THK

Phase A — Understand the business

Lens 1 · Company Overview

THK Co., Ltd. ("THK" = Toughness, High quality, Know-how) was founded in 1971 by engineer-inventor Hiroshi Teramachi, who commercialised the world's first Linear Motion (LM) Guide in 1972 — the recirculating-ball linear bearing that lets a machine axis move on a rail with near-zero friction and micron-to-submicron repeatability. It IPO'd on the Tokyo Stock Exchange in 1989. The company is a Tier-1 supplier of the "muscles and joints" of automated machinery — it does not sell finished machines (mostly), it sells the motion components inside them.

What it makes. A deep catalogue of precision mechanical components: LM guides, ball screws, ball splines, LM guide actuators, cross-roller rings, electric/linear actuators, cam and roller followers, linear bushings, and — until 2025 — automotive link balls, rod ends and suspension ball joints. Adjacent lines include seismic isolation / vibration-damping devices (buildings, data centres) and, through subsidiary SEED Solutions, robot hands/grippers (the General-Purpose Gripper TRX) and medical-robot components.

Who buys it. End markets are machine tools, general industrial machinery, semiconductor & LCD/OLED manufacturing equipment, industrial robots, precision instruments, electronic-device assembly, medical/assistive devices, aerospace, and transport systems. THK sells directly to industrial-machinery OEMs in Japan (machine-tool builders, robot makers, semi-equipment makers) and through distribution globally — a "spec-in" model where an LM guide grade gets designed into an OEM's machine and repeats with that machine's life cycle.

Contract structure. Predominantly book-and-ship component orders, not take-or-pay or recurring subscription. Revenue is therefore cyclical and order-driven, tracking machine-tool and semiconductor-equipment capex cycles — the single most important fact about the P&L. THK reports order trends because book-to-bill is the leading indicator; Q1 FY2026 commentary was explicitly that "orders for its industrial machinery business both in Japan and overseas have been tracking above initial projections". Customer concentration is low/diffuse (thousands of OEM and distributor accounts) — the concentration risk is at the end-market level (semi + machine tools), not single-customer.

The 2025 transformation (defining event). Effective the fiscal year ended Dec 31 2025, THK reclassified its entire Automotive & Transportation business as a discontinued operation and agreed to sell it. This is not a footnote — it re-bases the whole company. The old ~¥350B-revenue, two-segment THK (Industrial Machinery + Transportation Equipment) is becoming a ~¥260–276B pure-play industrial-motion company. See Lens 5.

Lens 2 · Supply Chain

Upstream (inputs → THK). The core input is high-grade bearing/alloy steel (chromium steel, case-hardening steels) machined into rails, balls, and screws; plus heat-treatment energy, grinding consumables, and increasingly electronics (encoders, motors, drivers) for its actuator and SEED robot products. THK is vertically integrated in the value-add step that matters — proprietary grinding, ball-circulation-path design, and heat-treatment know-how — so its supplier bargaining exposure is to commodity steel and energy, both of which it can largely pass through with a lag. Named steel/material suppliers are not publicly disclosed at line-item level (n/a — not disclosed), but the input is a commodity with many sources, which is a favourable position.

Midstream (THK's own footprint). A global manufacturing network: Japan — Yamaguchi (flagship LM-guide plant), Mie, Gifu, Yamagata, Kofu; China — THK Manufacturing of China at Wuxi, plus a Dalian R&D centre and plants and ~30 sales outlets nationwide; United States — Ohio (THK Manufacturing of America); Europe — production in France/Germany. This on-the-ground overseas capacity is a genuine competitive asset versus pure-exporter rivals and a hedge against tariffs/FX (a stronger yen and US tariffs were both cited as FY2024 headwinds).

Downstream (THK → end customer). THK → industrial-machinery OEMs → end user. Named buyer categories with the actual company archetypes: machine-tool builders (DMG Mori, Mazak, Okuma, Makino), industrial-robot makers (FANUC, Yaskawa, Kawasaki, ABB, KUKA), semiconductor & FPD equipment makers (Tokyo Electron, Nikon, Canon, ASML sub-suppliers, Applied Materials sub-assemblies), and — the future call option — humanoid-robot developers (Tesla Optimus, Figure, Agility, plus Chinese entrants) sampling linear actuators/roller screws. THK does not publicly break out named-customer revenue share (customers.csv is empty) — treat all named buyers as representative, not disclosed shares.

