Daifuku

Phase A — Understand the business

Lens 1 · Company Overview

Daifuku is the largest material-handling-systems company in the world — ranked #1 in Modern Materials Handling's "Top Material Handling Systems Suppliers" survey for the ninth consecutive year as of the 2023 ranking. It does not make humanoid robots. It designs, integrates, installs and services the automation plumbing of physical industry: the overhead rail and stocker systems that move silicon wafers inside a chip fab, the conveyor/sorter/AS-RS systems inside an Amazon or Uniqlo distribution centre, the assembly-line conveyance inside a Toyota or BMW plant, the baggage-handling systems inside an airport, and (a quirky high-share niche) automatic car-wash machines.

Business model. Project-based systems integration with a large and growing after-sales service annuity. A fab or DC operator commissions a multi-year capex project; Daifuku engineers, manufactures (in regional plants), installs and commissions it, then earns recurring maintenance, parts, retrofit and software revenue across the installed base's 10–20-year life. The strategic thrust under the current plan is explicitly to shift mix from "scale" to "quality" — selecting higher-margin projects and growing the service/software layer.

Scale. FY2025 (Jan–Dec 2025): net sales ¥660.7B, operating profit ¥100.8B (15.3% margin), profit attributable to owners ¥78.0B. Roughly ~68–74% of sales are outside Japan — North America and China/Korea/Taiwan are the largest overseas markets, tracking global semiconductor and e-commerce capex.

Customers / suppliers / competitors. Named blue-chip end customers span five verticals: TSMC, Samsung, SK Hynix, Intel, Rapidus (semiconductor cleanroom/AMHS); Amazon, Walmart, Costco, Uniqlo, Shimamura (e-commerce/distribution); Toyota, Honda, BMW, Daimler/Mercedes, Ford (automotive); JFK, Heathrow, Changi, Narita, Haneda (airports). Suppliers are commoditised industrial inputs (steel, motors, controls, bearings — including precision actuators/reducers from Japanese peers Nabtesco/Harmonic Drive per the robotics supply-chain map ). Competitors differ by vertical (see Lens 7); the closest tracked peer in the research index is Symbotic (SYM) in US warehouse automation.

Contract structure. Large fixed-scope EPC-style project contracts (revenue recognised over time as work progresses) plus a recurring service tail estimated at a meaningful and rising share of the mix (the analyst teardowns attribute much of the margin step-up to "increased after-sales services"; precise service % n/a — not disclosed). Backlog is the key forward indicator: order backlog above ¥700B as of the May-2026 print.

Lens 2 · Supply Chain

Map: commoditised industrial inputs → Daifuku regional engineering/manufacturing → end-customer facility → multi-decade service tail.

Chokepoints / single-source dependencies. Daifuku is itself the chokepoint in its strongest niche: in semiconductor AMHS/OHT, Daifuku and Murata Machinery together hold a reported ~88% global share (Murata ~48.3%, Daifuku ~39.4% of AMHS), and in ultra-high-end logic-fab OHT Daifuku's share is reported >50%. It is not dependent on any single supplier; it is a near-single-source for some of its customers. Upstream commodity inputs are substitutable, so the supply chain's risk is demand-side (customer capex timing), not input scarcity. Names or it didn't happen — done.

Lens 3 · Competitive Advantages (moats)

Daifuku's moat is real, durable, and unusually wide for an industrial — but it is vertical-specific, not monolithic. Five sources of advantage:

1. Installed-base + switching costs (the core moat). Once an OHT rail network, stocker grid and control software are designed into a 300mm fab or a multi-million-square-foot DC, ripping it out mid-life is economically prohibitive. The customer is locked into Daifuku for retrofits, expansions, spares and the software stack. This converts a cyclical capex business into a partial annuity — and the service revenue carries higher margin than first-fit equipment.

2. Scale + reference moat in semiconductor AMHS. A fab is a >$10–20B asset where downtime costs millions/day; no chipmaker takes a flyer on an unproven wafer-transport vendor. Daifuku's reference list (TSMC global incl. Arizona advanced-node fabs, Samsung, SK Hynix, Intel, Rapidus) is itself the barrier. Result: a duopoly with Murata at ~88% combined AMHS share — among the most concentrated structures in capital equipment.

3. Cross-vertical breadth (diversification moat). Five end-markets (semi, e-commerce, auto, airport, car-wash) with uncorrelated-ish cycles let Daifuku smooth revenue: when auto-line capex paused on EV uncertainty (FY2025 North America −3.8% sales), AI-semiconductor cleanroom demand more than offset it (CFI Korea +45% sales). Few competitors play all five at #1/#2.

