Yaskawa Electric

Phase A — Understand the business

Company Overview

Yaskawa is the motion company — the one member of the robotics "Big Four" whose robots are a downstream expression of a deeper competence: making the servo motors, amplifiers and drives that turn electricity into precise, controllable movement. Founded in 1915 in Kitakyushu, it invented the word "mechatronics" (1969) and today frames its whole strategy as i³-Mechatronics — integrated, intelligent, innovative motion.

What it actually does — four reporting segments (year ended 28-Feb-2026):

1. Motion Control — ¥236.1B revenue, ¥24.4B OP, 10.3% margin (~44% of revenue). Two businesses: AC servo motors + controllers and the drives/inverter (AC drives) business. This is the crown jewel and the highest-quality earnings stream. Yaskawa holds the leading global share in AC servo motors and is one of the few makers vertically integrated from motor to drive to controller. The servo goes inside other people's machines — semiconductor wafer-handling robots, chip-making tools, machine tools, electronic-component assembly — and the inverter/drive goes into HVAC (notably US data-center air-conditioning) and solar power conditioners.
2. Robotics — ¥247.0B revenue, ¥20.4B OP, 8.3% margin (~46% of revenue). The Motoman industrial-robot line — handling, welding, assembly, painting — plus collaborative robots and the new Motoman NEXT adaptive/AI platform. Yaskawa is ~12% of the global industrial-robot market, the #4 of the Big Four behind FANUC (~17%), ABB (~13%) and KUKA (~13%). Note the tell in the numbers: Robotics is the larger segment by revenue but the lower-margin one (8.3% vs Motion Control's 10.3%) — the opposite of FANUC, whose robot business rides on a CNC-monopoly cash engine.
3. System Engineering — ¥38.7B revenue, ¥5.0B OP, 12.9% margin (~7%). Large-scale drive systems and social-infrastructure projects (steel-plant drives, environmental/energy systems) — the highest-margin segment, project-lumpy, and the beneficiary of the large steel-plant project cited in the FY order surge.
4. Other — ¥20.3B revenue, ¥2.0B OP (~4%). Logistics, environmental/energy, and misc.

(Segment OP sums to ¥51.8B vs the ¥47.3B consolidated operating profit — the ~¥4.5B gap is corporate/eliminations.)

Customers / suppliers / competitors. Customers: automotive OEMs and Tier-1s (large China/Asia auto lines were the FY robot driver), semiconductor-equipment makers (Yaskawa servos inside wafer handlers — the AI-capex tailwind), electronics/component makers, machine-tool builders, HVAC/data-center, and solar. Suppliers: precision reduction gears (Harmonic Drive, Nabtesco) for robot joints; semiconductors/power devices; and, newly, NVIDIA edge compute for Motoman NEXT. Competitors: the other Big Four (FANUC, ABB, KUKA) in robots; Inovance, Mitsubishi Electric, Delta in servo/drives; Estun/Inovance as the rising Chinese domestics. Contract structure is transactional + cyclical (unit sales and project contracts, no take-or-pay), with a maintenance/parts annuity under the installed base — but Yaskawa's recurring layer is thinner and less disclosed than FANUC's service annuity.

The one-line business model: Yaskawa sells precise motion — the servos and drives that move the world's machines, and the Motoman robots built on them — and is now betting its next decade on fusing that motion hardware with AI ("Physical AI") to defend margin against Chinese servo/robot commoditization.

Supply Chain

Map: upstream inputs → Yaskawa → machine builders / integrators / OEMs → end factory.

Upstream (into Yaskawa):

• Precision reduction gears / harmonic drives — the joint-torque chokepoint every Big-Four robot maker shares. Named suppliers: Harmonic Drive Systems (6324.T) and Nabtesco (6268.T).
• Servo motors, amplifiers, controllers — made in-house. This is the structural difference from KUKA/ABB, who buy motion control on the open market: Yaskawa (like FANUC and Estun) is vertically integrated into its own servos/drives, a permanent BOM cost edge.
• Edge AI compute — NVIDIA Jetson Orin autonomous control units + NVIDIA Isaac libraries/models, running on Wind River Linux, are the brains of Motoman NEXT. A new, concentrated single-vendor dependency shared across the whole robotics industry.
• Humanoid actuators — via the Tokyo Robotics acquisition (2026), Yaskawa is internalizing high-performance actuator tech it "previously lacked" for humanoids and Motoman NEXT expansion.
• Semiconductors/power devices, magnets, steel, castings, cabling — commodity, multi-sourced.

