Keyence

Phase A — Understand the business

Company Overview

Keyence sells the eyes, rulers, and nervous system of the automated factory — sensors, machine-vision systems, laser markers, measurement instruments, and digital microscopes — to roughly 350,000 customers across ~110 countries. It is, on the numbers, the most profitable hardware company on earth: FY2026 net sales ¥1.17 trillion (+10.4% YoY), operating income ¥595.7B (+8.4%) at a 51.0% operating margin, net income ¥445.1B (+11.7%) — the fifth consecutive record profit. Gross margin runs >80% (one source cites 83.0% ).

The business model is the moat, and it has two legs that reinforce each other:

1. Fabless manufacturing. Keyence does R&D, product planning, and final assembly/QC in-house but outsources ~90% of manufacturing to contract makers. That strips out the capital intensity that caps every other industrial's margin — capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. is a rounding error, so nearly every incremental yen of gross profit drops through. This is why a sensor company earns software-like margins.

2. Consultative direct sales. No distributors. Keyence hires engineering graduates and turns them into on-site consultants who walk the customer's factory floor, find the automation bottleneck the customer hadn't articulated, and spec the solution. ~3,000 sales staff (≈30% of ~10,000 employees) each file ~12 "Needs Cards" a year → ~36,000 fresh problem-statements annually that feed the product roadmap. The salesforce is simultaneously the distribution channel and the market-research org.

Contract structure: transactional, not recurring — Keyence sells hardware (10,000+ SKUs) with a same-day/next-day delivery promise it funds by carrying ~163 days of inventory. There is no big take-or-pay backlog and no single-customer concentration disclosed; revenue is spread across automotive, semiconductor, electronics, food/pharma, and general manufacturing. That breadth is a feature — no one end-market can sink it.

Reports as a single segment in statutory filings — it does not break out product-line P&L, which is itself a (deliberate) disclosure choice (see Lens 10).

Supply Chain

Keyence sits in the middle of the automation value chain and has engineered itself to be the low-capital, high-margin node:

Upstream (inputs → Keyence):

• Semiconductors / image sensors / lasers / optics — CMOS sensors, laser diodes, FPGAs, MCUs sourced from the broad electronics supply base. No single named critical supplier is publicly disclosed; the fabless model means Keyence is a buyer of chips and modules, not a fab operator.
• Contract manufacturers (EMS/ODM) — ~90% of physical production is outsourced. Keyence keeps final assembly, calibration, and QC. Specific CM partners are n/a — not disclosed (a governance opacity point, Lens 10). The Takatsuki logistics hub anchors distribution.

Keyence (the value-add node): product design, firmware/algorithms (increasingly AI-based defect detection), the applications-engineering know-how embedded in the salesforce, and the just-in-time inventory buffer.

Downstream (Keyence → end customer): sold direct to factory floors — automotive OEMs and Tier-1s, semiconductor and electronics fabs, food/beverage, pharma, logistics. No distributor layer to skim margin or dilute the customer relationship.

Chokepoints / single-source dependencies: the honest read is that Keyence's supply chain is deliberately un-concentrated on the input side (commodity electronics, multiple CMs) — its scarcity is on the output side (the salesforce and the accumulated applications library). The one genuine exposure is the same one every electronics buyer carries: an acute chip/optics shortage would pinch product availability, but Keyence's pricing power and inventory buffer cushion it better than peers.

Commercial-layer note: the robotics wiki files (positioning.md, supply-chain.md, market-structure.md) are pointers in the research index but resolve to empty on disk — no compiled supply-chain map to lean on, so this lens is web-derived.

Competitive Advantages (moats)

Keyence's moat is not one thing — it's the compounding interaction of four, which is precisely why it has been un-copied for 40 years:

1. The consultative salesforce (the real moat). Rivals sell products; Keyence sells solutions found on your factory floor. Switching away means giving up the engineer who knows your line. High switching costs, and a data flywheel: 36,000 Needs Cards/year → products that hit unmet needs → ~70% of new products carry a "world-first / industry-first" feature. That is a moat you cannot buy — you'd have to rebuild a 3,000-engineer culture.

2. Fabless economics as a weapon. ~50% operating margins mean Keyence can out-invest, out-price, and out-wait any competitor while still earning more. It can afford to sell a standardized premium product at a price a bespoke integrator can't match on ROI.