4. Niche monopolies. World #1 in airport baggage handling (500+ airports installed, incl. JFK/Heathrow/Changi); #1 car-wash machines in Japan (~40% share) and Korea (~60%) via Daifuku Plusmore. These are small but high-share, low-competition annuities.

5. Bargaining power. Strong over fragmented suppliers (commodity inputs); moderate over its largest customers — TSMC/Samsung are themselves monopsony-scale buyers and the duopoly structure means Murata is always a credible second source. The car-wash and airport niches give Daifuku the most pricing power; bleeding-edge logic fabs the least.

Moat verdict: A genuine wide-moat industrial in AMHS/cleanroom and the niche monopolies; a strong-but-contested integrator everywhere else. Ground-truth caveat: the official positioning.md on the shelf does not cover Daifuku (it covers humanoids), so this moat read is web-built, not research-layer-validated.

Lens 4 · Segments

Directional product mix (FY2025), from search snippets — treat as approximate:

• Intralogistics ~52% of revenue; Cleanroom ~26%; remainder Automotive + Airport — conflicts with a second snippet citing "Electronics & Semiconductors ~38% as the largest single division". The conflict is a definitional one (product segment vs end-market lens). Flagged; neither fabricated into a clean table.

Legal-entity segment cut (FY2025), reported consistently across two independent teardowns — this is NOT the official product segmentation; it is the reporting-entity view:

Trend & cause (accelerating / decelerating):

• Accelerating: Japan parent (margin ~22.5%, the profit engine — driven by AI-semiconductor cleanroom + service mix + cost discipline); Korea CFI (AI-semiconductor boom, orders +55.6%, sales +45.2%); China DSA orders +47.5% (self-sufficiency policy).
• Decelerating: North America (−6.6% OP) — the soft spot, on an automotive-investment pause amid EV-transition uncertainty; partially structural-mitigated by the doubled Indiana plant (tariff hedge).
• Geographic: ~68–74% overseas; North America ~26–30% of group revenue. Yen strength is therefore a direct translation headwind (FY2025 ~¥6.8B order headwind from FX).

---

Phase B — Measure performance

Lens 5 · Earnings Result (latest full-year print: FY2025, Jan–Dec 2025; reported 2026-02-12)

What drove it. This was a margin story, not a volume story — sales up only 2.6% but OP up 24.4%. The "structural profit reform" lifted gross margin by +¥16.1B via three levers: (1) price pass-through of input/labour inflation finally sticking; (2) mix — more high-margin semiconductor cleanroom + after-sales service replacing low-margin equipment; (3) cost discipline — "thorough order selection focused on profitability and elimination of low-profit projects" plus product standardisation cutting design/manufacturing hours. This is the single most important fact in the dossier: Daifuku has structurally re-rated its own margin from low-teens toward mid-teens, and management's FY2027/2030 plan assumes it holds (15.0% target).

Guidance/tone — FY2026 (Jan–Dec 2026): net sales ¥700.0B (+5.9%), OP ¥105.0B (+4.2%), net income ¥80.0B (+2.4%), DPS ¥82 (37.7% payout), orders ¥780–820B (+16–22%). The orders-vs-sales gap is the tell: management is guiding a big order acceleration (the negotiation pipeline is "extremely large with rising close probability") that converts to revenue with a lag — conservative near-term P&L guidance, large forward book.

Balance-sheet flags: clean. Equity ratio ~60%, D/E ~0.14x, total assets ¥754.2B, cash-conversion cycle improved 25 days to 74 days. No leverage risk; the risk is entirely demand cyclicality.

Market reaction (the most diagnostic line): On the Q1-FY2026 print (reported ~May 14, 2026) — sales +8% to ¥172B, margin 15.2%, orders +55% to ¥221B, backlog >¥700B — the stock fell ~6% the next day (May 15). A blowout order number sold off. That says expectations are rich and the stock now trades on future semiconductor-capex momentum, not on beating the current quarter — a classic late-cycle "good news, sold" signature.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the research layer (transcripts=0) and Japanese-domestic calls are not on Fool/Insider-Monkey, so this is inferred from successive IR releases, not a verbatim transcript read:

• Through FY2025: tone moved from "recovery/normalisation" to "record-high, structural-margin" confidence — "record sales and profit," "three profit figures at record highs for four consecutive periods".
• Recurring phrases (rising): "order selectivity / profitability focus," "structural profit reform," "AI-related semiconductor demand (HBM, CoWoS)," "local production for local consumption" (tariff narrative), "shift from scale to quality."
• Things they stopped saying / de-emphasised: pure top-line/volume targets — the FY2027 plan was revised to raise the profit target (OP ¥92B→¥120B) while holding the sales target (¥800B), i.e. management openly reframed the story from growth-at-any-cost to margin. New caution entered on FX (¥150/$ assumption) and US tariffs.