Midstream (Yaskawa itself): HQ Kitakyushu, Japan; manufacturing in Japan plus China (Changzhou robot plant), the US (Motoman, Ohio), and Europe (Slovenia) — more geographically distributed than FANUC's single Oshino complex, which cuts single-site earthquake risk but also dilutes the lights-out cost advantage.

Downstream (out of Yaskawa):

• Semiconductor-equipment makers — Yaskawa servos/robots go into wafer-handling and chip tools; this is the node driving the FY2027 Motion Control acceleration (AI-capex).
• Auto OEMs & Tier-1s — the China/Asia auto lines that carried FY robot revenue (but at "low value-added," hence the margin drop).
• Machine-tool builders, electronics/component makers, system integrators, HVAC/data-center, solar.
• The industry itself — Yaskawa is a merchant servo supplier to rivals' robots; this is a two-edged node — revenue today, but it means Yaskawa's servo share IS the battleground Inovance is winning at home in China (Lens 3/13).

Chokepoints & single-source risk:

Names or it didn't happen — delivered: Harmonic Drive, Nabtesco, NVIDIA (Jetson/Isaac), Wind River, Tokyo Robotics, Changzhou/Ohio/Slovenia plants, auto + semiconductor-equipment end-markets, Inovance/Estun as the domestic-China attackers. The vulnerability is the same as FANUC's, one segment over: Yaskawa's own servo leverage is precisely what Inovance is dismantling in the fastest-growing market.

Competitive Advantages (moats)

Yaskawa has a real but narrower and shallower moat than FANUC's — and it is being tested at its core, not just its flank.

The moats, ranked by durability:

1. Servo/motion-control technology + vertical integration (the real moat). A century of servo know-how, global-leading AC-servo share, and in-house motor→drive→controller integration give Yaskawa a genuine precision-and-cost edge over the buy-it-in players (KUKA/ABB). This is the crown jewel — but see the soft flank.
2. Big-Four robot brand + Motoman installed base. ~12% global share, decades of automotive-line reliability, a global service/integration footprint. Switching costs exist (programming, integration, spares) but are lower than FANUC's CNC lock-in — a robot arm is more swappable than a shop-floor CNC standard.
3. i³-Mechatronics data/software layer + Physical AI (the bet, not yet a moat). Motoman NEXT (NVIDIA Jetson/Isaac, adaptive/autonomous) + the Tokyo Robotics humanoid actuator play aim to move Yaskawa up the value stack from hardware to intelligent motion — the explicit Vision 2035 thesis. Optionality, not yet a defended position.
4. IFRS reporting + improving governance — a cleaner, more investable disclosure regime than FANUC's J-GAAP, and an outside-director-led nomination committee (Lens 9). A soft "trust" moat.

Bargaining power: Moderate. Strong in high-precision servo (semiconductor tools need the best), weaker in commodity robots and — critically — weak and weakening in China servo/drives, where Inovance now out-shares it at home.

The moat's soft flank (the thing that matters, and it's worse than FANUC's): FANUC's moat is attacked at its volume flank (mid-market robots) while its crown (CNC) stays safe. Yaskawa's crown itself — servo/motion in China — is under direct assault. Inovance holds >26% of the Chinese general-servo market, ranking #1, "often ahead of global peers like Yaskawa and Mitsubishi" on their home turf, growing revenue ~22% YoY (2025 rev ~RMB 42B) into a 6% sector. Chinese industrial policy explicitly favors local automation. So Yaskawa's deepest moat is being eroded in the largest servo market — a more dangerous setup than FANUC's, where the monopoly asset is untouched.