3. Pricing power / bargaining position. Over customers: Keyence wins because it quantifies ROI ("this sensor pays back in X months"), so it rarely competes on price. Over suppliers: as a fabless buyer of commodity electronics spread across many CMs, it holds the whip hand upstream. Gross margin >80% is the receipt.

4. The R&D paradox. Keyence spends only ~2.3% of revenue on R&D — versus Cognex ~14%, FANUC ~6.8%, Omron ~5.8%, Rockwell ~5.7%. It out-innovates on a fraction of the budget because the salesforce does the discovery — R&D is aimed, not speculative. Rivals spend 3–6x more and grow slower.

The scoreboard: FY2012–2023 revenue 15% CAGR vs. Cognex 10.9%, SMC 7.6%, Yaskawa 5.5%, FANUC 4.3%, Omron 3.2%, Rockwell 2.4%. Keyence has out-compounded every automation peer, overseas revenue growing ~22% CAGR — all organic.

Where the moat is thinnest: in commodity sensor/vision categories, emerging Chinese vendors are undercutting on price by 30–50%. Keyence's defense is to stay ahead on the solution layer, not the component — but this is the edge bulls should watch.

Segments

Keyence reports as a single statutory segment and does not publish product-line P&L — so any product-family split is n/a at the revenue-and-margin level. our figures on the shelf is empty (headers only). What is knowable:

By product family (qualitative, no P&L split disclosed): five groups — (1) Sensors (the core), (2) Measurement systems & microscopes, (3) Machine vision (described as a global 3D-vision leader), (4) Marking & printing (laser markers, traceability/code-reading), (5) Process control (PLCs, servos, HMIs). Machine vision and measurement are the higher-growth, higher-margin end; sensors are the volume base.

By geography (the split that is disclosed):

• Overseas ≈ 60% of revenue, all organic, up from ~48% a few years ago — the growth engine.
• FY2024 mix cited as Japan ~36%, USA ~18%, China ~15%, rest-of-world the remainder.
• China exposure is the swing factor. Direct China ~15%, but broader "Greater China + China-linked demand" is cited as high as ~30% of revenue — the single most important cyclical variable (Lens 12/13).

Regional trend (direction & cause): FY2024 quarterly YoY — Japan +8.5%, Americas +12.6%, Asia +17.7% (China recovering), but Europe −6.5%. FY2026 growth (+10.4%) was led by North America and Asia, specifically automotive and semiconductor capex. Read: the Americas + Asia semiconductor/automation build is the accelerant; Europe is the drag; China is the wildcard that can swing the whole print.

---

Phase B — Measure performance

Earnings Result (FY2026, year ended March 2026)

The latest full-year print was another record, with one landmark strategic addition.

• Net sales ¥1.17T, +10.4% YoY — first time comfortably clear of the ¥1T "sound barrier," reaccelerating after a soft FY2024.
• Operating income ¥595.7B, +8.4%; operating margin 51.0% — note the margin ticked down from ~51.9% (FY2025) as opex (salesforce expansion, overseas build-out) grew slightly faster than sales. Not alarming, but the +8.4% OP vs. +10.4% sales gap is the one line a bull should keep an eye on.
• Net income ¥445.1B, +11.7% — 5th straight record; net margin ~38%.
• ROE 13.5% — artificially depressed by the cash mountain (see Lens 9); operating ROIC ex-cash is an order of magnitude higher.
• Q4 was the fireworks: operating profit ¥179.4B vs. ¥163.8B consensus, and the stock jumped 16% to its daily limit on the beat.

Balance sheet (fortress): total assets ¥3.67T, equity ratio 94.6% — i.e. almost no debt. Marketable + investment securities ¥2.38T (gov't bonds ¥600B, corporate bonds ¥864B, CDs ¥891B, plus deposits/equities) on top of ~¥451B cash equivalents — so ~¥2.8–3.0T of the ¥3.67T asset base is liquid financial assets. Working capital is deliberately loose: ~163 days inventory (funds the delivery promise), ~108 days receivables (generous terms to small customers). No inventory/receivable quality flags — these are strategy, not stress.

Guidance/outlook: the ¥350→¥550 dividend hike (payout to 30%) was signaled to be maintained into next year, and management moved to amend the charter to enable buybacks (Lens 9) — the tone shifted from "hoard" to "return." Forward company guidance figures: n/a in parseable form.

Vs. its own history: the standout unusual item is not in the P&L — it's the capital-allocation pivot. For a company whose entire bear case was "dead money on the balance sheet," FY2026 is the year that thesis started to break in the bulls' favor.