Chokepoints / single-source dependencies. THK is itself a chokepoint for its customers more than it is dependent on any supplier — for ultra-high-precision LM guides there are effectively two credible global sources (THK and, at lower price/precision, Hiwin), so a machine-tool builder specifying THK grade has real switching friction. THK's own key dependency is cyclical end-demand (semi + machine tools), not a physical input bottleneck.

Lens 3 · Competitive Advantages (moats)

The moat is real, narrow, and process-based. THK co-created the LM-guide category and has compounded ~50 years of grinding, ball-recirculation and preload/accuracy know-how that does not transfer cheaply. Independent market maps put THK + Hiwin at ~30–40% combined global linear-guide share, with the top three (THK, Hiwin, NSK) at ~75% — a consolidated oligopoly. THK sits at the ultra-high-precision / high-reliability end; Hiwin (Taiwan) undercuts it by 20–35% at comparable accuracy grades and is the price-share taker; NSK is strong in semiconductor/metrology and matched-brand axis assemblies; Bosch Rexroth owns European machine-building; IKO/Nippon Thompson is the crossed-roller/miniature specialist; Schaeffler (INA), Ewellix (ex-SKF), PMI, Schneeberger round out the field.

Durable moats, ranked:

1. Switching costs / spec-in lock. Once an LM-guide grade is designed into an OEM's machine and validated for accuracy/life, re-qualifying a rival is costly and risky. This is the strongest moat and it is sticky, not toll-booth — it protects share, not pricing power against Hiwin.
2. Process/scale IP. Proprietary manufacturing (grinding, heat-treat, circulation geometry) + the broadest catalogue and global local-production/service footprint. Hard to replicate; slow to erode.
3. Brand = reliability in mission-critical uses (semiconductor, medical, aerospace) where a bearing failure scraps a wafer or a patient procedure — buyers pay up for THK there.

Where the moat is weaker than the multiple implies: it is a share moat, not a pricing moat. Hiwin's structural 20–35% price gap caps THK's ability to raise price, and Chinese entrants (e.g. Wuzhou Spring quoting planetary roller screws at ~1/3 of Japanese pricing; Beite mass-producing 2.6M roller-screw sets/yr from 2026) are attacking exactly the commoditising end that humanoid-scale volumes would create. The moat is excellent at defending the precision niche and mediocre at defending volume economics — which matters enormously for the humanoid thesis (see Lens 13).

Bargaining power. Strong over commodity-steel suppliers; moderate over customers (they need THK-grade precision, but have Hiwin/NSK as credible second sources and are themselves large OEMs).

Lens 4 · Segments

THK reports five geographic segments — Japan, the Americas, Europe, China, and Other (Taiwan, Singapore, etc.) — not product segments. Historically it also split by business: Industrial Machinery Business (IMB) vs Transportation Equipment Business (automotive). The FY2025 reclassification collapsed that to essentially IMB-only continuing operations — the automotive business is now discontinued/divested (Lens 1, Lens 5).

Rebased revenue (continuing operations = Industrial Machinery):

• FY2024 continuing ≈ ¥241B; implies the divested automotive business was ≈ ¥112B of the old ¥352.76B FY2024 combined revenue — i.e. THK just shed ~32% of its top line.
• FY2025 continuing ≈ ¥260B, +7.9% YoY.
• FY2026E continuing ¥276B guidance (revised up), +6.2%.

Regional trend within FY2025 continuing ops:

• Japan ¥110,859M, −1.8% YoY — the largest single region (~43% of continuing revenue ); domestic machine-tool demand still soft.
• Europe ¥67,516M, −0.1% YoY — flat, weak industrial demand.
• Americas + China + Other — carried the +7.9% group growth, i.e. the recovery is electronics/semiconductor-led in Asia ex-Japan and the US. This matches management's "strong electronics-related demand and recovery in key regions" framing.

Read: the growth engine has rotated from Japan (structurally mature, ~flat) to the semiconductor/electronics capex recovery outside Japan. The segment mix is improving — divesting low-margin automotive lifts the corporate margin and ROE (the entire strategic point), but leaves the company more levered to the semi/electronics cycle and less diversified.

---

Phase B — Measure performance

Lens 5 · Earnings Result (the transformation print)

Two things happened at once in the FY2025 accounts (year ended Dec 31 2025, reported Feb 12 2026): a large GAAP loss driven entirely by the automotive exit, sitting on top of a quietly recovering core.