Segments

Full-year segment marks, year ended 28-Feb-2026:

First-half detail (for the intra-year trend): MC 1H ¥112.8B (−5.5%), OP ¥12.0B (+9.2%), 10.7% margin ↑; Robotics 1H ¥119.2B (+6.4%), OP ¥10.5B (−0.5%), 8.8% margin ↓; SysEng 1H ¥18.7B, OP ¥1.9B. The pattern is consistent all year: Motion Control shrinking-but-fattening, Robotics growing-but-thinning.

Geography (year ended Feb-2026): Japan ~27% of revenue, overseas ~73%. Within Motion Control, Americas ~38%, China ~28%. Regional revenue-growth directions (1H): Japan +2.2%, China +2.6%, Asia-ex-China +3.4%, Americas −6.0%, Europe −4.5% — i.e. Asia (incl. China) the growth pole, the Americas/Europe the drag. Exact full-company regional revenue percentages beyond Japan-vs-overseas: n/a to a clean company-wide table.

The segment story in one line: Yaskawa is a servo company whose profit engine (Motion Control) is quietly improving on AI/semi mix, wrapped around a robot business (its largest segment) that is growing revenue but destroying margin to chase low-value China auto volume — the exact inverse of a quality-compounding mix.

---

Phase B — Measure performance

Earnings Result

The full-year print, year ended 28-Feb-2026 (announced 2026-04-10):

• Revenue ¥542.1B, +0.8% YoY.
• Operating profit ¥47.3B, −5.7% YoY (8.7% margin, down from ~9.3%).
• Net profit ¥35.2B, −38.2% YoY.
• EPS (ttm) ¥135.71.

Read the −38% net drop correctly — it is mostly a one-off, not operating collapse. The prior year included a ~¥26.7B gain from an equity/share sale that did not repeat; strip it and the underlying decline is the −5.7% operating move, itself driven by (a) FX, (b) rising indirect costs, and (c) the low-value-added China robot project mix. This matters for the multiple: the trailing P/E (~51x) is computed on depressed, one-off-suppressed earnings, so it overstates richness somewhat — but even the forward P/E (~34x) on the recovered ¥47B guide is a full multiple (Lens 7).

Vs its own guidance: Yaskawa had raised its full-year guide at 1H to ¥525.0B rev / ¥48.0B OP (from ¥515B/¥43B). Actual revenue beat (¥542.1B) but OP came in slightly below the raised ¥48.0B bar (¥47.3B) — a revenue-beat / margin-miss finish.

Guidance — the year ending 28-Feb-2027 (Yaskawa "FY2026", year one of Dash 35):

• Revenue ¥580B (+7.0%), Operating profit ¥60B (+26.8%), Net profit ¥47B (+33%).
• Dividend raised to ¥72/share (from ¥66–68; sources differ on the prior base).
• FX assumptions: ¥145/USD, ¥170/EUR, ¥20.5/CNY.
• The driver: management guides ~+19% Motion Control growth on AI/semiconductor-equipment demand, plus AI-robot monetization and the steel-plant project. The bull's whole case is that FY2027 is the earnings-bottom inflection ("turnaround to profit growth for the first time in four terms, driven by AI robots").

The Q4 tell: orders surged +20% YoY / +10% QoQ in the Dec-2025–Feb-2026 quarter on "expanding AI and semiconductor-related demand" + the steel-plant order. Order momentum is the leading indicator the guidance leans on.

Balance-sheet flags (NOT a FANUC fortress):

• Net DEBT of ¥63.6B (−¥245/share net cash, i.e. negative), D/E 0.25 — modest leverage, but a real contrast to FANUC's zero-debt / ¥603B-net-cash fortress. Current ratio 2.47 (liquid).
• Free cash flow just ¥5.93B against ¥35.2B net income — poor cash conversion this year (working-capital build in the cycle); P/FCF ~305x — a genuine flag that the cash isn't yet backing the earnings.
• ROE 7.84%, ROIC 6.17% — below a reasonable cost of equity; the business is currently earning sub-economic returns (Lens 9/13).

