At 49× earnings, the multiple decides on one variable — is 42% operating margin the new floor or the cyclical peak?

• The premium. At ¥63,500 the market pays 49× trailing P/E and 35× EV/EBITDA — above ASML (25×), KLAC (23×), AMAT (20×), and LRCX (19×) on EV/EBITDA despite Disco being one-tenth their scale. No margin of safety on multiple, yield (0.8%), or earnings base.
• Case for the floor. Gross margin compounded from 47% to 70% across two full cycles. FY26 operating margin of 42.3% sits 12 points above the FY18 prior peak of 30.5%. Q1 FY27 guide of 39.6% is the seasonal trough — already 13 points above the FY19 floor.
• Case against. AI/HBM mix is a single-application bulge. The published bear DCF anchors at ¥34,244 — roughly half spot — assuming reversion to a ~30% mid-cycle mean. The Q4 FY26 44.2% peak is fading into a 39.6% Q1 guide; mean reversion does not require new news.

Every cycle since FY02 has lifted the trough margin — the market is pricing cycles that no longer exist.

• The floor keeps rising. Trough operating margin: −7.7% (FY02) → +0.1% (FY09 on a 42% revenue collapse) → +26.2% (FY19) → 39.5% mid-cycle FY24. The lift has tracked consumables share rising from 13% (FY01) to 22% (FY26), now plus another 10% in parts and service.
• Razor-blade base, doubled. Consumables ¥96B in FY26 — roughly 2× the FY19 absolute level on a revenue base 3× larger. Management broke ground February 2026 on Gohara Phase 1, ¥33B for consumables capacity only. The largest single capital commitment in company history is for the annuity, not equipment.
• Sell-side moved; the tape hasn't. Six broker PT raises Jan–Apr 2026 lifted consensus to ¥77,150 against a ¥63,500 tape — a 21-point gap. The tape is down 22% from the late-February ¥81,000 peak across five months of clean beats. The convergence event is Q1 FY27 on July 23.

Higher margins than ASML and KLAC on a fortress balance sheet — and a 20-point operating-margin gap to Accretech on identical inputs.

Disco prints the highest operating margin in the semicap peer set (42.3% vs ASML 37%, KLAC 39%, AMAT 29%) on a balance sheet 79% equity-funded with no debt. Accretech — the only direct dicing/grinding rival, same Japan cost base, same customer roster — earns ~22% SPE-segment operating margin on the same inputs. The 20-point gap is the cleanest controlled experiment for pricing power in semicap. ROE of 25% on a clean balance sheet would be 35–40% under US-peer leverage; the conservatism is a choice, not a constraint.

Three structural threats sit under the floor — one visible, one invisible, one fresh.

• Hybrid bonding (visible). AMAT took a 9% stake in Besi in April 2025; Samsung committed hybrid bonding for HBM4 in May 2025. If hybrid bonding eliminates one thinning step per HBM stack, Disco's wallet per AI chip declines mechanically — but pre-bond grinding under 30µm still belongs to Disco, and HBM bit-growth at 70–100% YoY swamps any 10–20% step compression for now.
• Chinese consumables (invisible). China is roughly 20–30% of revenue, undisclosed inside Asia ex-Japan at 75% of FY26 sales. Domestic equipment share hit 35% in 2025; etch and deposition substitution is already past 40%. A NAURA or AMEC precision dicer qualifying at a Chinese leading-edge fab — or open-architecture blades on trailing-node fabs — erodes the razor-blade base under the floor.
• Receivables drift (fresh). DSO extended from 40 to 48 days in FY26; AR grew 33% on 11% revenue growth; the allowance for doubtful accounts grew 5.8× to ¥817M. First material yellow flag across 13 forensic categories in 26 years. The auditor downgrade trigger is DSO above 55 days.