -17.6% adj. OP) — the rally ran ahead of the fundamentals
- Memory concentration means a DRAM/NAND capex rollover hits revenue and multiple simultaneously
- China (historically a large share) is normalizing down (FY27 China revenue guided -2%) and structurally threatened by export controls plus Naura/Piotech local substitution
- Narrow platform vs diversified giants; less leading-edge single-wafer logic exposure caps the secular story
What it is worth
Peer-multiple comparison + reverse-earnings sanity check. Public on TSE Prime at JPY 2.63T (~$16.3B) market cap, ~87.7x trailing / ~54.6x forward earnings (stockanalysis.com, Jul 7 2026) vs a WFE peer group (AMAT ~20x, LRCX ~25x, TEL ~25x, ASMI ~30x forward).
Memory cycle rolls over or China weakness deepens; earnings dip and the multiple compresses toward peer-average (~25-30x forward), implying material downside from the JPY 10,625 spot.
Cyclical rebound (+19.1% FY27) materializes but the multiple normalizes toward the higher end of the peer band as growth is already priced; fair value roughly the consensus JPY 8,700-10,000 zone.
Memory/AI capex supercycle persists, GAA PORs convert, service annuity keeps compounding, and takeover optionality supports a premium — a >60x forward multiple can hold or the story grows into it; upside toward/above JPY 12,000+.
At ~55x forward the market is pricing a durable multi-year memory/AI (HBM-DRAM) supercycle plus a successful GAA-logic diversification and continued service compounding — roughly a 2x premium to larger, more diversified peers, on a smaller and more cyclical franchise. Analyst consensus is 'Buy' but with a target (~JPY 8,700) ~18% below the Jul 7 2026 spot of JPY 10,625, underscoring the valuation stretch. Not financial advice.
SWOT
Strengths
- Category leadership in batch/vertical-furnace ALD, LPCVD and oxidation/diffusion — high throughput per tool for memory
- Deep, sticky process-of-record relationships with the top memory makers; switching a qualified deposition step is costly and slow
- Recurring service/parts now ~40% of revenue (+27% YoY), smoothing the equipment cycle and lifting blended margins
- Clean, net-cash balance sheet (JPY 6.6B, reached a year early) funding R&D, a US demo center, and buybacks
Weaknesses
- Small and concentrated versus AMAT/LRCX/TEL — a single-product-family specialist, not a broad platform
- Heavy memory (DRAM/NAND) exposure makes revenue highly cyclical
- Historically large China revenue dependence, now exposed to export controls and local substitution
- Under-indexed to single-wafer leading-edge logic deposition where AMAT/ASMI/LRCX are stronger
Opportunities
- GAA transition and HBM/DRAM AI-capex supercycle multiply deposition intensity per wafer
- Grow North American / non-China footprint (US demo center) and diversify the customer base
- Expand single-wafer ALD and new-materials processes to widen the served market
- Continued service/upgrade attach on a growing installed base
Threats
- Memory-capex downturns and inventory cycles
- US/Japan export controls curbing China sales — Chinese localization (Naura, Piotech) displacing furnace demand domestically
- Larger rivals (TEL, ASM International, AMAT) pushing into batch/ALD niches
- Customer concentration — a capex pause at Samsung, SK Hynix or Micron moves the P&L materially
Moats, dependencies & bottlenecks
Moats
Once a deposition step is qualified into a customer's process flow, re-qualifying a competitor's tool is slow and risky — high switching cost.
Leading throughput and film quality in vertical-furnace ALD/LPCVD for memory; erosion risk from single-wafer alternatives and rivals.
Moderate-Strong ~40% of sales in parts/service/upgrades; grows with every tool shipped and is far less cyclical.
Sub-scale versus AMAT/LRCX/TEL; a specialist, not a full-flow platform — the anti-moat.
Dependencies
DRAM/NAND capital cycles drive the bulk of equipment revenue; concentrated buyer base.
Geographic/regulatory Historically a large revenue share; FY27 China guided -2% amid US/Japan controls and domestic substitution.
RF/power, gas-flow, quartz, robotics from firms like MKS, Advanced Energy, Entegris, Horiba — availability and cost feed COGS.
Technology roadmap Growth depends on winning new PORs each node; a missed qualification cedes a step to a rival.
Advantages
- Best-in-class batch thermal ALD/LPCVD throughput for memory
- Sticky, qualified process-of-record relationships
- High-margin, growing service/parts annuity (~40% of sales)
- Net-cash balance sheet and strong free cash flow
- Focused engineering culture and pure-play exposure to deposition-intensity growth
Weaknesses
- Sub-scale versus AMAT/LRCX/TEL/ASMI
- High cyclicality from memory concentration
- China revenue and regulatory exposure
- Thin diversification into leading-edge logic single-wafer deposition
- Rich valuation leaves little margin for execution or cycle misses
Bottlenecks
- Memory-capex cyclicality — revenue swings with DRAM/NAND investment
- China exposure squeezed between export controls and local competitors
- Customer concentration among a handful of memory/foundry buyers
- Narrow product platform; limited leading-edge single-wafer logic footprint
- Dependence on continuous POR wins to convert node transitions into revenue
Top signals & trends
Top signals
Direct read on Kokusai's largest demand pool; AI-memory buildout is the swing factor (FY27 DRAM +36% guided).
~JPY 10B GAA-related orders flagged; converting these validates the logic-diversification story (FY27 Logic/Foundry +38% guided).
FY27 China guided -2%; tighter controls or faster Naura/Piotech substitution pressures the franchise.
27% YoY growth to ~40% of sales de-risks the cycle and supports margins.
Forward ~54.6x and consensus target (~JPY 8,700) below spot signal a stretched multiple.
Trends
More deposition steps per wafer and a multi-year memory-capex upcycle.
New ALD steps open a logic market beyond Kokusai's memory core.
Higher layer counts multiply high-aspect-ratio deposition passes — structurally favors batch tools.
Near-term China revenue risk and a rising domestic-equipment competitive threat.
Recurring parts/service/upgrade revenue becomes a larger, steadier profit pillar.
Ecosystem & competitor graph
Suppliers feed the company; customers pull from it. Line thickness shows the strength of each tie (supply-chain dependency, customer earnings contribution). Hover to isolate a tie.
Vacuum, RF/power, and gas-delivery subsystems.
Precision power and RF for process modules.
Precursor materials, filtration, fluid-handling and specialty materials.
Gas and fluid-delivery subsystems for deposition tools.
Mass-flow controllers and process metrology components.
Top-tier DRAM/NAND/foundry buyer of deposition tools.
Leading DRAM/HBM maker; core memory customer.
US DRAM/NAND maker; key non-Asia customer.
NAND flash maker (ex-Toshiba Memory), important batch-furnace user.
Leading-edge foundry; GAA/logic deposition opportunity.
Logic/foundry customer for advanced-node deposition.
Much larger Japanese WFE leader; competes in batch/thermal and coater/developer and dwarfs Kokusai in scale.
The ALD pure-play leader, strongest in single-wafer ALD — the most direct deposition-technology rival.
Broadest deposition/CVD/PVD/epi platform; tried to acquire Kokusai in 2019 (terminated 2021 amid China antitrust review) and remains a large shareholder.
Deposition + etch giant, heavily memory-levered; overlaps in key film steps.
Korean furnace/ALD/CVD supplier close to Samsung/SK Hynix; regional competitor.
China's largest domestic WFE maker (context only, not a buy/own call) — rising furnace/deposition substitute inside China.