7826 — Furuya Metal Co., Ltd.
Thesis
Furuya Metal (7826.T, TSE Standard) is the cleanest small-cap pure-play on ruthenium adoption at advanced-node logic and DRAM. It is the global #1 or #2 ruthenium sputter-target maker, the global #1 iridium-crucible maker, and a moat-protected processor of two of the rarest precious metals in the world. The bull case: as TSMC, Samsung, and Intel migrate from copper to ruthenium for the bottom metal layers and contacts at 2nm / A14, the dollar content of Ru per wafer steps from "tens of milligrams as a liner" to "multiples more as bulk metal," and the new Sano sputter-target plant is timed to absorb that demand.
Stance / verdict: WATCH — partial scale-in possible, not a chase. Conviction: medium. The story is real, the management is high-trust, the moat is genuine, and the A14 Ru-content thesis has a credible path. But the stock has roughly 5x'd off the 2024 low (¥2,151 → ¥11,010, currently ¥9,640 as of 2026-05-15), the valuation has minimal margin of safety at 24x forward P/E, the FundamentEdge growth-primacy rule fails (it is a 5-8% organic grower, not an 8-12% compounder), and the technical setup is mid-range with no edge. The asymmetry that existed at ¥2,000 in early 2024 is gone — this is now a "watchlist for pullback" name, not a chase-it-here name.
What has to be true (the single most important thing). One or more leading foundries publicly selects Ru (vs Mo / Co / air-gap-only) as the material of record at the A14 / 14A node-decision window in 2026–2027. The clock that matters is that A14 materials-decision window: miss it and the bull thesis pushes out two years; hit it and Furuya's Thin Film segment can compound at 20%+ for 3–5 years.
The central interpretive problem. Furuya's reported margins have been compressed by the very ruthenium price strength the bulls cheer — gross margin fell from 35.6% (FY22) to 24.7% (FY25) even as revenue rose ~27%. This is metal-price pass-through, not operational decay. Reported revenue is a misleading volume signal during PGM price moves; the right volume proxy is gross profit, not revenue.
Build a starter position only if Japan-semi-thematic exposure is missing from the book; otherwise wait for a ¥7,500–8,000 pullback or A14 material-of-record confirmation before sizing up. Re-trigger the checklist on either event.
Snapshot
One-liner. A Japanese precious-metal artisan turned semiconductor materials supplier — the company's iridium crucibles still grow LED sapphire, but the share-price story is now ruthenium sputter targets for advanced-node interconnects.
- Ticker / exchange: 7826 (Tokyo Stock Exchange, Standard Market; 7826.T). Legal name Furuya Metal Co., Ltd. (株式会社フルヤ金属).
- HQ: Tsukiji, Chuo-ku, Tokyo (corporate); main plant Tsuchiura, Ibaraki; new fab Sano, Tochigi; R&D at Toride, Ibaraki.
- Founded: 1968 (incorporated; Furuya family in precious-metal trading since the early 1900s). IPO 2003 on TSE JASDAQ; moved to TSE Standard in the 2022 segment reshuffle.
- Fiscal year end: 30 June. Employees: 305 consolidated (FY25).
- Country: Japan.
Price / valuation snapshot (as of 2026-05-15, ¥9,640):
| Metric | Value |
|---|---|
| Share price | ¥9,640 |
| Market cap | ¥237B (~$1.6B USD at 145 JPY/USD) |
| Enterprise value | ¥277B |
| Shares outstanding | 24.6M |
| Float | 9.4M (38% — low; family + institutional lockup) |
| P/E (TTM) | 31.3x |
| Forward P/E (FY26E) | 24.1x |
| EV / Sales (TTM) | 17.2x (distorted by metal-price headline revenue) |
| EV / EBITDA (TTM, yfinance) | 159x (data artifact — see note) |
| EV / EBITDA (FY25 reported EBITDA ¥11.7B) | ~24x (the useful number) |
| P/B | 3.5x |
| Dividend yield | 2.92% (payout ratio ~31% normalized) |
| 52-week range | ¥2,151 – ¥11,010 |
| Insider ownership | 33.5% |
| Institutional ownership | 23% (mostly passive) |
Note on EV/EBITDA: the 159x reading is a yfinance artifact pulling a depressed trailing EBITDA against an inflated EV; using FY25 reported EBITDA of ¥11.7B gives ~24x, the more useful framing. The stock has roughly 5x'd off the 2024 low on the Ru-at-2nm thesis — most of the bull case is in the price.
Business
Furuya buys, refines, and fabricates platinum-group metals (PGM) — primarily ruthenium (Ru), iridium (Ir), and platinum (Pt) — into highly specialized industrial products. Two product families matter for the equity story: (1) sputtering targets (the disc of metal a fab vaporizes in a PVD tool to deposit a thin film on a wafer), sold to semiconductor and HDD customers, and (2) iridium and platinum crucibles used to grow single-crystal sapphire (LED substrates) and SiC seed, and to melt specialty glass. The rest is Pt-Rh thermocouples, fine chemicals, and PGM recycling.
Segments (FY25 estimate; company doesn't disclose segment margins cleanly)
| Segment | What it makes | ~% of revenue |
|---|---|---|
| Electronics | Iridium / platinum crucibles for sapphire & glass; precision tooling | ~30–35% |
| Thin Film | Ru / Ir / Pt sputtering targets and evaporation materials (HDD, semis, FPD, MRAM) | ~30–35% ← the GAA story |
| Thermal | Thermocouples, Pt-Rh temperature sensors | ~10% |
| Fine Chemicals & Recycling | PGM compounds, catalyst pre-cursors, scrap refining | ~15–20% |
| Supply Chain Support | PGM brokerage / leasing to external manufacturers | ~5–10% |
Thin Film (ruthenium sputter targets) is driving the recent multiple expansion. A single advanced-node Ru target runs $50K–200K; an iridium crucible runs $100K to $1M+ each (a 300mm crucible can contain 30+kg of Ir, typically consigned by the customer). Targets are a true consumable (weeks to months of life — structurally annuity-like once qualified); crucibles last 12–36 months before the metal is reclaimed and re-formed.
