SCREEN Holdings

Phase A — Understand the business

Company Overview

SCREEN Holdings is a 157-year-old Kyoto firm (founded 1868 as Ishida Kyokuho-do; formerly Dainippon Screen Manufacturing; adopted the holding-company structure and the SCREEN Holdings name in 2014) that has quietly become the world's #1 maker of single-wafer semiconductor cleaning equipment. It sells process tools, not chips — a capital-equipment vendor levered to fab capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs..

Four reporting segments, but the P&L is effectively one business:

• SPE — Semiconductor Production Equipment (~80% of sales, the crown jewel): FY2026 revenue ¥485.9B, down 6.5% YoY, at a 25.2% segment operating margin. Cleaning tools (single-wafer + batch), plus coater/developer (litho track), thermal processing, and a large, high-margin post-sales service/parts stream off a 15,000+ cumulative installed base of cleaning tools.
• GA — Graphic/network Arts (~8-9%): commercial printing — computer-to-plate (CTP) and inkjet digital presses. Low-growth, lower-margin legacy.
• FT — Display/Finetech (~8-10%): flat-panel/OLED production and coater equipment; OLED-driven display demand surged this line +224.9% to ¥44.7B in FY2026. Advanced-packaging equipment is being folded into FT from the holding company.
• PE — PCB/printed-electronics (~2%): direct-imaging and PCB-related tools.

Contract structure: capital-equipment sales (lumpy, tied to customer fab build-outs and recognized on shipment/setup) plus a recurring, high-margin service/consumables annuity off the installed base — service revenue is concentrated in Q4, a structural feature that makes intra-year margins lumpy. Not take-or-pay; demand is a derivative of leading-edge and memory WFE (wafer-fab-equipment) capex.

Customers: the world's fabs — TSMC, Samsung, SK Hynix, Micron, Intel, Kioxia, and increasingly Chinese memory/logic fabs (CXMT and domestic foundries). Suppliers: precision components, pumps/valves, chemical-delivery and fluidics vendors, plus Japanese materials/parts SMEs. Competitors: Tokyo Electron (TEL), Lam Research, and Samsung's captive SEMES (see Lens 3)..

Supply Chain

Upstream → SCREEN → end customer:

Precision components / fluidics / chemical-delivery / robotics (mostly Japanese SME suppliers) → SCREEN (designs, assembles & services the cleaning + track + thermal tools at Hikone, Kyoto, Toyama; new Yasu site under prep) → Chipmakers' fab lines → TSMC / Samsung / SK Hynix / Micron / Intel / Kioxia / CXMT → (their chips) → NVIDIA / AMD / hyperscaler accelerators → AI data centers.

SCREEN sits at the "Equipment" node of the hardware value chain. It is a picks-and-shovels supplier to the memory + leading-logic build-out that CoWoS/HBM demand is driving — the same bottleneck map (CoWoS advanced packaging 9/10; HBM3E/HBM4 9/10) that gates NVIDIA also pulls SCREEN's tools, because HBM stacking, TSV formation, and hybrid bonding each require multiple cleaning + surface-prep steps.

Chokepoints / single-source dependencies:

• Geographic concentration in demand: China 38% + Taiwan ~22% = ~60% of sales from two jurisdictions — the binding supply-chain risk is political, not physical (export controls, see Lens 10/13).
• Manufacturing footprint concentrated in Japan (Hikone/Kyoto/Toyama) — FX-exposed (weak yen has been a tailwind) and single-country.
• Cleaning is a chokepoint the OTHER way: with SCREEN+TEL+Lam >90% of the cleaning market, fabs have limited alternative single-wafer suppliers — a source of SCREEN's pricing power (Lens 3).

Gap (web-only): named tier-1 component suppliers not disclosed publicly at line-item level.

Competitive Advantages (moats)

The moat is real and specific: cleaning process know-how + an installed base + co-development at the leading edge.

