They beat plan by ¥7.5B with zero FX help, then called their own Q3 margin guide an arbitrary placeholder from historical trends.
Thesis: DISCO's pricing model ties equipment ASPs directly to customer device value, so every HBM, 2.5D, and CPO generation they co-develop lifts their own pricing power structurally. The ¥7.5B beat was FX-neutral and driven by consumables operating leverage at 71.7% gross margin, while the Q3 guide sits on a conservative ¥151 FX assumption and a placeholder margin build with no negative developments behind it. The factory is full and they are hiring 200 production support staff by March, mirroring the FY20/21 ramp, which is capital conviction that orders exist beyond what is booked.
Verdict: LONG — Conviction: MEDIUM
Catalyst: Q3 results landing above the ¥150B shipment guide at current FX rates (~¥140), confirming the placeholder margin guide has upside. Also watch the Q4 HBM bounce and whether the 200-person production ramp translates to shipment acceleration in FY2028 H1.
Key Risk: HBM equipment demand turns down harder than the one-quarter lumpiness management describes, and with no Q4 visibility ('クリアにご説明できない'), the 40%→30% memory mix swing becomes structural rather than cyclical. Capacity ceiling means no offsetting volume from other lines if AI tool orders stall.
The Tell: CFO on Q3 gross margin guide: '全ての製品の案件とかを正規に積み上げて利益計画を作っているわけではございません...とりあえずの数値を仮置きしている'. They are telling you they haven't bottomed-up the margin guide. Combined with 'ネガティブなことが発生しているわけではなくて' (no negative developments), this is an admission the guide is soft by construction, not because anything is wrong.
Friction Level: MODERATE_FRICTION — Both sides agree on the FX-neutral beat, record margin, and capacity strain. Disagreement centers on whether the factory ceiling is a bullish demand signal (bottleneck leverage) or a bearish execution limit that caps upside and makes guidance supply-driven rather than demand-driven.
2027Q2 · Oct 22, 2026
4 of 5 readings agreed4 LONG · 1 HOLDRead the full call
Record 71.7% Margin, Guide Is Literally a Placeholder
DISCO's pricing model ties equipment ASPs directly to customer device value, so every HBM, 2.5D, and CPO generation they co-develop lifts their own pricing power structurally. The ¥7.5B beat was FX-neutral and driven by consumables operating leverage at 71.7% gross margin, while the Q3 guide sits on a conservative ¥151 FX assumption and a placeholder margin build with no negative developments behind it. The factory is full and they are hiring 200 production support staff by March, mirroring the FY20/21 ramp, which is capital conviction that orders exist beyond what is booked.
CFO on Q3 gross margin guide: '全ての製品の案件とかを正規に積み上げて利益計画を作っているわけではございません...とりあえずの数値を仮置きしている'. They are telling you they haven't bottomed-up the margin guide. Combined with 'ネガティブなことが発生しているわけではなくて' (no negative developments), this is an admission the guide is soft by construction, not because anything is wrong.
¥7.5B above plan with zero FX contribution. Record 71.7% GM vs 70%+ modeled. Q3 guided at ¥151 FX (costing ~1pt margin) with CFO explicitly calling the guide a placeholder with 'no negative developments.' Y/Y Q3 shipments +23%. Management sandbagging with conservative FX and trend-based margin.
Gross margin printed 71.7%, a record, driven by consumables volume pulling fixed costs down the curve. CFO explained the mechanism directly: higher production volume on consumables reduces fixed cost burden per unit. OP margin 47.4% vs 42.9% in Q1 with no FX help.
Adding 100 production support staff on top of ~100 already deployed, targeting 200 total by end of March. Management explicitly compared this to the FY20/21 capacity ramp when blade dicer volumes surged. Hiring into a factory already at full load signals orders they haven't yet booked.
CFO stated equipment ASPs are linked to customer device value: 'お客様自身が付加価値の高いデバイスを手掛けられるかどうか...お客様が装置に支払ってくださるコストも変わってくる'. Every HBM, 2.5D, CPO generation co-developed with customers lifts their own pricing power. This is structural, not cyclical.
Factory at full load, grinders hardest to build, blade dicer demand surging. Hiring operators not engineers. When guidance is set by how fast support staff can be deployed to Hiroshima, the number is supply-constrained. Q4 HBM forecast explicitly refused: 'クリアにご説明できない'.
Q3 GM guide is not built bottom-up. CFO: '全ての製品の案件とかを正規に積み上げて利益計画を作っているわけではございません...とりあえずの数値を仮置きしている'. Also ¥30B of R&D tool revenue from Q1 sits under revenue recognition review with only ~half recovered into Q2.
CPO scale cannot be quantified: 'このICの内数に入っている部分に関しては定量的な解説が難しい'. Panel-level packaging and glass substrates are evaluation stage only: '本格量産として大きく出てきているかというと、ちょっとまだそこまで見えていない'. Every AI narrative option is unquantified or pre-revenue.
-20
CAUTIOUS"CFO used '仮置き' (placeholder) and 'とりあえずの数値' (just a number for now) for the GM guide, deferred Q4 HBM visibility with 'クリアにご説明できない', and conceded CPO cannot be quantified '社内でも区分が難しい'. Specific on backward numbers, hedged on forward."
Full transcript with specific numbers on beat, margins, FX assumption, and hiring plan. Critical gaps: no Q4 HBM forecast, no CPO/advanced packaging quantification, ¥30B revenue recognition review unresolved, no inventory or cash flow data disclosed.
Q3 results landing above the ¥150B shipment guide at current FX rates (~¥140), confirming the placeholder margin guide has upside. Also watch the Q4 HBM bounce and whether the 200-person production ramp translates to shipment acceleration in FY2028 H1.
HBM equipment demand turns down harder than the one-quarter lumpiness management describes, and with no Q4 visibility ('クリアにご説明できない'), the 40%→30% memory mix swing becomes structural rather than cyclical. Capacity ceiling means no offsetting volume from other lines if AI tool orders stall.
The verdict is public. The case behind it is Pro: the forensic reading of guidance and language, the bull and bear cases, the red flags, and the play. It opens to everyone on Jan 20, 2027.