Sentiment trajectory: rising and confident, but increasingly hedged on macro (FX/tariffs) — consistent with a management that has delivered the margin reset and is now managing expectations down on the cycle.

Lens 7 · Comps

Read: at ~27x forward earnings with ~18% ROE and a ~60% equity ratio, Daifuku is priced as a quality compounder, not a cheap cyclical — roughly a market-to-premium multiple for a business whose largest profit driver (semiconductor AMHS) is one of the more cyclical lines in capital equipment. The single sharpest comp observation: its closest economic peer in dominance is Murata, which is private, so the public market has no pure-play AMHS comparable — investors price Daifuku off intralogistics peers (KION/Jungheinrich, structurally lower-multiple) and off the semi-cap-equipment complex (higher-multiple). That valuation ambiguity is part of why it gyrates. Where multiples are blank above, they were not sourceable in this pass — a refresh should pull KION/Jungheinrich/Symbotic forward P/E and EV/EBIT from a financial-data source.

Lens 8 · Stock-Price Catalysts (moves >5%, last ~5 years)

Mostly ``; the research layer has no price history.

• 2022–early 2024 — semiconductor-capex pullback / China property + memory glut: material-handling orders soften; stock de-rates. (52-week low ¥3,150 stamped within the trailing year as of Feb 2026.)
• 2024 → Feb 2026 — the AI-semiconductor re-rating: stock roughly doubles off the lows to a 52-week high ¥6,516 (Feb 2026) as HBM/CoWoS fab-build demand + the margin reset compound; trading ¥6,472 late Feb 2026.
• 2026-02-12 — record FY2025 print + raised FY2027 profit target: positive catalyst, confirmed the margin-reset thesis.
• 2026-05-15 — Q1-FY2026: orders +55%, backlog >¥700B, yet stock −6%: the most informative move — a beat sold off, signalling rich expectations and a late-cycle posture.

Pattern the market actually reacts to: Daifuku is a high-beta semiconductor-capex and AI-build proxy. It moves on (1) the semi/AI WFE cycle (the dominant driver — fab-build announcements from TSMC/Samsung/SK Hynix and the HBM/CoWoS narrative), (2) order/backlog momentum more than the current-quarter P&L, and (3) yen/$ (overseas-heavy translation). It reacts less to airport/car-wash/auto news. Net: own this knowing the tape treats it as a chip-capex derivative, not a diversified industrial.

---

Phase C — Judge people & books

Lens 9 · Management

• CEO: Tomoaki Terai (b. 1968, CEO since Jan 2026). Cal State University business/economics grad (1993); joined Daifuku 1994 — a ~30-year lifer. Path runs through the operating core: director of the US holding co (Daifuku Webb / North America) from 2011, chairman of Taiwan Daifuku from 2015, GM of the Production Division (2017) and Cleanroom Division production (2021), then COO/EVP before stepping up. Implication: an operator steeped in the two highest-value franchises — semiconductor cleanroom and overseas (US/Taiwan) — exactly the areas driving the margin reset. A continuity appointment, not a turnaround outsider.
• Chairman: Hiroshi Geshiro — prior CEO/President, moved to Chairman Jan 2026. Orderly succession; the prior management owns the structural-profit-reform that just delivered.
• Track record (quantified): the team executed a margin reset from ~12.6% to 15.3% OP margin and ROE from ~15% to 18.4% while holding the FY2027 ¥800B sales target and raising the FY2027 OP target ¥92B→¥120B — i.e. credibility earned by under-promising on sales and over-delivering on profit. ROE 18.4% already exceeds the FY2027 17% target — a rare case of a Japanese industrial outrunning its own medium-term plan early.
• Capital allocation: disciplined and shareholder-friendly by Japanese standards — payout policy ≥35% (FY2026 37.7%), DPS rising ¥78→¥82, performance-linked; "periodic strategic share buybacks" referenced. Growth capex is self-funded (FCF ¥51.8B, net cash). The Indiana-plant doubling is a sensible localisation/tariff hedge. No value-destroying M&A flagged.
• Skin in the game (red-flag-adjacent): CEO direct ownership 0.005% (¥110M); total comp ~¥158M (31.6% salary / 68.4% bonus+equity). This is the standard Japanese large-cap pattern — negligible insider ownership and modest comp — so alignment runs through career/reputation and the governance code, not through an equity stake. Don't expect founder-style ownership alignment; do note the board has five outside directors (Franklin, Yoshida, Kanzaki, Hongo, Nakamura) and a PwC audit relationship.
• Archetype: professional-manager / salaryman-meritocracy, not founder-led. Appropriate for a mature, scaled, multi-vertical global integrator — the value-add here is operational selectivity and mix-shift, which this team has demonstrably delivered.