Market reaction: Stock +4.7% on the print (guidance beat expectations), and +138.8% over the trailing year into the ¥7,915 all-time high (22-Jun-2026). Translation: the market has already priced the FY2027 recovery and the Dash-35/Physical-AI story — the print rewarded the narrative, not a beat on the actual (weaker) year.

Earnings Calls (sentiment trend)

No transcripts on the research shelf (transcripts/ empty; Yaskawa IR posts English result PDFs + Q&A summaries — the PDFs are binary/flattened and unfetchable via WebFetch, confirmed this run). Sentiment reconstructed from result commentaries + the Q&A summary:

• ~2 years ago: cautious/cyclical — soft China/EV auto, semiconductor servo weakness in the Americas/Asia, "revenue up but profit down" the recurring frame. Four consecutive terms without profit growth.
• ~1 year ago (1H FY ending Feb-2026): stabilizing — Motion Control margin improving on mix even as revenue dipped; forecast raised; tone shifts from "downturn" to "mixed, margin-defended."
• Latest (year-end + Dash-35 launch, Apr–May 2026): decisively constructive and inflecting — "turnaround to profit growth for the first time in four terms, driven by AI robots," AI robots "developed over two years now generating commercial revenue," a new 10-year Vision 2035 + 3-year Dash-35 plan unveiled, dividend raised.

Recurring phrases: "Physical AI," "i³-Mechatronics," "AI robots," "semiconductor/AI-related demand," "high-value-added," "advanced AI-driven mechatronics." What they stopped saying: the defensive "weak demand" refrain. What's conspicuously understated (same blind spot as FANUC): the China domestic-competition threat — management frames China as a demand-recovery tailwind and "high-value-added robot systems" win, not as Inovance taking the servo crown. That gap is the short's edge (Lens 13). Sentiment trend: improved to the point of narrative optimism as a brand-new decade-long plan is being sold to the market at an all-time-high stock price.

Comps

Yaskawa vs the automation/robotics complex.

Read — the single most important valuation fact: Yaskawa is now MORE expensive than FANUC on trailing P/E (51.4x vs ~45.8x) and EV/EBITDA (27.7x vs ~26.3x) — despite being the structurally weaker franchise: 8.7% consolidated operating margin (vs FANUC ~20%), 7.84% ROE (vs FANUC ~9–13%), net debt (vs FANUC net cash), and a robot business whose margin is falling. The Estun dossier from 2025 called Yaskawa "the sane-valuation anchor... ~16–18x fwd P/E". That anchor is gone. The stock has re-rated from ~16–18x forward to ~34x forward while its earnings fell — the "cheap sensible Big-Four name" is now priced like a growth compounder it has not yet proven to be. Consensus's own average price target is ¥6,907 — below the ~¥7,000–7,915 spot: the sell-side, in aggregate, sees no upside from here.

Stock-Price Catalysts (what moves >5%)

Pattern:

• The AI / Physical-AI / semiconductor narrative — the dominant 2025–26 driver. The stock more than doubled (+138.8% in a year) into an all-time high on Motoman NEXT (NVIDIA), the "AI robots now monetizing" story, the Vision-2035 launch, and the semiconductor-servo recovery — not on the (weaker) reported earnings.
• The prior de-rating (2021→2024): peaked in early 2021 (the post-COVID automation/semi boom), then fell for ~3 years as China/auto/semi capexCapital expenditureMoney spent on long-lived things — buildings, machines, servers — rather than on running costs. cooled and profit stalled (the USD ADR YASKY shows a −16%/yr 3-yr CAGR — see FX note). The 52-week range low was mid-2025 before the doubling.
• Semiconductor-equipment capex cycle — Yaskawa servos into wafer handlers/chip tools; the AI-capex up-leg is the current tailwind and the guidance's load-bearing assumption.
• China automation/auto capex — the robot-segment swing factor (and the Inovance-share risk).
• Yen — a large exporter; yen weakness flatters translated earnings/stock (guidance set at ¥145/USD). FX distorts the tape across listings: the domestic 6506.T is at an all-time high ¥7,915, while the USD ADR (YASKY) looks weaker since 2021 purely because the yen fell ~105→150.
• Order momentum (Q4 +20% YoY) and guidance beats are repeatable near-term catalysts; the Dash-35 plan launch was a narrative catalyst.

