EQUITY RESEARCH  ·  JAPAN · SEMICONDUCTOR EQUIPMENT

DISCO CORPORATION (TSE: 6146)

The Back-End Chokepoint of the AI Era — Precision Cutting, Grinding & Polishing with a Global Share of 70–80%. Every chip on earth is diced and thinned on a DISCO machine before packaging.
Price ¥65,470 (Aug 14, 2026)
Market Cap ¥7.10T
PER (TTM) 52.6×
PBR 12.2×
Div. Yield 0.77%
52wk ¥37,260–91,680
FY2026 OM 42.3%
ROE 25.1%
For information purposes only — not investment advice. Figures from FY2022–FY2026 annual securities reports (EDINET, via ima KB) and timely disclosures; market data as of Aug 14, 2026. Report date: Aug 18, 2026.

1Executive Summary

Positioning · business nature · end-market landscape (from Segment Deep Dive D1)

DISCO is a near-monopoly supplier of precision processing equipment and consumables for the semiconductor back-end — dicing (cutting wafers into chips), grinding (thinning wafers to 30–100µm) and polishing. It is not a pure capital-equipment cyclical: ~30–35%+ of revenue is recurring (consumables + service), a "razor-and-blade" annuity on a global installed base. The business nature is capital equipment (precision machines) + high-margin consumables (diamond blades & wheels) + application know-how, sold into every device maker, OSAT and foundry on earth (TSMC, Samsung, Intel, NVIDIA…).

FY2026 (ended Mar-2026): revenue ¥436.9bn (+11.1%), operating profit ¥185.0bn (+10.9%, 42.3% margin), net profit ¥135.5bn, ROE 25.1%, net cash ¥184.6bn — shipments and sales a 6th consecutive record. Q1 FY2027 (Apr–Jun) beat guidance by a wide margin (+27.1% sales YoY, +42.2% OP YoY), and H1 FY2027 is guided to ¥242.8bn sales / ¥104.9bn OP (+24.8% / +33.0%).

FY2026 Revenue
¥436.9bn
▲ 11.1% YoY
Operating Profit
¥185.0bn
▲ 10.9% YoY
Op. Margin
42.3%
GM 70.2%
Net Profit
¥135.5bn
▲ 9.4% YoY
Q1 FY2027 OP
¥49.0bn
▲ 42.2% YoY
Verdict (summary): Accumulate / Moderate Buy — exceptional moat and AI-cycle earnings momentum, but the ¥7.1T valuation (PER ~52×, PBR 12×) already discounts much of it. We would add on pullbacks, with the key risks being an AI/HBM capex slowdown and 2028 depreciation ramp.

Recent Developments Timeline

DateEventSignificance
Apr 18, 2025Gohara Plant (Kure) construction decision announced — ¥33bn phase 1, precision-tool productionCapacity doubling for the AI cycle; BCM upgrade (tsunami risk removal)
Apr 22, 2026FY2027 guidance policy shifted to one-quarter-ahead disclosure; Q1 FY2027 forecast: sales ¥106.1bn / OP ¥42.0bnDemand-volatility-driven transparency change; quarterly cadence becomes the market rhythm
Feb 1, 2026Gohara phase-1 construction start; FY2027 capex plan ~¥33bn, R&D ~¥36bnRecord investment intensity
Jun 16, 2026FY2026 annual securities report filed — 6th consecutive record sales/profit; TSMC 11% disclosureBaseline for this report
Jul 21, 2026Q1 FY2027 results: sales ¥114.3bn (+27.1%), OP ¥49.0bn (+42.2%); H1 guide ¥242.8bn / ¥104.9bn; interim DPS ¥171Guidance beat; H1 shipments ¥276.9bn record
Jul 23–24, 2026Shares fell ~14% on "shipment miss" headline despite profit beat, then recoveredVolatility proof: headline-sensitive tape at high valuations