• Market structure: SCREEN, TEL, and Lam control >90% of the wafer-cleaning market; SCREEN is the #1 in single-wafer cleaning at ~31-45% share depending on the definition/source (SS/SU-3200-class single-wafer platform cited at ~31% of single-wafer; SCREEN's headline cleaning share quoted at ~45%). The overall wafer-cleaning-equipment TAM was ~$6.4B in 2025 → ~$9.9B by 2031 (~7.5% CAGR).
• Durable moats: (1) decades of accumulated cleaning recipes/process IP co-developed node-by-node with the top fabs — very high switching cost once qualified into a process flow; (2) installed base of 15,000+ tools generating sticky, high-margin service/parts revenue and deep customer entanglement; (3) scale + reference wins in single-wafer that compound (yield-sensitive customers stay with the qualified incumbent); (4) pricing discipline — management is explicit: "we do not want to compete through discounts," raising realized ASPs through new features/mix rather than price hikes.
• Bargaining power: strong over customers in single-wafer cleaning (few qualified alternatives, yield-critical step), but customers are giant and cyclical (TSMC/Samsung/SK Hynix/CXMT) and can defer capex — power is asymmetric across the cycle.
• The one structural crack: Samsung's captive SEMES supplies Samsung's own cleaning needs, structurally fencing SCREEN out of a meaningful slice of one of the three largest memory buyers.

Cleaning-intensity is the secular moat-widener: every node transition (GAA/gate-all-around, backside power delivery, 3D-DRAM, rising 3D-NAND layer counts) and every advanced-packaging step (TSV inner-wall cleaning, pre-hybrid-bond surface prep, particle removal) adds cleaning steps — cleaning demand grows faster than wafer volume.

Segments

FY2026 (ended March 2026), consolidated revenue ¥605.7B (−3.1% YoY):

*Segment granularity below SPE is a web-only gap — the FYE-Mar-2025 composition was SPE 78.8% / GA 9.3% / FT 9.8% / PE 1.8% / Other 0.4%; FY2026 line-item splits for GA/FT/PE were not cleanly sourced, and the reported "+224.9% to ¥44.7B" OLED figure does not reconcile to a full-FT-segment number.

The single most important segment fact: on the FY2027 guide (SPE ¥600B at ~25.5% margin → ~¥153B SPE operating income vs a total ¥150B OP guide), essentially 100% of group operating profit is SPE — GA/FT/PE net to roughly breakeven after corporate cost. SCREEN is, in economic substance, a semiconductor-cleaning company carrying three low-margin legacy attachments — which is both a focus argument (bull) and a latent sum-of-parts/divestiture question (Lens 14).

Geography (FY2026): China 38% (down from 42%), China+Taiwan ~60% — the segment story that matters most is geographic, not product (Lens 13).

---

Phase B — Measure performance

Earnings Result — FY2026 (year ended March 2026), reported 2026-05-13

A down year and a guide-down year — but the market looked straight through it to record orders.

• What drove it: SPE revenue −6.5% on project push-outs and revenue-recognition timing (setup vs. shipment basis), partly offset by a strong Q4 (SPE margin ~30%) and the OLED/display surge in FT. Gross margin actually expanded to 38.5% (from 37.6%) — the revenue decline was volume/timing, not price/mix erosion.
• Guidance / tone change: management reversed to a sharp FY2027 reacceleration — revenue ¥725B (+19.7%), OP ¥150B (+22.4%), margin 20.7%, SPE ¥600B (+23.5% at ~25.5% margin) — and framed FY2026's softness as "a heavy investment phase, not fundamental weakness".
• Balance-sheet flags: none negative — net cash (net debt/equity −0.45), current ratio 2.28, debt/equity 0.01, OCF ¥92.7B ≈ net income (clean cash conversion).
• Market reaction: violently positive on the record Q4 SPE order intake — the stock is +105% YTD and +56% in 30 days into this print. The tell: a revenue-miss + EPS-beat + record-orders print was treated as bullish — the market prices SCREEN off orders and the forward, not trailing revenue (see Lens 8).

