Sumitomo Electric Industries

Phase A — Understand the business

Lens 1 · Company Overview

Sumitomo Electric Industries is a ~¥5.1-trillion-revenue ($34B-class) diversified Japanese industrial — one of the original Sumitomo-group companies, founded 1897 as a copper-wire maker. It is not a focused optical name; it is a materials-and-cables conglomerate with ~400 group companies and ~300,000 employees. It makes money across five reportable segments:

• Automotive (57.5% of FY2025 revenue) — automotive wiring harnesses (the dominant product; SEI is the global #2 by volume behind Yazaki), plus anti-vibration rubber, hoses, electrical parts, and traffic-control systems. Run largely through subsidiary Sumitomo Wiring Systems. This is a low-margin, labor-intensive, volume business: ~6% operating margin.
• Environment & Energy (23.1%) — power transmission/distribution cable (incl. HV and submarine), substation systems, electron-beam/vacuum equipment, conductive products, and vanadium redox-flow batteries (SEI has researched these since 1982, guaranteeing 20-25-year life).
• Information & Communications / "Infocommunications" (6.4%) — optical fiber, optical cable, fusion splicers, optical devices, and compound-semiconductor substrates. This is the AI-optical engine and the optical-computing thesis.
• Electronics (8.0%) — electronic wires, flexible printed circuits (FPC), fluoropolymer products, electron-irradiation products, metal components.
• Industrial Materials & Others (7.6%) — steel wire (precision-spring, steel cord/tire wire), carbide cutting tools, sintered parts, semiconductor heat-spreader substrates, and optical components.

Contract structure: Automotive is design-win, line-fit, multi-year OEM contracts — most with copper pass-through clauses but limited mid-cycle price flexibility. Infocommunications/optical is a mix of hyperscaler/telco project sales and component supply; the InP-substrate franchise is closer to a merchant-semiconductor model (sell wafers to transceiver makers like Lumentum, Coherent, AXT's customers). Customer concentration is moderate at the group level (auto OEMs Toyota/Honda/Nissan/global; optical to hyperscalers via the transceiver supply chain) — no single-customer dependence disclosed.

Lens 2 · Supply Chain

Map upstream → SEI → end customer, with named stakeholders:

Upstream inputs (chokepoints in bold):

• Copper — the single largest BOM input; "well over half of total bill-of-materials cost in a conventional loom". Sourced from global cathode/refiners; SEI announced a 2026 initiative for 100% recycled copper in select cable lines to differentiate vs less-compliant rivals.
• Silica / glass preforms — for optical fiber; SEI is vertically integrated (it makes the glass, the cable, AND the fusion splicers that join them). This vertical integration is a genuine moat (Lens 3).
• Indium phosphide (InP) feedstock — the critical chokepoint. Indium is ~70% China-supplied. China's February 2025 export-licensing controls on indium/gallium drove 6-inch InP wafer prices +250% to ~$5,000. SEI is on the upstream-supply side here, not just a consumer — it grows InP crystal and polishes substrates.
• Aluminum — SEI is developing aluminum-based harness solutions to cut copper exposure/weight.

SEI → end customer:

• Wiring harness → auto OEMs (Toyota, Honda, Nissan, plus global; EV-specific HV harnesses to BEV makers, growing in China/Europe) → vehicles.
• Optical fiber/cable + multicore fiber → telcos, hyperscaler AI data-center campuses (via the SDM4 MCF MSA, see Lens 3) → AI networks.
• InP substrates → optical-transceiver/laser makers (Lumentum, Coherent, the merchant transceiver chain) → 800G/1.6T modules → AI data centers. SEI holds ~50% of the global InP substrate market — making SEI itself a chokepoint one layer up from the transceiver bottleneck (Lumentum's CPO backlog reportedly stretches into 2028; AXT, the #2 InP maker, was constrained by Chinese export-license delays into mid-2026).

Single-source / chokepoint summary: SEI sits on the supply side of two squeezes (InP substrates; high-performance multicore/optical fiber) and the demand side of one (copper for harnesses). The InP position is the most strategically valuable node in the entire chain.