What the market reacts to, distilled: the Physical-AI/semiconductor narrative + order momentum + yen — and it reacts less to the slow China-servo-share erosion, which is why (as with FANUC) that risk is under-priced.

---

Phase C — Judge people & books

Management

A live leadership transition — the most important management fact this cycle:

• Masahiro Ogawa — President since 2022 — retires as Representative Director & President on 27-May-2026 (at the 110th AGM), citing a chronic illness and a desire to focus as lead executive on the AI Robotics + New Mechatronics Application growth areas; he becomes Vice Chairman, Executive Officer.
• Hiroshi Ogasawara — Chairman (and Ogawa's predecessor as President) — re-adds the President role, becoming Representative Director, Chairman & President (a combined chairman-CEO seat).
• The change was run through a nomination committee dominated by outside directors — a genuine governance-quality signal for a Japanese industrial. Framed as doubling down on the AI-robotics/Physical-AI pivot.

Assessment:

1. Track record. The Ogawa era delivered the strategy (i³-Mechatronics → Physical AI, Motoman NEXT, Tokyo Robotics) but a weak financial record: four terms without profit growth, ROE ground down to 7.8%, robot margins compressing. Ogasawara's prior presidency (2016–22) spanned better cyclical years. The bet is that the architect of the AI-robot pivot (Ogawa) now runs it unencumbered while the steady chairman (Ogasawara) holds the P&L.
2. Tenure & skin in the game. Long-tenured company lifers; insider ownership is modest (typical Japanese-industrial, institution-dominated register). our figures absent — n/a.
3. Capital allocation. Committed ¥150B of investment over Realize 25 (FY2023–25); the new Dash 35 targets ROE ≥12% / ROIC ≥11% by the year ending Feb-2029 — i.e. management is explicitly acknowledging that today's 7.84%/6.17% is inadequate and must roughly double. Dividend policy: payout ≥40% under Vision 2035 (raised DPS to ¥72). Carries modest net debt (not a lazy-cash-hoard critique like FANUC — the opposite). The capital-allocation question is execution, not stinginess: can they actually lift returns from sub-8% to 12%+?
4. Red flags. Low on the governance axis (outside-director nomination committee, IFRS reporting). The substantive flag is performance, not integrity: sub-economic ROE and poor FY cash conversion (FCF ¥5.9B).
5. Founder vs professional manager. Long-professionalized (founded 1915; no founder-family operating control). Implication: governance-driven, plan-driven — but also no founder's-paranoia edge; the moat depends on institutional R&D discipline.

Forensic Red Flags

Accounting-risk scan (IFRS — a positive vs FANUC's J-GAAP; web-only, no filings on shelf):

• Revenue recognition: Hardware + project (System Engineering) revenue. Watch project-percentage-of-completion in SysEng and large China robot projects — lumpy, and the "low-value-added large projects" that compressed robot margin are a disclosed mix issue, not a rec-timing red flag. Low-moderate.
• Non-operating gains flattering the prior year: The ~¥26.7B share-sale gain in the year-before inflated that comparison and is the mechanical cause of the −38% net drop. Not manipulation — but a reminder to judge Yaskawa on operating profit, not headline net.
• Cash vs earnings — the real flag. FCF ¥5.93B vs net income ¥35.2B; P/FCF ~305x; OCF ¥52.2B. Earnings ran well ahead of free cash this year (working-capital/inventory build in the cyclical trough + capex for the Physical-AI build-out). Verify next print that FCF converges back toward net income — sustained divergence would be the thing to worry about.
• SBC / non-GAAP flattery: Not a US-style SBC model; IFRS, modest stock comp. Low.
• Goodwill/intangibles: Tokyo Robotics + prior deals add some goodwill/R&D capitalization to watch, but Yaskawa is largely organic. Low-moderate.
• Net debt (¥63.6B) + weak FCF together: not distress (D/E 0.25, current ratio 2.47), but it removes the FANUC-style "fortress" cushion — a downturn is less comfortably absorbed. Moderate.
• FX: large exporter; ordinary/net income swing on yen. Guidance ¥145/USD — yen strength is a direct earnings risk.