Earnings Calls (sentiment trend)

Tone across the FY2026 calls moved from cautious/soft (H1) → confident/order-backed (Q4):

• What management repeats now: "record-high order intake" (Q4, strengthening into Q1/Q2); "we're making the investment into future growth… in the next mid-term plan we expect to harvest the fruit"; "we do not want to compete through discounts." Focus themes: AI-related memory (HBM/DDR5), advanced + mature foundry, advanced packaging, OLED recovery.
• What they stopped saying / got more skeptical on: they openly doubt competitors' 20-30% China-growth forecasts and keep a "cautious" China stance even as demand improved from January — a notably un-promotional, "show-me" posture on their biggest market.
• Recurring risk framing: near-term margins pressured by a ¥32B fixed-cost step-up (incl. ¥10.7B labor + record R&D/capex/depreciation); ASP-improvement (features/mix) is the stated margin lever for the next mid-term plan — i.e., the harvest is deferred.
• Net: management is credible and conservative (skepticism on China, no hype), which raises confidence in the quality of the guide — but also signals the best margin years are one plan-cycle out, not now.

Comps

Peer set = global WFE + Japanese semicap.

. †TEL/Advantest/Disco EV/EBITDA vary widely by source/date. SCREEN's own 5-yr avg ROE ~21.5% is from the FY22-26 series (19.9/21.0/21.0/25.1/20.3%); peer 5-yr-avg ROE not sourced (US-peer figures are TTM ROE, inflated by buyback-shrunk equity — SCREEN's 20% ROE sits on a net-cash base, arguably higher quality).

Read: SCREEN at P/E 36.5 / 28.7 fwd trades ABOVE its bigger, broader-line peer TEL (~29x) and above AMAT/KLA (~27-29x), roughly in line with Lam (~39x), and below the Japanese AI-test/packaging darlings Advantest (~54x) and Disco (~58x). On EV/Sales 5.18 and P/B 6.9, SCREEN is priced like a secular grower — rich for a name that just printed −3% revenue. The multiple is defensible only if the FY2027 +20% reacceleration + advanced-packaging S-curve deliver; it is not a value entry.

Stock-Price Catalysts

52-week range ¥5,335 → ¥19,770; now ¥18,150 — a ~3.4x move off the low, sitting near the high. YTD 2026 +105%, 30-day +56%.

What actually moves the stock (pattern):

1. SPE order-intake prints — the record Q4 orders (not the revenue miss) drove the latest leg. SCREEN's tape reacts to bookings + forward guide, not trailing revenue.
2. Sector AI-capex re-ratings — SCREEN trades as high-beta to the Japanese semicap complex (moves with TEL/Advantest/Disco) and to hyperscaler-capex/HBM sentiment.
3. Margin surprises — Q4 SPE ~30% margin and the OP beat mattered more than the revenue shortfall.
4. China memory-capex + export-control headlines — a two-way catalyst (Chinese fab orders up = bull; control-tightening = bear).

Analyst posture is the caution flag: avg rating "Buy" (15 analysts), but the sell-side target was raised only to ~¥18,200 (≈ spot) and Simply Wall St's DCF fair value is ¥15,567 (−14% vs spot). The stock has run to/through the price targets — the momentum, not the models, is setting the price. Gap (web-only): a precise dated log of individual >5% moves over 5 years was not sourced; the above is the characterized pattern.