Earnings Calls (sentiment trend)

No transcripts on the shelf (transcripts=0), and Keyence is famously low-touch with investors — sparse guidance, minimal management access, no IR theatre. Sentiment must be read off actions and results commentary rather than call tone.

The trend across the last several prints (FY2024 → FY2026):

• FY2024: defensive — revenue declined ~6% in USD terms, Europe weak (−6.5%), China soft; management stayed quiet and let the model absorb the down-cycle. Message: "we compound through cycles."
• FY2025: reacceleration — back above ¥1.06T, +9.5%, led by North America and Asia semiconductor/automotive.
• FY2026: confidence — record profit #5, and the language of the announcement changed. The dividend near-doubling and the buyback-charter move are, for this management, a loud statement. The thing they stopped implying is "cash stays idle indefinitely."

The durable "recurring phrases" are structural: added-value, world-first products, sales-network strength, overseas expansion. The new note is shareholder return — a genuine tonal shift for a company that spent decades saying nothing about capital.

Comps

Machine-vision / factory-automation peer set. Multiples are with source/date or n/a. No multiple is fabricated.

Dividend yield: Keyence ~0.7% at the ¥550 dividend on a ~¥63–80k share price — negligible; this is a growth-compounder, not an income name. 5-yr avg ROE: mid-teens (13–15%), cash-suppressed.

Read: Keyence and Cognex trade at statistically identical forward multiples (~42–43x), and FANUC isn't far behind (~39x) — the whole quality-automation complex is priced for perfection. But Keyence earns a ~50% operating margin vs. Cognex's teens and FANUC's ~20s, grows faster and more consistently, and carries a net-cash fortress. On quality-adjusted terms Keyence is the cheapest expensive stock in the group — you pay ~43x for a business that genuinely deserves a premium. Historical P/E band: 10-yr avg ~30–35x, low ~20x (2016), high ~68x (2021) — so ~43x sits above mid-cycle but well below the 2021 bubble peak.

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

The pattern is unusually clean: Keyence trades on the global automation/semiconductor capex cycle and on earnings surprises — not on company-specific drama.

• 2021 (peak, ~68x P/E): rode the post-COVID automation + semiconductor super-cycle to an all-time high; briefly made founder Takizaki Japan's richest man.
• 2022–2023 (de-rate): global rate shock + automation down-cycle compressed the multiple hard from the 2021 peak; the business held up far better than the stock (revenue kept growing) — a valuation event, not a fundamental one.
• FY2024 (soft patch): USD revenue −6%, Europe −6.5%, China weak → stock pressured on the automation slowdown.
• 27 Apr 2026 (+16%, daily limit): Q4 OP ¥179.4B crushed ¥163.8B consensus on strong overseas sales — the single biggest one-day move, and it paired the beat with the dividend hike + buyback-charter news. The market reacted violently upward because two things it wanted — a beat and capital return — arrived together.
• 52-week range (ADR KYCCF) ~$303–$553 → an ~80% high-to-low swing, i.e. the name is volatile despite the fortress balance sheet, because it's a high-multiple cyclical.

What the market actually reacts to: (1) earnings surprises vs. consensus (semiconductor/auto capex is the driver), (2) China/Asia demand signals, and — newly — (3) capital-return news. The April 2026 reaction proves the market has been waiting for Keyence to address the cash; every step toward returning it is now a catalyst.

---

Phase C — Judge people & books

Management

• Founder: Takemitsu Takizaki (b. 1945), founded Keyence 1974, stepped back to Honorary Chairman in 2015. Net worth ~$16–19B — was briefly Japan's richest person (2021). Still controls ~18% directly + through holding company TT K.K., with a related asset-management vehicle holding another ~17%. Founder DNA is intact even without founder in the chair.
• CEO/President: transitioned to Yu Nakata (President & Representative Director, ~2025) from Akinori Yamamoto (President 2010–~2025). (Note: source variation on exact current title/name — treat as, medium confidence.) Keyence's model is deliberately CEO-light: the system (Needs Cards, salesforce discipline, fabless economics) runs the company, not a charismatic operator. Value Punks called the top job a "revolving door" (3+ CEOs since 2000) — and it hasn't mattered, which is itself the point.

(1) Track record: the machine speaks — 15% revenue CAGR, ~50% margins, 5 straight record profits, all-organic overseas expansion. Best-in-class execution, sustained across cycles and CEOs.