Regulatory findings (required sub-section):

• SEC (EDGAR EFTS — LR + AAER): Zero findings.. Not an SEC registrant.
• Non-SEC enforcement: Web scan ("Yaskawa" (lawsuit OR fine OR penalty OR settlement OR antitrust OR export control) enforcement) surfaced only private civil litigation, no regulatory enforcement action: Nidec v. Yaskawa (2008, Tokyo District Court, LCD-panel-handling-robot patent infringement); Kollmorgen v. Yaskawa (2001, W.D. Va., servo patent); and an employment case (Shobe v. Yaskawa America). No antitrust, consent decree, export-control penalty, or agency fine found. The relevant category risk is export-control / dual-use (robots + high-precision servos have military-adjacent uses; Japan METI + allied regimes) — industry-wide, no Yaskawa-specific action sourced.
• 10-K Item 3 equivalent: N/a — Yaskawa files a Japanese 有価証券報告書, not a 10-KForm 10-KA company’s audited annual report to the US regulator. The most complete thing it publishes.; not fetched (IR PDFs binary-unreadable). Verify litigation from the annual securities report on next refresh.
• Verdict: No material regulatory or legal findings — verified via SEC EDGAR EFTS (LR/AAER, zero) and web search (only private patent/employment litigation), as of 2026-07-10. Structural exposure limited to industry-wide dual-use export-control, no company-specific action.

---

Phase D — Project & stress-test

Forward Projection

Base actuals & guidance: Year ended Feb-2026 — rev ¥542.1B, OP ¥47.3B, NP ¥35.2B, EPS ¥135.71. Company guidance for the year ending Feb-2027 — rev ¥580B (+7%), OP ¥60B (+27%), NP ¥47B (+33%) → EPS ~¥181. Dash-35 target (year ending Feb-2029): rev ¥650B, OP ¥100B (15.4% margin), ROE ≥12%, ROIC ≥11%.

Three-year EPS path (, arithmetic shown; only the FY-ending-Feb-2027 line is company-guided):

• Year ending Feb-2027 — base EPS ~¥181 (company guide NP ¥47B). Load-bearing: MC +19% on AI/semi, robot margin stabilizes, ¥145/USD holds.
• Bull ~¥205.
• Bear ~¥140.
• Year ending Feb-2028 — base EPS ~¥215. Bull ~¥250; bear ~¥150.
• Year ending Feb-2029 (Dash-35 terminal) — base EPS ~¥270. Bull ~¥300 (target met + option value); bear ~¥180 (margin stalls ~12%, China servo share loss caps growth).

The projection's load-bearing assumption: that Yaskawa executes a near-doubling of operating margin (8.7% → 15.4%) and ROE (7.8% → 12%+) in three years while defending servo share in China against Inovance. That is a large ask against a four-year record of no profit growth. The base case IS the company's own plan — the risk is asymmetric to the downside because the market at ¥7,000 already prices the plan succeeding.

Forecast (Brier) — not logged per --watchlist unattended rules (skip our model create). If promoted to a thesis, the base call to log would be: "Yaskawa (6506.T) FY-ending-Feb-2027 operating profit ≥ ¥58B, p≈0.55, resolves 2027-04-15" — i.e. does the +27% OP inflection substantially land?