Lens 3 · Competitive Advantages (moats)

Optical / Infocommunications — strong, widening moat:

• Vertical integration in optical fiber — SEI manufactures glass, cable, and the fusion splicers to join them. Few peers control the full stack; this is a durable process/scale moat.
• Top-3 global optical fiber — the top-five (Prysmian, Corning, Sumitomo, Furukawa/Proterial, YOFC) hold ~45% of 2025 revenue; Corning ~10.4%, Prysmian ~9%. SEI and Furukawa lead in high-performance/telecom-grade fiber.
• ~50% of global InP substrate market — the foundational material for co-packaged optics and optical I/O. SEI controls crystal growth → polished substrate. This is the single most important moat for the optical-computing thesis, and it has been strengthened by China's export controls knocking out merchant Chinese supply.
• Multicore-fiber standard-setter — SEI is a founding member of the SDM4 MCF MSA (Feb/Mar 2026) alongside Corning, AFL/America Fujikura, and TeraHop, defining the 4-core multicore-fiber design for AI data-center interconnects. Co-authoring the standard is a network-effect moat.
• Compound-semiconductor IP — world-first GaN substrate mass production; engineered-GaN JV with Soitec; GaN-HEMT high-frequency research with University of Tokyo.

Automotive — moat = scale + switching costs, but commoditizing at the low end:

• Wiring-harness design-wins are sticky (years of co-engineering per platform), and SEI/Yazaki/Aptiv/Leoni form a tight oligopoly with Asia-Pacific ~51% share. But low-voltage harnesses are commoditizing; the value migrates to EV high-voltage harnesses and power modules.

Bargaining power: Strong over transceiver makers (InP near-monopoly amid shortage). Weak-to-balanced over auto OEMs (Toyota et al. resist mid-cycle price hikes; copper pass-through softens but does not eliminate margin risk).

Lens 4 · Segments

FY2025 (year ended March 31, 2026) reportable-segment table — net sales and segment profit, both years. All OPM / YoY / share figures ``:

*FY24 consolidated revenue shows as ¥4,789.3B in the segment-table base vs ¥4,679.8B in the IR Financial Highlights — a reclassification/restatement difference between the comparative-segment base and the headline; I flag the ~¥110B gap rather than silently picking one. The +9.2% headline growth is consistent with both bases on the FY25 ¥5,110.2B figure.

The single most important fact in this dossier: Infocommunications is only 6.4% of revenue but generated ~59% of the group's FY2025 operating-profit increase. Its OPM nearly tripled to ~24% as generative-AI data-center optical demand (optical wiring components, multicore fiber, InP) flooded in. This one segment is the entire re-rating story. Every other segment grew respectably (Electronics +35% OP, Industrial +53% OP off a small base, Environment & Energy +15% OP on power-cable/renewables), but the optical mix-shift is what moved the stock.

The structural tension: the segment driving the narrative and the OP growth is the smallest by revenue, while the segment that defines the company (Automotive, 57.5% of revenue) earns ~6% margins and grew OP only +4%. The bull thesis is a mix-shift bet — that high-margin optical compounds faster than the low-margin auto base dilutes it.

---

Phase B — Measure performance

Lens 5 · Earnings Result (FY2025, year ended March 31, 2026)

The headline print was a clean blowout and a record across the board:

• Revenue ¥5,110.2B (+9.2% YoY) — first time above ¥5 trillion.
• Operating profit ¥418.2B (+30.4%), OPM 8.2% (up from ~6.7%).
• Ordinary profit ¥431.3B (+39.3%); net income attributable to owners ¥369.5B (+90.7%) — the net-income near-doubling reflects operating leverage plus below-the-line items (FX, equity-method, lower one-offs).
• EPS ¥473.78 (vs ¥248.47).
• ROE 14.7% (from 8.6%); ROIC 12.2% (from 9.8%).
• Balance sheet: total assets ¥4,824.5B; net assets ¥2,835.0B; equity ratio 56.9% (strong); interest-bearing debt ¥709.8B (down from ¥775.9B) — net leverage is low. CFO ¥425.2B, capex ¥243.2B, R&D ¥162.9B.
• Drivers: management explicitly cited optical components for AI data centers + automotive cost pass-through as the swing factors.
• Guidance (FY2026, year ending March 2027): revenue ¥5,300B, operating profit ¥425B, ordinary profit ¥432B. This is the tell-tale: guidance is conservative — OP +1.6% after a +30% year. SEI is guiding a flat year even as the AI-optical demand it just rode is accelerating. Either management is sandbagging (the note.com analysis flags the guide as ~11% above prior-year consensus but a deceleration) or they see auto/macro/tariff offsets. The market is paying ~25x for the optical optionality, not the guide.
• Market reaction: the stock is up +314.5% over the trailing year, all-time high ¥14,850 on June 3, 2026. The print was expected to be strong; the move had largely happened ahead of it.