Company at a Glance

AttributeDetailAttributeDetail
Founded / Listed1937 (Kure, Hiroshima) / TSE Prime since 1999Fiscal yearApril–March (3月期)
BusinessBack-end precision processing: dicing saws, laser saws, grinders, polishers + diamond consumables + serviceGlobal shareDicing ~70–80%, grinding ~65–75%, polishing ~60–70%
Employees7,379 (consolidated, Mar-2026)PlantsHiroshima (Kure, Kuwabata, Gohara*), Nagano (Chino) — *under construction
ManagementPresident & CEO: Kazuma Sekiya (founder family, 1.94% stake)SegmentSingle segment — Precision Processing Systems
Net cash¥184.6bn (no interest-bearing debt)Dividend policy25% of half-year NP + ¥20 floor + ⅓ excess cash

2Investment Thesis

Three theses built from segment-level evidence (Segment Deep Dive D1–D7)

Thesis 1 — Irreplaceable step, near-monopoly pricing power

Every semiconductor die must be cut from its wafer and ground to final thickness before packaging. DISCO holds an estimated 70–80% share of dicing saws, 65–75% of grinders, 60–70% of polishers (CR3≈95% in dicing: DISCO, Tokyo Seimitsu, ADT). Its 20µm kerf blades (competitors 30–35µm+) and 5µm-grade thinning drive yield economics that fabs cannot replicate elsewhere. The result: gross margin ~70% — above ASML (~50%) and Tokyo Electron (~45%) — and 11 straight years of margin expansion (GM 54%→70%, OM 21%→42%).

Thesis 2 — Razor-and-blade recurring annuity cushions the cycle

Equipment sales (~63% of revenue) open the account; consumables (blades, grinding wheels — ¥95.0bn FY2026) plus service/training/recycling (~15.6%) lock it in. A 300mm fab burns thousands of blades a year at hundreds of dollars each. Switching costs (recipe + application engineers + re-qualification risk) are enormous, so customers cut new-equipment orders in a downturn but keep buying consumables while fabs run — the structural floor under DISCO's profit.

Thesis 3 — Direct, high-beta exposure to the AI/HBM super-cycle

HBM stacks require wafers thinned to 30–50µm; hybrid bonding demands another order of magnitude of TTV control. DISCO is the "chokepoint" for this pre-bond step. Evidence: FY2025 +27.9% revenue surge, FY2026 +11.1% (6th record), Q1 FY2027 sales +27.1% / OP +42.2% with shipments at a record ¥132.0bn; H1 FY2027 shipments guided at ¥276.9bn; a ¥80bn 10-year plan to triple tooling capacity (Gohara plant, phases through 2028+).

Key caveat: the thesis is increasingly single-cycle (AI/HBM + TSMC/Taiwan). Management itself notes general-memory capex has not yet recovered — if AI capex cools, there is no visible second engine today.

3Financial Trajectory

Consolidated P&L FY2022–FY2026 (+ FY2027 H1 forecast) · ¥ million · JGAAP

Item (¥M)FY2022FY2023FY2024FY2025FY2026FY27 H1 Fct
Net Sales253,781284,135307,554393,313436,889242,800
YoY+38.8%+12.0%+8.2%+27.9%+11.1%+24.8%
Gross Profit154,011184,506208,642277,570306,477
Gross Margin60.7%64.9%67.8%70.6%70.2%
Operating Profit91,513110,413121,490166,834184,989104,900
Op. Margin36.1%38.9%39.5%42.4%42.3%43.2%
Ordinary Profit92,449112,338122,393168,943184,936104,800
Net Profit (parent)66,20682,89184,205123,891135,52173,800
Net Margin26.1%29.2%27.4%31.5%31.0%30.4%
EPS (¥)611.67765.47777.291,143.261,249.84680.39
DPS (¥, split-adj.)269305307413505171 (int.)
Payout (DPS/EPS)44.0%39.8%39.5%36.1%40.4%

FY2022/23 DPS adjusted for the Apr-2023 3-for-1 split (actual paid ¥808/¥916). FY2027 = quarterly guidance only (no full-year disclosure). Revenue CAGR FY2022–26: 14.5%; OP CAGR: 19.2%. Payout shown on a reproducible DPS/EPS basis (split-adjusted).