(2) Tenure & skin in the game: founder family ~35% combined economic interest via direct + vehicles — enormous alignment. Insider-transactions file n/a — not on shelf.

(3) Capital-allocation history — the crux, and it's changing:

• Historically the one blemish. Keyence hoarded. Dividend payout was 6% of net income in 2018, crawling to ~20%, described as "immaterial". It built a ~¥2.4T securities / ~¥2.8–3T total liquid pile earning near-nothing while the operating business earned ~100% ROIC — a textbook value-destroying idle-cash drag that suppressed group ROE to the mid-teens.
• FY2026 = the inflection. Dividend near-doubled ¥350→¥550 (payout to 30%), signaled to hold; and management moved to amend the charter to enable buybacks. Under TSE's "cost-of-capital-conscious management" reform (the ¥/P-B campaign pushing Japanese firms to return capital), Keyence is finally addressing its single structural weakness. This is the most important development in the whole dossier.

(4) Red flags: the 2016 gift-tax affair — founder's elder son failed to report ~¥150B in gift tax on shares of the family asset-management firm (which holds ~17% of Keyence), settling ~¥30B in back taxes. It's a family-succession/tax-structuring matter, not corporate fraud — but it flags aggressive family-wealth engineering and the opacity of the ownership vehicles. Separately, the salesforce culture draws heavy criticism (relentless cold-calling, extreme quotas, "customers hate the daily calls") — a reputational/retention risk, not an accounting one.

(5) Archetype: founder-built, now system-run — the rare case where the founder's genius was encoded into a process durable enough to outlast him. That's the bull's core comfort and the reason CEO churn doesn't scare the stock.

Forensic Red Flags

Forensic-analyst lens. Every figure /; no SEC filings exist to cross-check, which is itself a limitation — treat conclusions as medium-confidence.

Income statement: margins are so high they invite skepticism, but ~50% operating / >80% gross has held for years across independent sources and cycles — consistency argues against a one-off distortion. Revenue recognition is point-of-sale hardware (clean, low judgment). No non-GAAP/SBC games of the US-tech kind — this is Japanese-GAAP hardware; stock-based comp is not a material margin flatterer here (a genuine positive vs. the SBC-inflated US comps). The one thing to watch: OP grew +8.4% vs. sales +10.4% in FY2026 — mild opex deleverage, worth monitoring but immaterial.

Balance sheet: the flag is not fraud — it's capital inefficiency. ~¥2.4T in securities + ~¥451B cash (94.6% equity ratio) is a fortress that was, until FY2026, a value trap — a mountain of low-return financial assets depressing ROE. Receivables (~108 days) and inventory (~163 days) run high by deliberate strategy (small-customer terms + the delivery promise), not deterioration — but a forensic analyst notes that generous receivables to small customers can mask channel-stuffing risk in a downturn; no evidence of it, but it's the place to look if growth ever stalls. Goodwill/intangibles are negligible (organic growth, ~no M&A) — a clean balance sheet with no acquisition-accounting risk.

Cash flow: FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. conversion ~80–90% of net income — cash earnings track book earnings well; no divergence between earnings and cash (the classic red flag) is absent. Capex is trivial (fabless), so FCFFree cash flowCash left after paying to run and maintain the business. Unlike profit, it is hard to flatter with accounting choices. ≈ operating cash flow ≈ net income — high quality of earnings.

Segment/disclosure opacity: Keyence reports one segment and does not name its contract manufacturers or break out product-line economics. This is legal under J-GAAP and consistent with its secretive culture, but it reduces external verifiability — you cannot independently check where the margin comes from by product. A forensic minus for transparency, not for integrity.

Regulatory findings (required sub-section):

• SEC (EDGAR LR/AAER): none possible — Keyence has no CIK and is not an SEC filer. regulatory/regulatory-findings.md (fetched 2026-07-06) confirms total_sec_findings: 0, noting "Keyence has no CIK … no EDGAR enforcement search is possible".
• Non-SEC / web ("Keyence" (FTC OR DOJ OR FDA OR consent decree OR fine OR penalty) enforcement): no material antitrust, product-safety, or enforcement actions surfaced. The only material legal/tax matter is the 2016 family gift-tax settlement (~¥150B assessed, ~¥30B paid) — a shareholder/family tax matter, not a corporate enforcement action against Keyence Corp.
• Item 3 Legal Proceedings: n/a — no 10-K exists (non-US filer).
• Verdict: No material corporate regulatory or fraud findings — verified via the (empty) SEC EDGAR EFTS pull and web search as of 2026-07-06. The standing flags are governance/disclosure opacity and family-wealth structuring, not accounting fraud.