Unusual vs own history: net income +91% and ROE jumping ~6pts in one year is a regime change for a company that historically compounded mid-single-digits. The question (Lens 12/13) is whether ~24% Infocommunications margins are a new plateau or a shortage-driven peak.

Lens 6 · Earnings Calls (sentiment trend)

No transcripts on the research-layer shelf and SEI's IR webcasts/PDFs are Japanese-GAAP CEO presentations (binary PDFs that don't scrape); sentiment is reconstructed from the president's messages and results commentary:

• Tone shift over the last ~4 reporting cycles: rising conviction on Digital & AI. FY2024 (Apr 2025) framed growth as "data center optical devices + copper/FX tailwind." By the FY2025 H1 update (Nov 2025) and the FY2025 full-year + MTMP2028 launch (May 2026), the language hardened into a structural three-pillar strategy — "Digital & AI, Energy, Mobility" with a ¥1T investment commitment. That is management institutionalizing the AI-optical tailwind into capex and strategy, not treating it as a windfall.
• Recurring phrases: "Goho Yoshi / five-way win" (the Sumitomo stakeholder philosophy — appears in every message), "collective strength of the Group," "data center demand."
• What they stopped saying: the defensive "challenging environment / cost pressure / yen" framing that dominated FY2022-23 has receded; FY2025 messaging is offensive (records, milestones, growth investment). U.S. tariffs are acknowledged as a headwind but not a thesis-breaker.

Lens 7 · Comps

Peer table — SEI vs optical-fiber peers and wiring-harness peers. Multiples are `` with source/date or n/a. None fabricated.

Read: SEI at ~25x sits well below Corning (36-44x EV/EBITDA) and Fujikura's AI re-rating, but well above the auto-harness peers (Aptiv ~9.5x). That is exactly the valuation of a company the market is repricing from an auto-parts multiple toward an optical multiple — and is roughly halfway there. The re-rate has happened; further multiple expansion requires the optical mix to keep proving itself.

Lens 8 · Stock-Price Catalysts (moves >5%, last ~5y → concentrated in the last 12 months)

The defining feature is the +314.5% trailing-year run — a multi-year re-rating compressed into ~12 months, driven almost entirely by the AI-data-center optical narrative. Identifiable catalysts:

• Generative-AI data-center optical demand inflection (2025→2026) — the dominant driver; each quarter that Infocommunications OP accelerated re-rated the stock.
• China indium/InP export controls (Feb 2025) — repriced SEI's ~50% InP share from a commodity line into a scarce, pricing-power asset.
• SDM4 MCF MSA announcement (Feb/Mar 2026) — validated SEI as an AI-interconnect standard-setter.
• FY2025 record results + MTMP2028 + 4-for-1 split + dividend hike (May 12, 2026) — the catalyst cluster that took the stock to its ¥14,850 ATH on June 3, 2026.
• Sumitomo Riko absorption-type split agreement (June 26, 2026) — a fresh structural/portfolio move.

What the pattern reveals: for SEI, the market reacts to optical/AI-mix datapoints and supply-chain scarcity signals, not to the auto cycle. The auto business is treated as a stable-but-boring anchor; the marginal buyer is buying the optical call. That makes the stock sensitive to AI-capex sentiment and any sign the InP/optical shortage is easing — the same factors that would de-rate Corning/Fujikura.

---

Phase C — Judge people & books

Lens 9 · Management

• Track record: Osamu Inoue — President & COO (and CSO) since 2017, joined SEI 1975; career spanned U.S. and Germany operations. Masayoshi Matsumoto — Chairman & CEO. This is a classic Japanese promote-from-within bench: 40+-year company veterans, not lateral hires. They delivered the MTMP2025 plan (FY2025 explicitly "achieved goals announced in 2023's plan") and then set a more ambitious MTMP2028.
• Tenure & skin in the game: long tenure; inside directors "encouraged to hold shares." But no controlling shareholder/founder — ownership is institutional: Master Trust Bank 17.07%, Custody Bank 9.52%, Nippon Life 3.17%, with Sumitomo-group insurers (Sumitomo Life 2.08%) present; foreign ownership ~31.2% (Mar 2025). This is professional-manager governance, not founder-led — implying steadiness and capital discipline over bold risk-taking.
• Capital allocation: improving and shareholder-friendlier. DPS ¥154 (FY2025) vs ¥97 (FY2024) — a +59% hike; payout ~32.5%. 4-for-1 split (record date June 30, effective July 1, 2026; authorized shares raised 3B→12B) to broaden the investor base. ROE rose to 14.7% and ROIC to 12.2% — genuinely good for a Japanese industrial. The MTMP2028 commits ¥1T cumulative FY2026-28 to Digital & AI / Energy / Mobility — disciplined growth capex into the proven-hot areas.
• Red flags (governance): none material. The historic auto-parts cartel is covered in Lens 10 (and SEI came out clean on it — leniency applicant). The Goho-Yoshi multi-stakeholder framing is sincere Sumitomo doctrine but can be code for "shareholders are one of five constituencies" — i.e., don't expect aggressive buybacks/activism-style returns; this is a steward, not a maximizer.
• Archetype: professional-manager steward of a 125-year-old keiretsu institution. Implication for this stage: SEI will invest into the optical opportunity methodically (the ¥1T plan) and return cash gradually — it will not bet the company, and it will not financial-engineer the stock. Good for durability, a ceiling on explosive upside.