Lens 10 · Forensic Red Flags

Acting as forensic analyst. Daifuku screens clean on the obvious vectors, with the usual caveats of a non-SEC IFRS reporter and project-accounting model.

• Revenue recognition (the one to watch): systems integration recognises revenue over time / percentage-of-completion on long-duration fixed-price contracts. This is the inherent soft spot of every EPC-style business — % completion and cost-to-complete estimates are judgement-heavy and can flatter or defer profit. Mitigant: the margin step-up came alongside an improving cash-conversion cycle (−25 days to 74 days) and FCF of ¥51.8B vs ¥78.0B net income — cash is lagging earnings somewhat (FCF/NI ~66%), which is normal for a growing project business funding working capital, not a divergence flag. OCF ¥76.1B ≈ net income — healthy. No evidence of receivables/inventory outrunning revenue; CCC improved.
• Segment reporting: here is a genuine transparency caveat — the company reports a legal-entity segment cut that does not cleanly map to the product segments (Intralogistics/Cleanroom/Auto/Airport), and the official product-segment P&L was not machine-readable in this pass. That is a disclosure-granularity friction, not a fraud signal, but it limits external scrutiny of where margin truly sits (parent ~22.5% vs DNA ~9.2% tells you Japan/semi carries the group). Flagged.
• SBC / non-GAAP: Japanese GAAP/IFRS reporting; no aggressive US-style non-GAAP adjustment culture; SBC is immaterial (negligible options). Low risk.
• Goodwill/intangibles, leases, related parties, contingencies: no material flags surfaced; ~61 subsidiaries add consolidation complexity but the equity ratio (~60%) and low D/E (0.14x) argue against hidden leverage. Item-3-equivalent legal-proceedings disclosure: n/a — no 10-K (non-SEC filer); reviewed via web instead.

Regulatory findings (required sub-section):

• SEC (EDGAR LR + AAER): None possible — Daifuku has no CIK and is not an SEC registrant; fetch-regulatory-findings.ts returned total_sec_findings: 0 and noted "no EDGAR enforcement search is possible".
• Non-SEC enforcement (web): the search "Daifuku" (FTC OR DOJ OR FDA OR CFPB OR consent-decree OR settlement OR fine OR penalty) enforcement and a governance/scandal/litigation/recall/fraud search returned no material findings — results surfaced only Daifuku's own governance/compliance disclosures and unrelated Japanese corporate scandals (Olympus, Daiwa). Auditor: PwC.
• 10-K Item 3: n/a — non-SEC filer (no Form 10-K exists).
• Conclusion: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR/AAER → not applicable, no CIK), targeted web enforcement/scandal search, and the absence of any product-recall or fraud hit, as of 2026-06-30. Caveat: as a foreign domestic filer, JFSA/TSE-level disclosure is not indexed in EDGAR; a deeper pass would check Japanese-language regulatory sources.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (EPS, next three fiscal years: FY2026E / FY2027E / FY2028E, all Dec-year)

Built bottom-up from FY2025 actuals + management guidance. Share count derived from market cap ÷ price: ¥2.59T ÷ ¥6,472 ≈ 400M shares. FY2025 actual EPS ≈ ¥78.0B / 400M ≈ ¥195. (One teardown cites FY2026E EPS ~¥217.6 implying ~358M shares — a share-count discrepancy I flag; I use the market-cap-derived ~400M for internal consistency and show both.) All outputs ``; inputs labelled.