Q1 FY2027 — the guidance-beating quarter (Apr–Jun 2026)

Item (¥M)Q1 FY27 ActualQ1 FY27 ForecastBeatQ1 FY26 ActualYoY
Net Sales114,308106,100+7.7%89,914+27.1%
Operating Profit49,03342,000+16.7%34,480+42.2%
Ordinary Profit48,44142,300+14.5%33,999+42.5%
Net Profit (parent)34,22129,500+16.0%23,767+44.0%
EPS (¥)315.51271.98219.23

Management: "machine acceptance (検収) progressed faster than assumed". Shipments hit a record ¥132.0bn in Q1 — a 3rd consecutive record quarter. H1 FY2027: sales ¥242.8bn (+24.8%), OP ¥104.9bn (+33.0%), shipments ¥276.9bn; interim DPS ¥171 (+32.6%).

Revenue (¥bn, bars) vs Operating Margin (%, line) 0%50%OM (right) 0220440 254 284 308 393 437 FY22 FY23 FY24 FY25 FY26 36.1% 38.9% 39.5% 42.4% 42.3%
Revenue grew 14.6% CAGR with operating leverage — OP CAGR 19.2%; margin plateaued at ~42% in FY2026 (mix + FX effects fading).
Shipment value (出荷額, ¥bn) — DISCO's leading indicator 0 225 450 264 292 317 402 443 FY22 FY23 FY24 FY25 FY26
Shipments lead recognized revenue by the acceptance (検収) cycle — 6 consecutive record years; Q1 FY2027 ¥132.0bn (3rd straight record), H1 FY2027 guided ¥276.9bn. Shipments are the single best leading indicator for DISCO.
Driver decomposition (FY2025→FY2026, Δsales +¥43.6bn): Volume — shipments +10.3% (¥401.6bn→¥442.8bn) is the dominant driver; Mix — product/application mix nudged GM down 0.4pp; Cost — COGS ratio 29.4%→29.8%, absorbed; Price — favorable FX (yen weakness) contribution faded after FY2025. OP elasticity > revenue elasticity: +67.5% OP vs +53.8% sales over FY2023–26.

4Business Deep Dive

Seven dimensions, directly expanded from Segment Deep Dive D1–D7

① Product Portfolio D1 · Business Essence

DISCO's domain is "Kiru (cut) · Kezuru (grind) · Migaku (polish)". Core products:

Product GroupKey ProductsTechnology / SpecRole
Precision Equipment (62.7%)Dicing saws (DFD auto / DAD semi-auto), laser saws (incl. Stealth Dicing), grinders, polishers, wafer mounters20µm kerf blades @ 40,000rpm; 5µm-grade thinning; 300mm TTV ±1.5µmBack-end cutting & thinning chokepoint
Precision Tools (21.7%)Diamond dicing blades, grinding wheels, dry-polish wheelsProprietary bond + diamond-grit recipes; $100s per bladeRecurring consumables (razor-blade)
Other (15.6%)Maintenance, training, disassembly/recycling, leasing & used-equipment, paid processing57+ global service pointsService annuity + installed-base support
② Market Position & Moat D2
CategoryDISCO Share#2 & others
Dicing saws~70–80%Tokyo Seimitsu ~25%; ADT (CR3 ≈95%)
Grinders (thinning)~65–75%Tokyo Seimitsu, G&N (CR3 ≈85%)
Polishers~60–70%Niche Japanese players

Moat sources: ① decades of process know-how + free "test-cut" application support; ② consumable recipe lock-in (swap the blade brand and yield collapses); ③ enormous customer switching cost (re-qualification, yield risk worth hundreds of millions of dollars); ④ global application-engineer network. In AI-advanced/HBM processes DISCO is effectively a monopoly — even TSMC, Samsung and SK Hynix have no practical second source for the most demanding steps.

③ Profit-Driver Decomposition D3 · the key dimension
DriverFY2025→FY2026Reading
VolumeShipments +10.3% (¥401.6→442.8bn)Dominant — AI/HBM-driven machine units
Price / MixGM 70.6%→70.2% (−0.4pp)Mix shift; FY2025 FX tailwind faded
CostCOGS ratio 29.4%→29.8%Slightly up; raw-material exposure small
SG&AR&D ¥31.7bn→¥34.2bn; headcount +5.5%Absorbed by operating leverage

Sustainability: margin is supported by monopoly pricing power, not cyclical operating leverage → a downturn would compress volume but the 70% gross margin and consumable floor should protect profitability far better than a typical equipment maker. Upside to the 42% OM plateau is now limited near-term (mix drag), with 2028 depreciation (Gohara) as a headwind.