Lens 10 · Forensic Red Flags

Web-only forensic read (no filings on shelf; SEI reports Japanese GAAP). Areas of attention:

• Revenue recognition: auto-harness line-fit revenue is straightforward; optical project sales can carry timing/percentage-of-completion judgment. No disclosed issue. n/a — not independently verified against primary statements.
• Margin durability / non-recurring tailwind: the genuine analytical flag (not an accounting one) is that ~24% Infocommunications margins are shortage-inflated. InP prices +250% on China controls flatter SEI's substrate economics; if Chinese supply normalizes or AXT's licenses free up, that margin could mean-revert. This is a quality-of-earnings caution: a chunk of the FY2025 OP surge is scarcity rent, not durable structural margin.
• Balance sheet: clean. Equity ratio 56.9%, interest-bearing debt down YoY to ¥709.8B, CFO ¥425.2B > capex ¥243.2B → strong free cash conversion. No receivables/inventory blowout flagged. SBC is not a material non-GAAP distortion (Japanese-GAAP, modest equity comp). No goodwill-impairment red flag surfaced.
• Net income vs operating: net +91% ran well ahead of OP +30% — driven by ordinary-profit items (FX, equity-method, lower one-offs). Worth watching that below-the-line tailwinds (a weak yen) don't reverse; the operating +30% is the cleaner number.

Regulatory findings (required sub-section). Read from regulatory/regulatory-findings.md (Stage 1) + web:

• SEC EDGAR (LR + AAER): none possible. SEI has no CIK and is not an SEC registrant; the regulatory-findings file confirms 0 SEC findings and notes no EDGAR search is feasible.
• Non-SEC / antitrust history — material but resolved favorably. SEI was implicated in the 2000-2009 automotive wiring-harness cartel (prices/supply allocation to Toyota, Honda, Nissan, Renault). The European Commission fined Yazaki €125M, Furukawa €4M, Leoni €1.38M (total ~€141M, 2013) — but granted Sumitomo Electric and SEWS-E full immunity because SEI was the leniency applicant that alerted regulators to the cartel. In the parallel U.S. DOJ matter, Yazaki/Furukawa pleaded guilty (Yazaki paid $470M); SEI is not recorded among the fined. Net read: a dark industry history in which SEI behaved as the whistleblower and emerged with zero fines — a positive governance signal, not a liability.
• Current/other enforcement: no material live FTC/DOJ/EU action against SEI surfaced as of 2026-06.
• Conclusion: No material unresolved regulatory or legal findings. The one historic cartel matter resolved with SEI receiving immunity. Verified via the SEC-EDGAR-not-applicable note, web search (antitrust/enforcement), and public reporting as of 2026-06-29.

---

Phase D — Project & stress-test

Lens 11 · Forward Projection (FY2026 → FY2028, year-ends March)

Built bottom-up from FY2025 actuals + management's own MTMP2028 targets, which provide an unusually concrete anchor:

• Management's MTMP2028 (FY2028) target: net sales ¥6T, operating profit ¥600B, before-tax ROIC >15% — implying a ¥6T/¥5.11T = +17% revenue and ¥600B/¥418B = +43% OP over three years (a ~12.5% OP CAGR), funded by ¥1T of investment.
• Near-term guidance (FY2026): revenue ¥5,300B, OP ¥425B — deliberately conservative (a near-flat OP year).

EPS scaffold (pre-4:1 split; divide by 4 post-July-1-2026). FY2025 base EPS ¥473.78; growth rates are judgment, output:

Post-split equivalents (÷4): Base ~¥125→¥141; Bull ~¥134→¥171; Bear ~¥116→¥112.