Base / Bull / Bear (FY2027 EPS, ~400M-share basis):

• Base ~¥230 — management plan delivered, 15% margin holds, orders convert on the ¥780–820B FY2026 book; semi/AI capex stays firm into 2027. P/E 27x → ~¥6,200.
• Bull ~¥270 — AI-fab supercycle (HBM/CoWoS, US/Japan/Taiwan/Korea fab waves) pushes sales past plan and margin toward 16%; multiple holds or expands on "structural compounder" recognition.
• Bear ~¥160 — a 2027 semiconductor-capex air-pocket (a real risk given >50% logic-OHT concentration) cuts orders 20–30%, fixed-cost deleveraging drops margin to ~13%, and the multiple compresses to ~18–20x as the market re-cyclicalises the name → a double de-rate (lower E and lower multiple) that could halve the stock from the highs.

Forecast log (per skill): in --watchlist (unattended) mode the skill says skip forecast.ts create — only log when genuinely committing to the base case. Not logged this pass. Suggested base call for a future committed log: 6383.T FY2026 net income >= ¥80B (mgmt guide), p≈0.6, resolves 2027-02 (FY2026 results).

Lens 12 · Bull vs Bear

Bull case. Daifuku is a world-monopoly-grade industrial that the market still half-prices as a cyclical integrator. It owns ~88% (with Murata) of semiconductor AMHS and >50% of ultra-high-end logic OHT — the picks-and-shovels of the AI-fab buildout — plus #1 global airport baggage handling and #1 Japan/Korea car-wash, all feeding a high-margin installed-base service annuity. Management just structurally reset margins (12.6%→15.3%) and ROE (→18.4%, already past the FY2027 target), guided a +16–22% order acceleration for FY2026, and runs a fortress balance sheet (net cash, ~60% equity ratio) with rising shareholder returns. Three secular tailwinds — AI-semiconductor fab capex, developed-market labour shortages driving warehouse automation, and aging-airport replacement — all point one way. The contrarian earnings surprise the bulls bank on: order-to-revenue conversion of the record backlog drives FY2027 sales past plan while the margin reset proves permanent, not cyclical.

Bear case (2–3 permanent-impairment-or-derate risks).

1. Semiconductor-capex concentration = cyclical beta dressed as a compounder. The profit engine (Japan parent margin ~22.5%, CFI Korea, the cleanroom mix) rides the most cyclical end-market. A WFE downturn — memory gluts and fab-build pauses are historically violent and recurring — would deleverage fixed cost and simultaneously trigger a multiple de-rate from ~27x toward cyclical-trough multiples. This is the dominant risk and it is a valuation risk as much as an earnings one.
2. Expectations are now rich and late-cycle. A +55% order quarter sold off −6% (May 2026) — the tape is no longer rewarding beats. At ~27x, the bar is "supercycle continues," and any 2027 air-pocket re-rates hard.
3. Margin-reset durability is unproven across a downcycle. The 15.3% margin was built in an up-cycle on price pass-through + mix + order selectivity. Selectivity is easy when the book is full; in a downturn, the temptation to chase low-margin volume to fill plants returns. The 15% "structural" margin has not yet been tested by a recession.

Pre-mortem (it's late 2027, the thesis broke): AI-fab capex digested its 2024–26 pull-forward and TSMC/Samsung/SK Hynix paused new starts; Daifuku's record backlog converted but the next book collapsed; North America stayed weak (auto EV-pause persisted, e-commerce DC build cooled); the yen strengthened to ¥130/$ on BoJ normalisation, gutting translated overseas profit; margin slipped to ~13% on deleveraging; and the multiple halved as the market re-labelled it "Japanese semi-cap cyclical." Stock −40–50% from the highs.

Are multiples too high? At ~27x for an ~18% ROE net-cash compounder, fair-to-full in an up-cycle, expensive if you (correctly) weight the cyclicality. The honest read: the quality justifies a premium to KION/Jungheinrich; the cyclicality argues you don't pay peak multiple on peak margin on peak orders — wait for the cycle to give you a better entry.

Contrarian view (what the market refuses to see): Both bull and bear over-index on the semiconductor line. The genuinely under-appreciated asset is the boring annuity — the >500-airport baggage installed base, the 40–60%-share car-wash monopoly, and the cross-vertical service tail — which provides a through-cycle floor the market ignores when it trades the name as a pure AI-fab proxy. The market is refusing to see that the diversification is the moat, not the semiconductor share.

Lens 13 · Devil's Advocate (short-seller)

Dismantling the bull case.