---

Phase D — Project & stress-test

Forward Projection

Built bottom-up from FY2026 actuals (net income ¥445.1B, ~243M shares, EPS ≈ ¥1,830 ). All output lines with arithmetic; no consensus EPS was cleanly sourceable, so this is a model, not a print.

Assumptions (labeled):

• Machine-vision/automation TAM +8.3% CAGR to 2030; Keyence historically grows ~1.5–2x the market via share gains.
• Margins ~stable at ~50–51% (slight opex deleverage offset by mix toward vision/measurement).
• DilutionDilutionIssuing new shares, so each existing share owns a smaller slice of the same company. ~0; buybacks now a tailwind to per-share figures if the charter amendment passes.

. ~3-year base EPS CAGR ~8–9%; at ~43x that leaves little valuation cushion — the stock needs the bull path (or multiple support from capital return) to make double-digit forward returns.

Brier forecast: skipped create step (unattended --watchlist rule — only log when genuinely committed to the base case). Would-be forecast for the record: "6861.T FY27 (Mar-2027) net income ≥ ¥480B, p≈0.60, resolves 2027-05-31" — logged here as a note, not written to our model.

Bull vs Bear

Bull case. You are buying the single best hardware business model ever assembled, and its one flaw is finally being fixed. The consultative-salesforce + fabless-economics moat has out-compounded every automation peer for 40 years at ~50% margins and ~100% operating ROIC, and it survives CEO turnover because it's a system, not a person. Secular tailwinds — reshoring, factory automation, semiconductor capex, labor scarcity, AI-based inspection — all pull Keyence's way, and it grows ~2x the market via share gains. The FY2026 catalyst is real and repricing-worthy: the dividend near-doubled and buybacks are coming, so the ~¥2.8T of dead cash that suppressed ROE to the mid-teens is about to start working — a mechanical boost to ROE, EPS, and the multiple, exactly what TSE governance reform is engineering across Japan. Earnings-surprise history is a positive skew (8+ consecutive beats, +16% on the last one). This is a compounder you hold for a decade.

Bear case (permanent-impairment risks).

1. China (~30% of demand) structurally decelerates. A prolonged Chinese industrial slump plus domestic Chinese vision/sensor champions undercutting by 30–50% in commodity tiers could permanently reset Keyence's growth and its pricing power in its biggest growth market. This is the crack.
2. Valuation compression. ~43x forward with ~8–9% base EPS growth = priced for perfection. The 2022–23 de-rate showed the stock can lose a third of its value on no fundamental deterioration — a multiple event. If growth merely normalizes, the multiple does the damage.
3. The moat's commodity edge erodes. If AI-based vision becomes "good enough" as a cheap software/commodity layer, Keyence's premium-hardware ROI pitch weakens at the low end.

Pre-mortem (18 months out, thesis broke — what happened?): China demand rolled over hard, a global automation capex freeze hit orders (peers see 30–40% order drops in a quarter in true down-cycles), Keyence printed a flat-to-down year for the first time in a while, and the ~43x multiple compressed toward its ~20x floor — a 40–50% drawdown with the business still intact. The killer isn't the model; it's cyclicality meeting a rich multiple.

Are multiples too high? Above mid-cycle (~43x vs. ~30–35x 10-yr avg), but defensible given the margin/ROIC quality and the capital-return catalyst. Not a bubble (2021 was 68x). The risk is timing, not overvaluation-to-the-point-of-absurdity.

Contrarian view (what the market refuses to see): the consensus fixates on "expensive + China risk." What's under-appreciated is the capital-allocation regime change: a company that returned 6% of profit in 2018 and 20% recently just went to 30% and opened the door to buybacks. If Keyence follows the TSE-reform playbook to its conclusion, the ¥2.8T cash is a loaded spring — years of buybacks that mechanically lift ROE from the mid-teens toward the 20s and re-rate the multiple. The market is pricing the cash as dead; it's about to come alive.

Devil's Advocate (short-seller)

Skeptical short-seller dismantling the bull case.