Sanity vs MTMP2028: management's ¥600B FY2028 OP, taxed/below-the-line at FY2025-like ratios, would imply EPS roughly in the high end of my base-to-bull range — i.e., the company's own plan is more aggressive than my base. That gap is the bet.

No forecast.ts create in this unattended --watchlist run, per the SKILL — the base call is documented here for a later human-gated forecast log if promoted.

Lens 12 · Bull vs Bear

Bull case. SEI is a mispriced AI-infrastructure picks-and-shovels play hiding inside a boring industrial wrapper. It owns ~50% of the InP substrate chokepoint that every co-packaged-optics and 1.6T transceiver roadmap depends on, is vertically integrated in high-performance optical fiber, and co-authors the multicore-fiber standard for AI campuses (SDM4 MSA). The Infocommunications segment tripled its margin to ~24% and drove ~59% of group OP growth in a single year, and management has institutionalized the tailwind into a ¥1T, ¥6T-sales/¥600B-OP MTMP2028 plan. Meanwhile the auto-harness base provides ¥2.9T of stable cash and an EV-HV-harness growth option, the balance sheet is fortress-grade (57% equity ratio, falling debt, CFO 1.75x capex), ROE is up to 14.7%, and the company is sweetening returns (DPS +59%, 4-for-1 split). At ~25x it trades at a steep discount to Corning's 36-44x EV/EBITDA — there's room to re-rate further as the optical mix grows and the market stops valuing it like a wire-harness maker.

Bear case (permanent-impairment risks).

1. The optical margin is partly scarcity rent. A large slice of the FY2025 OP surge is the China-export-control-driven InP shortage (prices +250%). When Chinese indium supply or AXT's export licenses normalize — already underway in mid-2026 — that ~24% Infocommunications margin can compress materially. This is the single biggest risk to the thesis: the market is extrapolating peak-cycle optical margins.
2. 57% of revenue is structurally ~6%-margin, commoditizing auto harness. Low-voltage harnesses are commoditizing; SEI competes on labor cost (Mexico, SE Asia) against Yazaki/Aptiv/Leoni with limited pricing power vs Toyota et al. Any auto-production downturn or copper spike (incomplete pass-through) hits the largest segment.
3. Expectations are now baked in. After +314% in a year at an ATH, the stock prices the optical optionality, while management guides a flat FY2026 OP. The asymmetry has inverted: a single soft optical quarter or any "InP shortage easing" headline could de-rate hard.

Pre-mortem (18 months out, thesis broke): It's late 2027. China quietly expanded indium/InP export licenses through 2026-27, AXT and Chinese substrate makers refilled the channel, and InP wafer prices round-tripped most of the +250%. SEI's Infocommunications margin fell from ~24% back toward the low-to-mid teens; the segment that drove 59% of OP growth stopped growing OP. Simultaneously, a 2027 global-auto air-pocket and a stronger yen squeezed the harness base and reversed the FX tailwind that flattered FY2025 net income. The stock, having priced ~30%+ earnings growth, de-rated from ~25x to ~14x on flat-to-down earnings — a >40% drawdown — without anything "going wrong" operationally; the peak was simply a scarcity-and-narrative peak.

Are multiples too high? ~25x is not demanding for a genuine AI-optical compounder, but it is demanding if Infocommunications margins are at a cyclical peak and 57% of the company grows low-single-digits. Fair if the optical mix-shift + MTMP2028 deliver; rich if InP normalizes.

Contrarian view (what the market refuses to see): Bulls treat SEI as "Japan's AI-optical play" and price the Infocommunications segment as the story — but they under-weight that the InP windfall is a China-policy artifact, not a durable structural margin, and they ignore that the ¥2.9T auto anchor caps blended margins no matter how well optical does. The non-consensus call is that the easy re-rating is done: SEI was a screaming buy at an auto-parts multiple before the optical mix was understood; at ~25x after +314%, it's a hold-quality compounder, not a fresh double — and the next big move is more likely a scarcity-unwind de-rate than another leg up.