④ Demand Structure / Downstream D4

End-market mix:AI data-center / AI infrastructure + advanced logic + HBM (thinning to 30–50µm, hybrid-bonding pre-steps) — the current growth engine; ② power semiconductors (EV/SiC/GaN — SiC hardness near diamond, DISCO is the de-facto choice); ③ legacy logic / general memory — China legacy fab build-out supports volume, but general-memory capex has not fully recovered (management's own words).

Region (¥M)FY2025FY2026YoY
Japan41,04345,606+11.1%
China125,375134,975+7.7%
Korea42,34133,069−21.9%
Taiwan74,404117,378+57.8%
Other Asia33,88844,115+30.2%
Americas48,38334,903−27.9%
Europe27,87626,839−3.7%
Total393,313436,889+11.1%

Cycle profile: highly cyclical (customer capex) but with 1–2 quarters of shipment→revenue visibility and a consumable cushion; geography is concentrating — Taiwan+China = 57.8% of FY2026 revenue.

⑤ Strategic Shift / Capacity / Mid-term Plan D5

Gohara Plant (new, Kure, Hiroshima): phase-1 building ¥33.0bn (2026/2–2028/4), 133,570 m², seismically isolated; consolidates Kure + Kuwabata precision-tool lines (also removes tsunami risk = BCM). Three phases, land 218,539 m² (~2× Kuwabata). 10-year total investment ~¥80bn to triple precision-tool capacity. Haneda R&D Center rebuild in progress (the source of the FY2024 ¥7.5bn impairment).

  • FY2027 (FY ending Mar-2027) capex ~¥33.0bn; R&D ~¥36.0bn — record investment intensity
  • FY2027 guidance policy shifted to one-quarter-ahead disclosure only (demand volatility)
  • Watch item: 2028 plant ramp → depreciation drag on margins, partially offset by capacity for the next AI cycle
⑥ Competition & Risk D6
CompetitorThreat LevelWhere
Tokyo Seimitsu (Accretech)MediumDicing/grinding #2 — price/modular; customers nurture 2nd source
Plasma-Therm / PanasonicLow-MedPlasma dicing on ultra-thin wafers — technology substitution risk
Hamamatsu (laser) / niche laserLowStealth dicing in targeted applications
Chinese: JSG, 光力科技, 迈为, 华海清科Low (rising)Low-end dicing/thinning; localization push; local-procurement price pressure

Key risks quantified: TSMC = 11.0% of FY2026 revenue (first-ever >10% disclosure); HBM trio (TSMC/Samsung/SK Hynix) accounts for a large share of revenue; Taiwan 26.9% + China 30.9% = 57.8% geographic concentration. China self-sufficiency policy is the long-dated bear case (per Morningstar).

⑦ Governance & Capital-Return Signals D7

Will system: since 2011, DISCO runs an internal-currency, manager-less market economy (task bidding, existence cost, ~40% of bonus linked to Will) — widely credited for the OM lift from ~16% to 42%. Unique culture, hard to replicate, a genuine soft moat.

How Will works (context): every task, desk, meeting room and even an umbrella stand is priced in Will; employees bid for work at the lowest price; the president can print Will for special projects (acting as "central bank"); a weekly "Coliseum" lets employees pitch improvement ideas and win execution budgets. The system removed traditional managers and made cross-team resource flow automatic — 772 chargeable behaviors and 337 rewardable behaviors govern it. Sony, Toyota and Panasonic have studied it; none have replicated it. This cultural engine is why DISCO sustains 40%+ operating margins without conventional bureaucracy.