Headline (as reported):

• GAAP net loss attributable to owners: −¥69,891M, caused by an ¥81,639M loss on business liquidation of the discontinued automotive segment (booked in "Loss from discontinued operation"). Net margin −25.7%.
• Continuing operations were profitable: continuing-ops basic EPS ¥89.85 ⇒ continuing net income ≈ ¥10.1B; continuing revenue ≈¥260B, +7.9%; continuing operating income −9.3% YoY (cost inflation, restructuring, stronger yen, US tariffs).
• Group gross margin 30.44%, operating margin 8.60% (trailing, blended).

Guidance / outlook (FY2026, revised UP in May 2026 after Q1):

• Revenue ¥276,000M (+6.2%), operating income ¥31,000M (+19.2%), net attributable ¥22,700M (+5.6%). FY2026E operating margin ≈ 11.2% and net margin ≈ 8.2%. Note net income is guided to more than double off the ~¥10B FY2025 continuing base — an aggressive recovery assumption.

Q1 FY2026 (Jan–Mar 2026, reported May 11 2026):

• Revenue (continuing) ¥69.0B, operating income up >3x YoY, net attributable ¥4.5B, EPS ¥40.09. A clean cyclical-rebound quarter that justified the guidance raise; orders "tracking above initial projections."

Balance sheet: cash ¥95.52B, total debt ¥140B, net debt ¥44.48B (modest leverage), implied book equity ≈¥253B — the ¥69.9B loss took a real ~20% bite out of equity, which is why P/B optically jumped.

What's unusual vs THK's own history: (1) the first GAAP loss in memory, entirely one-time; (2) a discontinued-ops reclassification that shrinks revenue ~32%; (3) the market rewarding the loss — the stock nearly doubled through the year the company printed a ¥70B loss, because the market is buying the post-transformation THK. The print is bifurcated: poor on GAAP, strategically constructive on continuing ops.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the shelf (transcripts=0); reconstructed from reported call/press summaries [all web]:

• FY2024 (Feb 2025): defensive — "revenue flat, OP −26.8%," blaming cost hikes, tariffs, stronger yen, weak machine-tool and automotive demand. Tone: cyclical trough, cost-defence.
• FY2025 (Feb 2026): pivot to transformation — the language shifts to "structural reforms and resource reallocation to improve profitability," ROE, capital efficiency, and the automotive exit. Management stops apologising for the cycle and starts selling the portfolio reshape. A "workforce renewal initiative" (up to 120 staff, special retirement) signals a leaner cost base.
• Q1 FY2026 (May 2026): confident — guidance raised, "orders tracking above projections," "industrial machinery demand recovering." Tone: recovery confirmation.

Trend: cost-defence (2024) → strategic-reshape (2025) → recovery-confirmation (2026). The consistent through-line management now leads with is ROE / capital efficiency / focus — a deliberate re-messaging toward the Tokyo-Stock-Exchange "improve capital efficiency" agenda. What they've stopped saying: apologies for automotive. What they keep saying: robotics/DX/automation as the secular demand story.

Lens 7 · Comps

Peer table — global precision-motion / factory-automation / robot-component makers. Multiples are `` with source/date; where a figure isn't cleanly sourced it is marked n/a. No multiple is fabricated.

Read: on forward P/E, THK (22.5x Street) is the cheapest of the motion/humanoid cohort — below Hiwin (38x), Yaskawa (32x), NSK (25.5x), Nabtesco (24x), and Harmonic Drive (~127x trailing). The entire Japanese/Taiwanese automation cohort has re-rated on the humanoid theme; THK is not the frothiest name in it. But two provenance-critical caveats: (1) THK's "forward P/E 22.5x" implies a Street net-income estimate of ~¥38B — versus the company's own FY2026 guidance of ¥22.7B; on management's number THK trades at ~38x. The Street is modelling a recovery ~65% above guidance (or rolling to FY2027). Surface this, don't paper over it — it is the single biggest number-risk in the whole file. (2) On EV/EBITDA (20.2x) and P/B (3.41x), THK is expensive versus its own sub-book cyclical history and versus Nabtesco's 9.4x. THK is "cheap within an expensive cohort, expensive versus itself."