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

Dismantling the bull case:

• Revenue concentration risk: the profit growth is concentrated in one 6%-of-revenue segment whose margins depend on a Chinese export-control shortage. Take away the InP scarcity premium and ~59% of the OP-growth story weakens — the company reverts to a ~6-9%-margin diversified industrial growing OP mid-single-digits. The market is paying an optical multiple for a windfall.
• The moat is weaker than bulls think where it matters most for "optical computing": SEI leads multicore fiber and InP substrates, but it is not the leader in hollow-core fiber — that niche is led by Microsoft (via its Lumenisity acquisition), with Prysmian and others also in it. And the highest-value CPO laser/engine IP sits with Lumentum/Coherent/Broadcom, not SEI — SEI is the substrate supplier to them, a strong but commoditizable position if InP capacity floods back.
• Most dangerous competitor bulls underestimate: AXT + Chinese substrate makers. AXT is the #2 InP producer; once China's export licenses normalized (first licenses granted June 2025, ramping into 2026), merchant InP supply returns and SEI's ~50% share + pricing power erode. SEI's own 40-100% capacity expansion adds to a future glut.
• Capital allocation / incentives: no controlling owner and a five-stakeholder ("Goho Yoshi") philosophy means shareholders are explicitly one of five priorities — don't expect buyback aggression or asset-light discipline. The Sumitomo Riko absorption-type split (June 2026) is a portfolio reshuffle worth scrutinizing for value transfer within the keiretsu.
• Assumptions that must hold for ~25x: (a) Infocommunications margins stay near peak; (b) AI-capex keeps accelerating optical demand; (c) auto base stays stable with successful copper pass-through; (d) yen stays weak (FX tailwind). Break any two and the multiple is unjustified.
• −20-30% growth-disappointment scenario: if FY2027-28 OP comes in ~25% below the MTMP2028 trajectory (¥600B → ~¥450-480B), the stock likely de-rates from ~25x toward the mid-teens — a 40-50% downside from the ¥14,850 ATH, given how much narrative is priced.
• Single permanent-impairment scenario: the least plausible (the company is too diversified and cash-rich to be impaired) — the realistic bear is a de-rate, not impairment. SEI's downside is "expensive compounder reverts to fair value," not "business breaks."

Lens 14 · Management Questions (ordered by information value)

1. Of the ~¥57B Infocommunications operating-profit increase in FY2025, how much is volume/mix vs. InP-substrate price (the China-export-control shortage), and what is the durable structural margin for that segment once InP supply normalizes? (The whole thesis turns on this answer.)
2. How does SEI's InP-substrate market share and pricing evolve as AXT regains export licenses and Chinese merchant supply returns — and how much of your 40-100% capacity expansion risks a 2027-28 oversupply?
3. MTMP2028 targets ¥600B OP, yet FY2026 guidance is a near-flat ¥425B. Walk us through the OP bridge from ¥425B to ¥600B — how much comes from Digital&AI vs Energy vs Mobility, and in which years?
4. Where exactly does the ¥1T FY2026-28 investment go by pillar, and what ROIC do you underwrite for the Digital&AI portion specifically?
5. In hollow-core fiber and co-packaged optics, where does SEI add the most value vs. competing on the substrate layer below Lumentum/Coherent/Broadcom — and is hollow-core a market you intend to lead or supply into?
6. Automotive is 57.5% of revenue at ~6% margin. What is the path to structurally higher harness margins — EV high-voltage mix, aluminum, automation — and what blended Automotive OPM do you target by FY2028?
7. How complete is copper pass-through in your auto contracts, and what is the P&L sensitivity to a sustained +20% move in copper that OEMs resist re-pricing mid-cycle?
8. How much of FY2025 net income's +91% (vs OP +30%) was FX/below-the-line, and how exposed is FY2026-28 earnings to a yen reversal?
9. What is the strategic rationale and shareholder-value case for the Sumitomo Riko absorption-type split — and how do you ring-fence minority shareholders from intra-group value transfer?
10. With ROE at 14.7% and a 57% equity ratio, why is the payout ratio only ~32%? Under "Goho Yoshi," what would make you lift returns or buy back stock?
11. What share of Infocommunications revenue is now tied to AI data centers specifically, and how concentrated is it among the top hyperscaler/transceiver customers?
12. How should we think about U.S. tariff exposure across the harness and optical footprints, and what is the mitigation (localization, Mexico, pricing)?
13. What is the commercialization timeline and revenue ambition for vanadium redox-flow batteries within the Energy pillar — is this a real third leg or a long-dated option?
14. Where are you in GaN/SiC power-device scale-up (incl. the Soitec engineered-GaN JV), and when does compound-semi become a material P&L contributor vs. an R&D showcase?
15. Which one external variable (InP supply, AI-capex cycle, auto production, yen, China policy) keeps you up at night as the biggest risk to MTMP2028?

---