IndicatorFY2026Comment
Dividend policy25% of half-year consolidated NP + ¥20 floor + ⅓ of excess cashPayout 40.4% (DPS/EPS); DPS ¥505 (+22.3%)
FY2027 interim DPS¥171+32.6% YoY — earnings-led
ROE / Equity ratio25.1% / 78.9%No interest-bearing debt
Net cash¥184.6bn cash~25% of total assets; no buybacks
Shareholder structureDispersed; CEO holds 1.94%No activist overhang

5Segment Structure

Product-group revenue mix FY2022–FY2026 (DISCO reports as a single operating segment)

Product Group (¥M)FY2022FY2023FY2024FY2025FY2026FY26 Share
Precision Equipment143,909180,741194,879249,494273,95762.7%
Precision Tools58,47562,33669,16585,90194,97621.7%
Other (services etc.)51,39641,05743,51057,91767,95515.6%
Total253,781284,135307,554393,313436,889100%

FY2022 as originally reported; FY2023 report reclassified FY2022 (equipment +¥10.7bn, tools +¥4.1bn, other −¥14.8bn).

Product-group revenue mix (¥bn, stacked) 0 220 440 254 284 308 393 437 FY22 FY23 FY24 FY25 FY26 Precision Equipment Precision Tools Other (services etc.)
Equipment dominates growth; tools + other (~37% of revenue) provide the recurring/service annuity. FY2025-26: equipment +9.8%, tools +10.6%, other +17.3%.
FY2026 geographic revenue (¥bn) Japan China Korea Taiwan Other Asia Americas Europe 45.6135.0 33.1117.4 44.134.9 26.8
China 30.9% + Taiwan 26.9% = 57.8% of revenue; Korea (memory) and Americas weakened in FY2026 — concentration toward AI-logic customers.

6Anomaly Analysis

From Anomalies Report — each with severity, cause and risk assessment

#AnomalyData PointCause (per annual reports)Severity
1FY2024 net-profit stall (+1.6%)NP ¥84.2bn vs +8.2% salesOne-off impairment ¥7.5bn — Haneda R&D Center rebuild (below OP line)Medium
2FY2025 growth spike (+27.9%)Sales ¥393.3bn; OP +37.3%AI/HBM demand + high-value mix + FX (yen weakness)Medium
3FY2026 investment CF doubles (+99.7%)Invest CF −¥135.8bn; FCF −¥2.2bn¥100bn time-deposit placement + Gohara plant capexLow-Mid
4TSMC single-customer 11.0%¥48.2bn (first >10% disclosure)AI advanced-logic capex concentrationMedium
5Dividend jump +22.3% vs EPS +9.3%DPS ¥413→¥505Policy: 25% of half-year NP + excess-cash ⅓Low
6Q1 FY2027 guidance beat (+7.7% sales / +16.7% OP)Sales ¥114.3bn vs ¥106.1bn guideFaster-than-expected machine acceptance (検収)Low (positive)
7Geographic shift (Korea −21.9%, Americas −27.9%)vs Taiwan +57.8%Memory/auto capex soft; AI logic (TSMC) boomingMedium
83:1 stock split (Apr 2023)EPS/DPS history distortionDisclosure caliber — model split-adjustedLow
Interpretation: the "false alarms" (FY2024 net-profit stall, FY2026 FCF dip) are one-off/accounting items — underlying operating trends stayed healthy. The real structural watch-items are: TSMC/Taiwan concentration (anomaly 4 + 7), the margin plateau (mix/FX), and the 2028 depreciation ramp. The positive surprise (anomaly 6) shows guidance conservatism — management discloses only one quarter ahead.

7Valuation

Market data Aug 14, 2026 · ¥65,470 · consensus from 21–23 analysts

PER (TTM)
52.6×
10-yr median ~22×
Forward PER
38.5×
vs FY2027E EPS ~¥1,700
PBR
12.2×
ROE 25%+ supports premium
EV/EBITDA
34.1×
Analyst target ¥84,580
ScenarioAssumptionFY2027E EPSMultipleFair Valuevs Current
ConservativeAI capex decelerates H2; OM ~40%¥1,45038×¥55,100−15.8%
BaseH1 run-rate holds; H2 moderates; OM ~42%¥1,55042×¥65,100≈ −0.6%
Target (bull)AI/HBM super-cycle extends; margin upside on complexity¥1,60048×¥76,800+17.3%
Analyst ConsensusBuy — avg 12-mo target¥84,580+29.2%

EPS scenarios are estimates based on H1 FY2027 guidance (¥680.39 H1 EPS), FY2026 H2 seasonality and margin assumptions. Analysts' target implies stronger H2 than our base case.