• What structurally breaks the money machine? Keyence's margin is a function of pricing power in a premium niche. The most dangerous scenario isn't a cyclical dip — it's commoditization from below: Chinese vision/sensor vendors (backed by state industrial policy) climbing the value chain at 30–50% lower prices, turning Keyence's "world-first feature" premium into a shrinking island. In China specifically — its single biggest growth market — "buy domestic" policy is a structural, not cyclical, headwind.
• Revenue concentration: not customer-concentrated, but geography-concentrated on the cycle — ~30% China-linked, heavy semiconductor/auto capex exposure. Both are among the most volatile demand pools in the global economy. A synchronized China + semicap downturn hits the growth and the multiple at once.
• Why the moat is weaker than bulls think: the salesforce moat is a culture — and cultures are fragile to scale and geography. The relentless-cold-call, extreme-quota model that works in Japan draws open customer hostility abroad ("customers hate the daily calls" ); replicating 3,000 elite engineer-consultants across 110 countries as the company scales is not guaranteed, and salesforce burnout/turnover is a real degradation vector.
• Most dangerous competitor bulls underestimate: not Cognex (same premium game, similar multiple) — it's the Chinese long tail plus AI-vision-as-software players who could make "good-enough" inspection cheap and undercut the hardware-ROI pitch at the volume base.
• Worst capital-allocation history: a decade-plus of hoarding ¥2T+ at near-zero return while earning ~100% operating ROIC — a colossal, quantifiable value-destruction that only now is being addressed, and only under external (TSE) pressure. Plus the 2016 gift-tax affair and opaque family holding vehicles — aggressive family-wealth structuring and disclosure opacity (single segment, unnamed CMs). Governance-minded investors should not wave this away.
• Assumptions that must hold for today's price: sustained ~8–10% growth, ~50% margins intact despite Chinese price pressure, China recovering, and the buyback promise actually executing. That's a lot of "and."
• Valuation if growth disappoints 20–30%: at ~43x, a growth scare that resets the multiple toward the ~20x historical floor is a ~40–50% drawdown even with earnings only modestly lower — the 2022–23 episode is the template.
• Single scenario that permanently impairs: Chinese domestic champions + AI-commodity vision permanently cap Keyence out of its biggest growth market and compress its premium — turning a 15%-CAGR compounder into a low-single-digit, ex-growth industrial that de-rates from ~43x to ~15–20x. Plausibility: moderate — the low-end erosion is real and observable; the complete impairment is not the base case but is the tail worth respecting.

Management Questions (ordered by information value)

1. The charter amendment enables buybacks — what is the actual capital-return policy and target payout/net-cash ratio? Will you commit to a multi-year buyback that draws the ¥2.4T securities pile down, or is this optionality you may never use? (This answer most changes the thesis — it's the entire capital-allocation crux.)
2. In China (~30% of demand), how much of recent softness is cyclical vs. structural "buy-domestic" policy, and how are you defending share against Chinese vendors pricing 30–50% below you?
3. Where are AI-based/software vision approaches most threatening to your premium-hardware ROI pitch, and how does your software layer defend the value base — not just the frontier?
4. Operating profit grew slower than sales in FY2026 — is ~50% the durable margin floor, or should we model gradual deleverage as you scale the overseas salesforce?
5. Can the consultative-salesforce culture be replicated at full fidelity across 110 countries as you scale, and what is engineer-consultant turnover outside Japan?
6. What is the right net-cash level for this business? Why hold ¥2.8T of low-return financial assets against a ~100% operating-ROIC core?
7. How do you sustain ~70% "world-first" new-product hit rate on just ~2.3% of sales in R&D as the SKU base and TAM expand?
8. Which product family (vision, measurement, microscopy, marking, sensors) has the best 5-year growth/margin outlook, and why won't you disclose product-line economics?
9. What's the succession/development plan for the engineer-consultant salesforce given documented burnout and customer-relationship friction?
10. How exposed is the model to a synchronized semiconductor + automotive capex down-cycle, and what did FY2024 teach you about the trough?
11. Reshoring/friend-shoring — quantify the incremental automation demand you expect in the US/India/SE Asia over five years.
12. How do you think about M&A now that you're returning capital — is inorganic growth permanently off the table, and why?
13. What governance changes (beyond buybacks) are you making under the TSE cost-of-capital initiative — board independence, disclosure, the family holding vehicles?
14. Contract-manufacturer concentration and geographic footprint — how resilient is the fabless supply chain to a chip/optics shortage or a Taiwan/China disruption?
15. If you had to name the one thing most likely to end the 15% compounding, what would it be?

---