Lens 8 · Stock-Price Catalysts

Specific dated intraday moves aren't cleanly sourced without the tape (n/a on precise per-event %); the pattern over the last ~2 years is well-evidenced:

• +98.76% over the trailing 52 weeks — the stock roughly doubled. This is the dominant fact.
• Humanoid-robot theme (through 2025): inclusion in Morgan Stanley's "Humanoid 100" value-chain map (May 2025) and the Feb-2026 CNBC "Musk stressing robots over cars — buy the parts suppliers" cycle put THK on every humanoid screen. This — more than fundamentals — drove the multiple.
• Automotive-divestiture announcement (FY2025 results, Feb 12 2026; definitive AP87 agreement Feb 2 2026): received as ROE-accretive portfolio clean-up — the market cheered a strategy that reduced revenue.
• Q1 FY2026 guidance raise (May 11 2026): recovery confirmation, order strength.
• Semiconductor/electronics capex recovery (2025–26): the real cyclical tailwind under the theme.
• TSE "capital efficiency / P/B > 1" reform + yen weakness: broad re-rating tailwind for Japanese industrials.

What the market actually reacts to for THK: (1) the humanoid narrative (highest beta — it re-rated the multiple), (2) semiconductor-cycle order signals (the fundamental driver), (3) capital-return / ROE self-help. It reacts less to any single customer. The risk in that pattern: a multiple built substantially on theme #1 is vulnerable to theme rotation before the humanoid revenue is real.

---

Phase C — Judge people & books

Lens 9 · Management

Founder-family controlled, second generation in both top seats. Founder Hiroshi Teramachi (invented the LM guide, 1972) built THK from scratch. Today Akihiro Teramachi is Chairman & CEO and Takashi Teramachi is President & COO — family continuity, which in Japanese precision-manufacturing usually means long-horizon, engineering-led stewardship.

1. Track record: grew THK into the global LM-guide co-leader and a diversified motion-components house; navigated multiple machine-tool/semi down-cycles. The current regime's signature move is the 2025 portfolio reshape — exiting a chronically low-return automotive business to lift group ROE.
2. Tenure & skin in the game: Akihiro Teramachi personally owns 5.05% (6,014,000 shares) — a materially aligned, ~¥40B+ family stake at the current price. Top institutional holders are nominee trusts (Master Trust Bank ~16.2%, Custody Bank ~7.8%). Real founder-family ownership + alignment is a positive.
3. Capital allocation (the current story): a genuine, credible pivot to shareholder returns and capital efficiency — (a) auto divestiture explicitly "to boost capital efficiency and ROE"; (b) a buyback authorised for up to ¥40,000M / 20,000,000 shares (16.31% of capital) — share count already −7.34% YoY; (c) a ~30% payout-ratio dividend policy, DPS ¥184, yield ~2.5%; (d) a stated ROE >10% target versus 5.19% actual. Historically THK's ROE/ROIC have been low (5.2%/5.5%) — the whole equity story is management delivering the ROE step-up they're promising.
4. Red flags: none acute. The ¥81.6B liquidation loss is a cost of exit, disclosed and one-time, not an accounting surprise. Founder-family control can entrench, but here it's paired with an unusually shareholder-friendly agenda.
5. Archetype: founder-family, engineering-led, now overlaid with a capital-efficiency mandate — the good version of the "Japanese industrial waking up to ROE" trade.

Lens 10 · Forensic Red Flags

Regulatory findings (required sub-section). Per regulatory/regulatory-findings.md (Stage 1, 2026-07-10): THK has no SEC CIK — it is not an SEC registrant, so no EDGAR Litigation Releases or AAERs are searchable; total SEC findings = 0. Non-SEC web search ("THK" (FTC OR DOJ OR... settlement OR fine OR penalty) enforcement) returned no material enforcement actions against THK Co./6481.T. As a Japanese filer, the relevant regulator is the FSA/TSE, and no material securities-enforcement or accounting-restatement action was surfaced. Conclusion: no material regulatory or legal findings — verified via SEC EDGAR EFTS (n/a — no CIK), web search, and issuer disclosures as of 2026-07-10. (Caveat: no filings/ on the shelf means I could not read a 10-K/20-F Item-3 equivalent directly; TSE tanshin legal-proceedings disclosure was not fetched — flagged as an open item.)