Bull vs Bear

Bull case. Yaskawa is a century-old motion-control franchise inflecting from cyclical trough to an AI-robotics growth chapter. The crown-jewel Motion Control segment is already proving margin expansion on AI/semiconductor and data-center-inverter mix (10.3% and rising); the Q4 order surge (+20%) confirms an order-led recovery; and the Physical-AI stack — Motoman NEXT on NVIDIA Jetson/Isaac, plus the Tokyo Robotics humanoid-actuator acquisition — gives real optionality to climb from commodity hardware into intelligent motion, exactly as the market rewards. Vertical servo integration is a durable cost edge over ABB/KUKA. The new Dash-35 plan targets a credible near-doubling of OP to ¥100B and ROE to 12%+, with a rising 40%+ payout. If the semiconductor/AI-capex up-cycle persists and the humanoid option gets any value, a re-rating extends. Management just doubled down (Ogawa → AI-robotics lead).

Bear case (2–3 things that could permanently impair, or at least de-rate hard):

1. China takes the servo crown (the core-moat attack). Inovance is #1 in Chinese general servo (>26%), ahead of Yaskawa on that turf, growing ~22%/yr into policy tailwinds that favor domestics. Unlike FANUC (whose monopoly CNC is safe), Yaskawa's deepest moat — servo/motion — is the exact thing being commoditized in the largest market. If China servo share keeps sliding, Motion Control's margin story caps out.
2. Valuation air-pocket. 51x trailing / 34x forward / 28x EV-EBITDA after +139%, on 7.8% ROE, with consensus PT (¥6,907) below spot. A cyclical wobble, a missed +27%-OP guide, or a cooling AI narrative could de-rate the stock toward a Big-Four-normal ~18–22x forward — a 35–50% drawdown even on flat EPS.
3. Robot margin structurally low + weak cash. Robotics (the biggest segment) earns only 8.3% and falling, bought with low-value China auto volume; FY FCF was ¥5.9B (P/FCF 305x). This is not a high-quality cash compounder at a compounder's price.

Pre-mortem (18 months out, thesis broke): The FY-ending-Feb-2027 print misses the +27% OP guide (yen strengthens through ¥145, tariffs bite, China robot mix stays low-value, Inovance keeps taking servo share); the Physical-AI/humanoid narrative cools as Motoman NEXT revenue stays de minimis; the stock de-rates from 34x forward toward ~20x on ~flat EPS — a 40%+ drawdown from ¥7,000. The tell will be Motion Control margin rolling over while China-domestic servo share rises in the same quarter, or a guide-down at 1H.

Are multiples too high? Yes — Yaskawa now trades above FANUC on trailing P/E and EV/EBITDA while being materially lower-quality (margin, ROE, balance sheet). The premium is a pure Physical-AI-narrative + turnaround-hope premium with negative consensus upside; there is no margin of safety.

Contrarian view (what the market refuses to see): The market is pricing Yaskawa as a Physical-AI/humanoid winner at the same moment its core servo franchise is being out-executed at home in China by Inovance and its robot margins are structurally lower than FANUC's. Yaskawa could win the narrative (NVIDIA partner, humanoid actuators) and lose the margin war — a re-rate built on a story whose foundation (servo/motion dominance) is quietly eroding.

Devil's Advocate (short-seller)

Dismantling the bull case:

• Revenue concentration & the shift: ~46% Robotics (China-auto-heavy, low-value, margin-compressing 10.0%→8.3%) + ~44% Motion Control (servo, the segment Inovance is taking in China). Both of the two big segments are exposed to the single most contested market on earth — and one of them is Yaskawa's crown jewel.
• Why the moat is weaker than bulls think: Yaskawa's edge is servo/motion. But the marginal servo/robot sold in China is a "good-enough, 20–40% cheaper, locally-favored" Inovance/Estun product — and Inovance is vertically integrated too, so it doesn't need Yaskawa. The moat protects legacy installed base better than net-new volume, and the volume is in China.
• The most dangerous competitor bulls underestimate: Inovance (300124.SZ) — #1 in Chinese general servo (>26%), already ahead of Yaskawa on that turf, and a ~25x-forward compounder attacking exactly Yaskawa's crown jewel from below, with Estun taking the robot arm. A two-front war on both of Yaskawa's core segments, in the market that must grow for the thesis to work.
• Worst capital-allocation reality: ROE 7.84% / ROIC 6.17% — below cost of capital. The company is currently destroying economic value on invested capital, carries net debt, and generated only ¥5.9B FCF. Management's own Dash-35 plan concedes returns must roughly double — an admission that today's returns are inadequate. A short frames it: "sub-economic-return industrial, re-rated to 51x by an AI-narrative momentum crowd."
• Assumptions that must hold for ¥7,000: (i) the +27% OP recovery lands; (ii) Physical-AI/humanoid narrative persists; (iii) China servo share stabilizes vs Inovance; (iv) yen doesn't strengthen through ¥145; (v) semiconductor-capex up-cycle continues. Break any one and 34x forward is untenable.
• If growth disappoints 20–30%: guided EPS ¥181 → ~¥125–145; at a de-rated ~18–22x → ¥2,500–3,200 → a 55–65% drawdown from ~¥7,000. The downside math is severe precisely because both the multiple and the earnings are elevated/hoped-for.
• The single scenario that permanently impairs: Yaskawa becomes a sub-scale #4 robot maker AND a share-losing #2/#3 servo player in China — an ex-growth cyclical whose "Physical-AI" optionality never scales past pilots. Not bankruptcy (D/E 0.25), but a structural de-rate from narrative-growth (34x) to Big-Four-cyclical (15–18x) on lower normalized earnings. Plausibility: medium-high on a 3–5 year view, given the China servo-share data already printing.

Management Questions (ordered by information value)

1. China servo/robot share: What is Yaskawa's servo and robot unit share in China today vs 3 years ago, and what does your Dash-35 plan assume it is in FY-ending-Feb-2029, given Inovance is now #1 in Chinese general servo? (The single answer that most changes the thesis.)
2. Dash-35 targets OP margin 8.7%→15.4% and ROE 7.8%→12%+ in three years after four terms of no profit growth — what specifically bridges that gap, and how much depends on Physical-AI revenue that barely exists today?
3. Robotics revenue grew +4% but OP fell −14% (margin 10.0%→8.3%) on "low-value-added" China auto projects — do you defend robot share on price (margin) or cede volume and hold price (share)? Which, explicitly?
4. Motion Control margin is expanding on AI/semiconductor mix — how cyclical is that semiconductor-equipment servo demand, and what is the through-cycle floor if the AI-capex wave pauses?
5. FCF was only ¥5.9B vs ¥35.2B net income this year — when does cash conversion normalize, and how much of the gap is Physical-AI capex vs working capital?
6. On Motoman NEXT + NVIDIA: does open/foundation-model robotics lower the barrier for cheaper competitors and erode your motion-precision premium? How do you monetize AI without commoditizing your own servos?
7. Tokyo Robotics / humanoids: what is the realistic revenue timeline, and is this a growth vector or a defensive R&D hedge? How much capital will it absorb before it earns?
8. Inovance and Estun are vertically integrated like you but 20–40% cheaper — what is your durable answer in the China mid-market beyond "high-value-added"?
9. You carry ¥63.6B net debt and a 7.8% ROE — why is the balance sheet levered at all given sub-economic returns; is there a deleveraging or capital-return re-think?
10. Guidance assumes ¥145/USD — what is the earnings sensitivity per ¥5 move, and how hedged are you against yen strength?
11. What share of the FY-ending-Feb-2027 +27% OP guide is volume/mix vs price vs FX, and how much is the one steel-plant project?
12. Post-transition (Ogasawara chairman-president, Ogawa VC on AI robotics) — how are decision rights split, and does combining chairman+president weaken board independence despite the outside-director nomination committee?
13. Manufacturing spans Japan/China/US/Slovenia — how do tariffs and supply-chain regionalization change your cost structure and where you build?
14. Service/maintenance annuity: what % of revenue is recurring, and what is its steady-state growth — can you build a FANUC-like service moat?
15. What is the one competitive or technological development (Chinese domestic, a foundation-model platform shift, a humanoid entrant) that keeps you up at night?

---