Peer Comparison (Aug 2026, Morningstar-normalized)

MetricDISCO (6146)Advantest (6857)ASMPT (0522.HK)BESI (BESI.AS)
PER (normalized)52.3×71.1×65.6×
PBR12.2×33.6×4.3×
P/Sales16.3×23.9×4.6×
Op. Margin (FY2026)42.3%~25%~15%~30%
BusinessBack-end cut/grind/polishTesters/handlersPackaging & bondingHybrid bonding

DISCO is not cheap vs peers, but its operating margin (42% vs mid-teens to low-30s for peers) is the structural differentiator that justifies the premium.

Valuation read: at ¥65,470 the stock trades at ~52× TTM earnings and ~12× book — rich even for a monopoly; the market is pricing continued AI-led growth. PER compression risk is real if quarterly momentum stalls (the stock already fell ~14% after the Q1 "shipment miss" headline on Jul 24 before recovering). We see fair value near current levels on base case, with upside only in the bull scenario — hence "Accumulate, not chase".

8Catalysts

9Risks

RiskDescriptionSensitivityLevel
AI/HBM capex slowdownRevenue is increasingly driven by one cycle; general memory not yet recoveredHigh beta: a 10% capex cut by the HBM trio could shave ~¥40–50bn sales (est.)High
Customer / region concentrationTSMC 11.0% single customer; Taiwan+China = 57.8% of salesTSMC capex adjustments flow straight through; Taiwan geopoliticsHigh
Valuation compressionPER ~52× vs 10-yr median 22×; rich expectationsAny momentum miss → 15–30% drawdown potentialMed
Technology substitutionPlasma dicing / stealth laser on ultra-thin wafers; new materialsLong-dated; DISCO defends by owning laser/plasma tooMed
China self-sufficiency / geopoliticsLocalization push (光力, 迈为, 华海清科) + US-China controls; China = 30.9% of salesLow-end share erosion 5–7pp possible in China over yearsMed
2028 capacity/depreciation rampGohara + Haneda rebuild raise depreciation just as AI-cycle could coolOM down 1–2pp potential in FY2028–29 (est.)Med

10Profit Quality

Quality metrics FY2022–FY2026 — earnings quality is exceptional (cash-backed, debt-free, recurring-heavy)

MetricFY2022FY2023FY2024FY2025FY2026
Gross Margin60.7%64.9%67.8%70.6%70.2%
OP / Ordinary ratio99.0%98.3%99.3%98.8%100.0%
Operating CF (¥bn)83.781.897.5120.4133.5
FCF (¥bn)40.168.781.152.4−2.2
ROE24.3%25.9%22.4%27.6%25.1%
Payout (DPS/EPS)44.0%39.8%39.5%36.1%40.4%
Net cash (¥bn)125.8163.1215.5229.2184.6

FY2026 FCF −¥2.2bn includes a ¥100bn time-deposit placement (capital allocation, not operating burn); ex-that, operating FCF is strongly positive.

Overall assessment: excellent. Operating profit ≈ ordinary profit (no financial leverage games); OCF covers profit 99%+; debt-free with ¥184.6bn net cash (~25% of assets); ROE sustained 22–28%; payout transparent and earnings-linked. The only quality blemish is the one-off FY2024 impairment and the FY2026 FCF dip — both explained and non-recurring in nature.

11Conclusion

ACCUMULATE
Verdict: Accumulate (Moderate Buy) · Horizon: 6–18 months · Conviction: Medium-High. A genuine near-monopoly at the chokepoint of the AI era with world-class margins (OM 42%), a recurring consumable annuity, and record momentum — but the ¥7.1T valuation leaves limited margin of safety. Add on pullbacks toward ¥55–60k; re-evaluate if AI/HBM capex signals turn.

Rating Grid

Moat
★★★★★
Growth
★★★★☆
Profit Quality
★★★★★
Valuation
★★☆☆☆
Balance Sheet
★★★★★

Recommended Actions

Scenarios to Revisit This View