Accounting-risk scan (from reported financials; all , none — no filings on disk):

• Discontinued-operations / disposal accounting (the live one): the ¥81,639M liquidation loss and the mechanics of transferring subsidiaries + loan receivables to an Advantage Partners-backed SPC (AP87) deserve scrutiny — did THK retain any vendor financing, guarantees, or earn-out/clawback exposure to the divested automotive entities? A PE buyer taking the business plus the loans can imply THK effectively funded its own exit. Watch the FY2026 cash-flow statement and any retained-interest disclosure.
• Earnings quality on the rebound: a Simply Wall St flag noted the Q1 FY2026 EPS surge raised "earnings-quality" questions — confirm the recovery is operating cash-backed, not disposal- or FX-flattered. With net income guided to double off ¥10B, the gap between GAAP recovery and cash generation is the thing to reconcile once the FY2026 cash-flow prints.
• FX / tariff translation: a global cost base + yen and US-tariff sensitivity means reported margins carry translation noise; not a red flag, a volatility source.
• SBC / goodwill: not a notable distortion for THK (a hard-asset manufacturer, minimal SBC vs US-tech; no large goodwill balance flagged).
• Working capital: classic for a cyclical component maker — inventory and receivables swing with the order cycle; watch inventory build if the semi recovery stalls.

Net: clean by forensic standards — the one genuinely non-trivial item is the structure and retained exposure of the automotive disposal, which I cannot fully verify web-only. Flagged, not alleged.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection

Bottom-up from the FY2026 guidance base (all ; guidance inputs ). Shares assumed to keep shrinking ~2–3%/yr on the buyback. No forecast.ts logged — this is an unattended --watchlist run (per skill, skip the Brier create step).

FY2026E (guidance-anchored): revenue ¥276B, OP ¥31B (11.2% margin), net ¥22.7B, shares ~112M → EPS ≈ ¥203.

FY2027E (base): semi/electronics recovery continues, mild machine-tool normalisation; revenue +7% to ≈¥295B, operating leverage lifts OM to ~12.5% → OP ≈¥37B, net ≈¥27B, shares ~109M → EPS ≈ ¥248.

FY2028E (base): revenue +6% to ≈¥313B, OM ~13% → OP ≈¥41B, net ≈¥30B, shares ~106M → EPS ≈ ¥283.

• Bull path (FY2028): semi supercycle + first real humanoid component volumes + margin mix → revenue ~¥345B, OM ~15% → EPS ≈ ¥370.
• Bear path (FY2028): semi capex pause + machine-tool rollover + humanoid stays negligible → revenue ~¥270B (flat), OM back to ~9% → EPS ≈ ¥165.

Valuation implication: at ¥7,688, the base FY2028E ¥283 EPS is ~27x three-years-out earnings; the bear case (¥165) is ~47x. Even the bull case (¥370) is ~21x FY28 — i.e. the stock already discounts the bull operating case. That is the crux: you are paying a full multiple on the optimistic earnings path, with the humanoid upside as an unpriced-if-it-works / overpriced-if-it-doesn't call option layered on top.

Lens 12 · Bull vs Bear

Bull case. THK is the irreplaceable precision layer of automation — a ~50-year process moat in a consolidated oligopoly, at the exact inflection where (a) the semiconductor/electronics capex cycle is recovering (orders above plan), (b) the auto divestiture structurally lifts group margin and ROE toward the >10% target, (c) management is shrinking the share count ~7%/yr and returning capital, and (d) humanoid robots — if they scale to Morgan Stanley's multi-trillion-dollar TAM — need exactly THK's ball screws, roller screws, cross-roller rings and linear actuators in every limb, plus THK has its own robot-hand/SEED business. A cleaner, higher-return, robotics-levered THK re-rates and compounds. Earnings doubling into FY2026 and expanding thereafter is the base, not the bull.

Bear case (2–3 permanent-impairment / de-rate risks).

1. The multiple, not the business, is the risk. ~3.4x book, ~20x EV/EBITDA, ~38x company-guided earnings, after a +99% one-year run, on a business earning 5% ROE. If the ROE step-up disappoints or the Street's ~¥38B net estimate proves ~65% too high versus ¥22.7B guidance, the de-rate is severe — this is a valuation/expectations impairment, the most probable one.
2. Humanoid is priced, not proven. Current humanoid component revenue is negligible (n/a — not disclosed, almost certainly <1–2% of sales). If humanoid timelines slip (Optimus/Figure are still pre-mass-production) or — worse — humanoids scale on cheap Chinese roller screws (Wuzhou at ~1/3 THK pricing, Beite at 2.6M sets/yr), THK captures share at commodity margins, not the premium the multiple assumes. The volume prize and the margin moat may be mutually exclusive.
3. Cyclicality unchanged, now less diversified. Post-auto, THK is more levered to the semi/machine-tool cycle. A capex pause takes revenue and the multiple down together (operating deleverage + de-rate).