---

Phase C — Judge people & books

Management

• CEO: Masato Goto — Representative Director, President & CEO since 20 June 2025 (~1 year tenure). An SPE lifer and internal promotion: President of SCREEN Semiconductor Solutions (the crown-jewel subsidiary) 2014-2019, then Head of Corporate Strategy. Succeeded Toshio Hiroe (CEO 2019-2025, now Chairman). Archetype: professional operator with deep domain roots, not a founder — appropriate for a 157-year-old, widely-held blue chip.
• Track record: ran the SPE business through its ascent; the group's operating margin roughly doubled from ~15% (FY2022) to ~20-22% (FY2025-26) and ROIC ran 28% → 48% → 36% over the same window. Goto inherits a strong hand rather than a turnaround.
• Skin in the game: insider ownership is small (professional-management, institutionally-held) — specific insider %/transactions not sourced (web-only; no our figures).
• Capital allocation — disciplined and shareholder-friendly: 13 consecutive years of >30% dividend payout; FY2027 dividend guided to a record (¥175 post-split); net-cash balance sheet, no value-destroying M&A on the record; heavy but organic reinvestment (record R&D/capex ~¥43B each, capacity expansion at Yasu/Hikone/Kyoto/Toyama). ROE 20-25% / ROIC 36-48% says the reinvestment earns its keep.
• Red flags: none material — conservative guidance, un-promotional tone, clean cash conversion. Watch-items rather than flags: (a) the service-revenue Q4 concentration that makes margins lumpy and (b) management's own framing of a fixed-cost step-up that defers the margin "harvest" to the next mid-term plan.

Forensic Red Flags

Accounting posture looks clean; the genuine risks are cyclicality and geopolitics, not the books. (Reports under Japanese GAAP; no SEC filings exist to cross-check, which is itself a transparency limitation for a US-lens analyst.)

• Revenue recognition: the FY2026 SPE softness and the reported single-wafer-share dip were both attributed by management to "timing differences in revenue recognition (setup vs. shipment basis)". This is plausible for capital equipment — but it is management's characterization, and it conveniently explains away a share loss; a skeptic verifies it against next year's realized share (Lens 13). Watch-item, not a violation.
• Cash vs. earnings: OCF ¥92.7B ≈ net income ¥92.0B — no divergence; earnings are cash-backed.
• Balance sheet: net cash, D/E 0.01, current ratio 2.28 — low financial-engineering risk; no evidence of receivables/inventory outrunning revenue in the sourced data (line-item working-capital detail not on the web-only shelf — gap).
• SBC / non-GAAP: Japanese reporters lean on statutory numbers; no aggressive non-GAAP adjustment culture evident. Margin lumpiness from Q4 service concentration is the main "watch the intra-year optics" item.
• Goodwill/intangibles / related parties: none flagged in sourced material (web-only limitation on the notes to accounts).

Regulatory findings (required):

• SEC (EDGAR LR + AAER): N/A — SCREEN has no CIK; no EDGAR enforcement search is possible.
• Non-SEC (FTC/DOJ/antitrust/export-control) web search: no material enforcement actions, settlements, fines, or penalties found against SCREEN Holdings / Dainippon Screen.
• 10-K Item 3 (Legal Proceedings): N/A — no SEC filings.
• Verdict: No material regulatory or legal findings — verified via SEC EDGAR EFTS (no CIK, N/A), non-SEC web search, as of 2026-07-10. The live regulatory risk is prospective export-control exposure on China revenue (Lens 13), not any existing action.

---

Phase D — Project & stress-test

Forward Projection

Base year — FY2026 actual (ended March 2026): revenue ¥605.7B · OP ¥122.5B · NI ¥92.0B · EPS ¥486.62 · ~189M shares. Anchor: management's FY2027 guide (revenue ¥725B, OP ¥150B). NI conversion ≈ 74-75% of OP (FY25 73.3%, FY26 75.1%). (A stock split is underway — "post-split ¥175" dividend — so per-share figures may re-base; shown here in current-share terms; P/E unaffected.)

All forward figures — arithmetic shown; base anchored to the sourced guide. Consensus check: forward P/E 28.7 on ¥18,150 implies a FY2027 consensus EPS ≈ ¥633 — above my guide-based ¥587 base, i.e. the street is modeling a beat and/or buyback-driven share reduction. Per --watchlist rules, no our model Brier forecast is logged in this sweep.