• What structurally breaks the money-machine: a synchronised semiconductor-capex downturn. AMHS is discretionary, lumpy, fab-build-triggered — when fabs stop breaking ground, OHT orders don't slow, they stop. >50% logic-OHT share means Daifuku has nowhere to hide in that scenario; you can't gain share when the market goes to zero.
• Revenue concentration / what if it shifts: the profit is concentrated in the Japan parent + semiconductor/cleanroom entities; North America (~26–30% of revenue) is the low-margin, decelerating piece (OP −6.6%). If semi rolls over while NA stays weak, the high-margin engine and the volume engine fail together.
• Why the moat is weaker than bulls think: Murata is co-equal (48% vs 39% AMHS) and private — it doesn't need to earn a public multiple and can price aggressively to win marquee fabs. In a downturn, a credible second source means pricing power evaporates exactly when Daifuku needs it. In intralogistics, AutoStore, Locus, AMR challengers and Honeywell/Dematic are real; Daifuku is not #1 in every DC bake-off.
• Most dangerous competitor bulls underestimate: Murata in semi (the duopoly can become a price war), and in warehouse automation the cube-storage/AMR cohort (AutoStore, Symbotic, Locus) that wins greenfield e-commerce on flexibility/cost.
• Worst capital-allocation / governance knocks: none egregious — but ~0.005% CEO ownership means management is not financially aligned with shareholders the way a founder is; the legal-entity-vs-product segment opacity reduces external accountability; and the ~61-subsidiary sprawl is hard to scrutinise.
• What must hold for today's ~27x: AI-fab capex stays elevated through 2027–28, the 15% margin survives untested, the yen doesn't strengthen materially, and orders keep converting. Three of four are macro/cyclical — outside management's control.
• If growth disappoints 20–30%: as modelled in Lens 11 bear — EPS to ~¥160 and multiple to ~18–20x = a stock that halves from the highs. The downside is geared because it's a double de-rate.
• Single permanent-impairment scenario & plausibility: a structural loss of AMHS share to Murata or a Korean/Chinese entrant on a major node transition — plausibility low (reference/switching moat is strong), so the realistic bear is cyclical+valuation, not terminal. The business doesn't break; the stock can, badly, on the cycle.

Lens 14 · Management Questions (ordered by information value)

1. The 15.3% operating margin was built in an up-cycle on price pass-through, mix and order selectivity — what is the trough margin you would defend in a 20–30% semiconductor-capex downturn, and which lever (selectivity vs volume) gives first?
2. What share of group revenue and group operating profit is now recurring after-sales service, and what is its margin versus first-fit systems? (The single missing number in your disclosure.)
3. Please reconcile the legal-entity segment cut with the product segments (Intralogistics/Cleanroom/Automotive/Airport) — where does operating profit actually sit by product?
4. In semiconductor AMHS you and Murata are ~88% of the market. How does pricing behave in a downturn with a co-equal private competitor that doesn't answer to public shareholders?
5. What is your order-to-revenue conversion lag on the ¥780–820B FY2026 order guide, and how much of that backlog is cancellable or re-timeable if a customer pauses a fab?
6. North America operating profit fell 6.6% on the auto pause. What is the path back — and how much of the Indiana capacity doubling is utilised vs speculative?
7. Your FY2026 plan assumes ¥150/$. Quantify the OP sensitivity per ¥10 move, and your hedging/natural-hedge posture as overseas production rises.
8. ROE is 18.4%, already past the 17% FY2027 target. What is the ceiling, and at ~60% equity ratio and net cash, why not lever modestly or buy back more aggressively to lift it further?
9. What is your customer concentration in semiconductor AMHS (TSMC/Samsung/SK Hynix combined as % of segment revenue), and how do you de-risk a single-customer fab-plan delay?
10. Where are the cube-storage/AMR challengers (AutoStore, Symbotic, Locus) winning DC bake-offs against you, and what is your competitive answer in flexible/greenfield e-commerce?
11. On capital allocation: criteria for buyback vs dividend vs M&A vs organic capex, and would you do a transformative acquisition or stay disciplined-organic?
12. What is the software/digital-twin attach rate today and the margin/stickiness uplift you expect from it by 2030?
13. Walk through the ¥1T-sales / ¥150B-OP 2030 vision bridge — how much is semi vs intralogistics vs airport vs service, and what's the biggest execution risk?
14. CEO insider ownership is ~0.005%. How is management financially aligned with minority shareholders through a cycle, beyond annual bonus?
15. Which one end-market would you most want to reduce dependence on over five years, and what are you doing about it?

---