Pre-mortem (18 months out, thesis broke): the semi order momentum that justified the FY2026 guidance raise faded by 2H26; humanoid remained a 2027+ story with no THK design-win of scale; the Street's above-guidance estimate got cut to ¥22.7B; and the humanoid theme rotated. THK fell from ¥7,700 toward book-value-plus (~¥4,000–5,000), a 35–45% drawdown, with the business barely changed — a pure multiple unwind.

Contrarian view (what the market refuses to see): the bulls treat "THK is in the Humanoid 100" as THK will earn humanoid economics. The overlooked truth is that THK's moat is a precision/share moat, not a cost/volume moat — the very thing humanoid mass-production commoditises. The market is paying a growth-at-scale multiple for a company whose edge is strongest exactly where scale isn't. Conversely, the under-appreciated real story is the boring one: a disciplined ROE self-help + buyback + semi-recovery that is worth owning — at a much lower price.

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

Dismantling the bull case. Where the money is made: ~90%+ of revenue is cyclical industrial components (LM guides, ball screws) into semi/machine-tool/general-industrial OEMs — a good but structurally low-ROE (5%), price-capped business (Hiwin sits 20–35% below on price at equal accuracy). The moat is weaker than bulls think on volume economics: it defends the precision niche, not commodity pricing. The most dangerous competitor bulls underestimate: not Hiwin — the Chinese roller-screw entrants (Wuzhou Spring, Beite) purpose-built for humanoid-scale volume at ~1/3 Japanese pricing; if humanoids standardise on "good enough at 1/3 the cost," THK's humanoid TAM is a margin-dilutive share war, not a premium annuity. Worst capital-allocation reading: THK transferred automotive subsidiaries and their loan receivables to a PE SPC and ate an ¥81.6B loss — a defensible clean-up, but confirm THK isn't retaining exit financing/guarantees. Assumptions that must hold for today's price: (1) net income roughly doubles to ¥22.7B in FY2026 and keeps climbing; (2) ROE actually reaches >10%; (3) humanoids become a real, premium-margin revenue line this decade. If growth disappoints 20–30% (semi pause, FY26 net ~¥16–18B not ¥22.7B), the stock is ~48–54x that number and de-rates hard. Single permanent-impairment scenario (and plausibility): humanoids scale on Chinese components while THK's core semi cycle simultaneously rolls over — moderate probability on any 2–3 year window, and it would collapse both the earnings and the multiple. Short verdict: not a fraud, not a broken business — a great company at a short-able price on expectations, best expressed as "fade the humanoid multiple," not "the business is bad."

Lens 14 · Management Questions (ordered by information value)

1. Your FY2026 net-income guidance is ¥22.7B, yet Street forward multiples imply ~¥38B — what specifically bridges that gap, and is guidance conservative or is the Street too high?
2. Quantify humanoid-robot-related revenue today (¥ and % of sales) and your FY2028/FY2030 targets — and name the design-win stage with the major humanoid programs.
3. When humanoids reach mass volume, do you expect to earn your historical precision margins, or will you meet Chinese roller-screw pricing? Which volumes do you walk away from?
4. Bridge me from 5.2% ROE to your >10% target: how much is auto-exit mix, how much buyback, how much operating margin — and by when?
5. On the automotive disposal: does THK retain any vendor financing, guarantees, earn-outs or residual exposure to the AP87/Advantage Partners entities?
6. Is the ¥40B / 20M-share buyback a one-off or the start of a standing return policy? What's the through-cycle payout + buyback framework?
7. What is current book-to-bill by region, and how much of the FY2026 order strength is semiconductor versus broad industrial — i.e. how cycle-dependent is the guidance?
8. Where can group operating margin structurally settle post-auto — is 11% a floor or a ceiling?
9. How do you defend LM-guide pricing against Hiwin's persistent 20–35% discount as its precision closes the gap?
10. What's the China strategy — Wuxi/Dalian localisation as offence, or hedge against domestic Chinese competitors moving up-market?
11. How large can SEED Solutions / the robot-hand business realistically become, and would you acquire to accelerate it?
12. Capex plans for FY2026–28 and the incremental ROIC you underwrite on new precision-component and roller-screw capacity?
13. What's the succession/governance plan given both top roles are held by the Teramachi family?
14. FX and US-tariff sensitivity: yen and tariff impact per given move on operating income?
15. If the semiconductor recovery stalls in 2H26, what's the downside-protection playbook for margins and returns?

---