Key swing factors: (1) China order durability (±¥50-80B of revenue); (2) advanced-packaging ramp (¥100B FY2027 → ¥400-500B FY2033 aspiration — the multi-year EPS-mix upgrade); (3) operating leverage — capacity supports ¥680B + 10-20% growth with no new fab, so incremental margins can surprise up; offset by the ¥32B fixed-cost step-up.

Bull vs Bear

Bull case. SCREEN is the purest listed play on rising cleaning-intensity-per-node — the one process step that scales super-linearly with device complexity (GAA, backside power, 3D-DRAM, NAND layer counts) and with advanced packaging (TSV cleaning, pre-hybrid-bond prep). It owns #1 single-wafer share inside a >90%-concentrated three-firm oligopoly, monetizes a 15,000-tool installed base with a sticky service annuity, and compounds at 20-25% ROE / 36-48% ROIC on a net-cash balance sheet with 13 years of dividend growth and disciplined ("no discounts") pricing. Two S-curves stack: the cyclical FY2027 +20% WFE reacceleration (order-backed, record Q4 bookings) and the secular advanced-packaging ramp toward ¥400-500B by FY2033. A quiet-monopoly compounder that re-rates in an up-cycle.

Bear case (2-3 permanent-impairment or de-rating risks).

1. China is ~38% of sales and geopolitically fragile. Much of the FY2026-27 China order strength looks like a pull-forward ahead of tightening controls (BIS added China-wide HBM/DRAM equipment controls in 2024; management flags the "MATCH Act" could hit 10-15% of China investments). A control step-up or a CXMT-capex pause air-pockets the single largest revenue pool.
2. It is still a cyclical selling to a handful of cyclical buyers. The FY2026 −3% revenue / −10% OP down-year proves the moat does not flatten the WFE cycle. Six customers (TSMC/Samsung/SK Hynix/Micron/Intel/CXMT) set the tape.
3. Valuation has priced the recovery. 36x trailing / 29x forward / 5.2x EV/Sales / 6.9x P/B, +105% YTD, near all-time highs, consensus targets AT-or-BELOW spot. To make money from here SCREEN must beat a robust guide and hold the multiple.

Pre-mortem (18 months out, thesis broke — what happened?). A US/Japan export-control tightening or a CXMT/Chinese-fab digestion pause vaporized ~1/3 of the order book; the FY2027 +20% guide was cut mid-year (echoing FY2026's guide-down); the "share dip was just timing" line proved to be real single-wafer share ceded to TEL/Lam/SEMES; and the stock de-rated from ~36x to ~22x on a lower earnings base — a 40-50% drawdown with no company-specific fraud, just cycle + geopolitics + a popped multiple.

Are multiples too high? For the quality, no — for the entry, yes. 29x forward is not egregious for a 20%+ ROE oligopolist with an AP option, but the entry is at the top of a 3.4x move with targets at spot. Contrarian view (what the market refuses to see, both ways): the crowd trades SCREEN as generic AI-semicap beta, under-appreciating the durable, less-cyclical cleaning + AP-cleaning annuity (upside on a 5-year view) while simultaneously under-pricing the China pull-forward / control risk embedded in the next four quarters (downside near-term). Both can be true — which is exactly why the call is watch and buy the dip, not chase.

Devil's Advocate (short-seller)

Dismantling the bull case:

• Structural break in how it earns: ~60% of revenue rides China+Taiwan; the single biggest incremental buyer this cycle is Chinese memory/mature-logic (CXMT et al.) — precisely the cohort in the crosshairs of US/Japan/NL export controls. One BIS/METI rule change on China DRAM/HBM equipment can remove a chunk of the order book overnight, and the FY2026 China decline (42%→38%) shows the sensitivity is live.
• Concentration: demand is set by <10 fabs; a Samsung/Intel capex air-pocket or a TSMC digestion quarter hits hard — and Samsung's cleaning spend is largely captive to SEMES, so SCREEN is already fenced out of one giant.
• Moat weaker than bulls think? The reported single-wafer-share dip in FY2026 is the crack. Management says "revenue-recognition timing." Maybe. But if TEL and Lam (both bigger, better-capitalized, broader-line) are pushing single-wafer, or SEMES is expanding beyond Samsung, the "durable #1" narrative erodes — and SCREEN can't defend on price without breaking its own margin story.
• Most dangerous competitor bulls underestimate: Lam Research (plasma-enhanced/dry-cleaning adjacencies + scale + US-fab relationships) and TEL (co-sells cleaning alongside its dominant coater/developer track — a bundling threat).
• Capital-allocation / accounting knocks: genuinely few — the honest short is not an accounting short. It's cycle + geopolitics + multiple.
• What must hold for today's price: FY2027 +20% delivers, China orders don't pull-forward-then-collapse, single-wafer share re-expands as promised, and the 29x forward multiple survives contact with any WFE wobble.
• If growth disappoints 20-30%: revenue ~¥580-640B vs the ¥725B guide → OP margin compresses on the ¥32B fixed-cost step-up → EPS closer to ¥450-500 → at a de-rated 22-25x, a ¥11,000-13,000 stock (−30-40%).
• Single scenario that permanently impairs: a durable China decoupling (controls + Chinese domestic-equipment substitution, e.g. local cleaning-tool champions maturing) that both removes ~1/3 of demand and hands the vacated China share to domestic rivals — plausibility medium and rising with each control cycle.

Management Questions (ordered by information value)

1. Of the record Q4 SPE orders, what share is Chinese memory/mature-logic, and how much would you characterize as pull-forward ahead of potential export-control tightening vs. sustainable demand?
2. The FY2026 single-wafer-cleaning share dip — quantify it: was it purely revenue-recognition timing, and what is your actual unit/share position vs. TEL, Lam, and SEMES today?
3. What is your base/bear plan if BIS or METI extend China DRAM/HBM equipment controls — how many yen of the FY2027 ¥600B SPE guide is at risk, and how fast can non-China demand backfill?
4. Advanced packaging: bridge the ¥100B (FY2027) → ¥400-500B (FY2033) target — what is the revenue split across AP-cleaning, the Lemotia PLP coater, and LeVina exposure, and what share do you hold in each?
5. Margins: you won't "compete on discounts," but you also can't raise base prices — what realized-ASP and mix improvement underwrites the next mid-term plan's margin expansion, and by how many bps?
6. The ¥32B fixed-cost step-up — when does the "harvest" arrive, and at what revenue level does it become a margin tailwind rather than a headwind?
7. What is the incremental (drop-through) operating margin on revenue between today and the ~¥680B capacity ceiling before Yasu adds fixed cost?
8. SEMES/Samsung: is any structural path opening to win Samsung cleaning share, or should we model that spend as permanently captive?
9. Capital allocation: with net cash and 20%+ ROE, why the modest ~30-35% payout — what is the bar for buybacks or larger M&A vs. continued organic reinvestment?
10. The three legacy segments (GA/PE, and non-AP FT) contribute revenue but ~no group operating profit — what is the strategic case for keeping them vs. divesting to refocus on SPE + AP?
11. On the US Albany (ATCA) hub — is this about being designed-in to North American leading-edge fabs (Intel/TSMC AZ/Micron), and what revenue does it unlock by when?
12. How exposed is the FY2027 guide to FX — what USD/JPY and KRW/JPY assumptions underpin it, and what is the yen-sensitivity of operating profit?
13. What is your read on Chinese domestic cleaning-equipment substitution (local champions) over five years — where do you lose first, and how do you defend?
14. Hybrid bonding / sub-2nm: where does cleaning-intensity growth most favor SCREEN specifically, and what new process steps are you being designed into for GAA/backside-power and 3D-DRAM?
15. What single leading indicator (bookings, book-to-billBook-to-billNew orders divided by orders filled. Above 1 means the backlog is growing; below 1 means the company is working through it faster than it is replacing it., a customer segment) would you watch to know the cycle has turned down before it shows in revenue?

---