Business model — read carefully
Furuya operates on a metal-content + processing-fee model. Customers either (a) consign the precious metal — Furuya processes it for a fee and the metal price washes through — or (b) buy the finished product with metal embedded, in which case Furuya bears working-capital and price risk on the inventory. The company hedges embedded inventory on the LBMA / TOCOM, but hedges are imperfect — ruthenium has no liquid futures market. Reported revenue is heavily distorted by spot PGM moves; the processing fee — the "real" margin — is a small slice of the headline number. Investors who buy Furuya as a pure semis play often miss this.
Geographic mix: ~55% Japan, ~30% Asia ex-Japan (Taiwan, Korea, China for HDD and semis), ~10% North America, ~5% Europe.
Why it matters technically
Below the 7nm node copper interconnects break down: copper needs a barrier layer (TaN) that, at sub-15nm pitches, consumes too much cross-section; line resistance and RC delay explode; electromigration worsens. The industry tried cobalt (Intel 10nm/Intel 7) and walked it back at Intel 4. Ruthenium is now the leading materials-roadmap candidate for the bottom interconnect layers because it needs no barrier (Ru-on-low-K is stable), has lower line resistance than Cu at sub-10nm pitches, and tolerates higher current density. At the contact level, Ru is explored as a contact-fill alternative to tungsten, which also needs a non-scaling barrier. TSMC has used a Ru liner on N3 (per SemiAnalysis IEDM 2022 coverage: "decreased contact resistance by 20-30%, via resistance by 60%"); Intel has used Ru in 10nm/Intel 7; IBM-Samsung demonstrated subtractive Ru interconnects at 18nm pitch. The open question is whether Ru goes from liner (tens of mg/wafer) to bulk metal (5–10x more mass) at A14 / 2nm and below.
Ru (atomic #44) is mined ~30 tonnes/year globally — about a hundredth of platinum, a thousandth of copper; ~80% from South Africa, ~15% Russia. Melts at 2,334°C; targets are made by powder metallurgy / hot isostatic pressing (HIP), not casting. Iridium (#77) is rarer still (~8 tonnes/year mined), melts at 2,466°C, the most corrosion-resistant metal known — the only crucible that survives growing single-crystal Al₂O₃ (sapphire) at 2,050°C in oxygen.
Moat (specific, not broad — three layers)
- Process IP in Ru/Ir powder metallurgy — extreme melt points, oxidation, and machining issues; decades of accumulated know-how hard to replicate in 3 years.
- PGM working-capital scale — a permanent precious-metal inventory worth multiples of revenue is itself a barrier to entry.
- Customer qualification — getting a sputter target qualified at a fab is an 18–24 month process; switching costs are real. (Plus recycling closure — Furuya buys back used targets at a discount, important PGM working-capital arithmetic for the customer.)
The moat doesn't prevent substitution at the chip-design level: if cobalt, molybdenum, or air-gap-only schemes out-compete Ru at A14, the bull case is lost. Quality verdict: durable but not transformative — a high-quality niche, not a software-quality compounder.
Customers (inferred — Furuya doesn't disclose named customers; <10% per customer stated policy)
| Customer (inferred) | Ticker | Est. rev share | Relationship |
|---|---|---|---|
| Western Digital / Seagate / Toshiba HDD | WDC, STX, 6502.T | 15–20% | HDD media sputter targets (Ru for granular PMR / HAMR) |
| TSMC / Samsung / Intel (via distributors + direct) | 2330.TW, 005930.KS, INTC | 10–15%, growing | Semiconductor sputter targets — the GAA story |
| Nichia / Sumitomo Electric / Wolfspeed | private, 5802/5802 | 5802.T, WOLF | 10–15% |
| Samsung Display / LG Display / BOE | 005930.KS, 034220.KS | ~5% | FPD sputter targets (declining mix) |
| Japanese auto / chemical catalyst makers | various | ~5% | Pt / Rh catalyst pre-cursors |
No single customer disclosed >10% of revenue — unusual for a niche supplier and a genuine positive. HDD exposure is double-edged: a real franchise but a secularly declining unit market (HAMR / Seagate Mozaic 3+ extends the runway).
Competitive position
Direct competitors: Tanaka Kikinzoku (Tanaka Holdings, private JP — closest competitor; Ru targets and Ir crucibles), Heraeus Precious Metals (private DE), Mitsubishi Materials (5711.T — diversified, PGM is one of many lines), Johnson Matthey (JMAT.L — catalyst-heavy, limited target overlap), Solar Applied Materials (8048.TW — lower-cost Asian target competition). Furuya is the most pure-play public name in PGM sputter targets and Ir crucibles — the entire reason it trades as a Ru-thematic stock, since there is no other listed alternative (Tanaka is private, Heraeus is private, Mitsubishi too diversified).
Porter five forces: rivalry moderate (few disciplined players); supplier power high (PGMs from ~4 South African and Russian mines); buyer power moderate (fabs leverage at re-bid but qualification costs are real); substitutes are the structural risk (Mo, Co, air-gap-only at the materials-roadmap layer); new entrants low (capex + know-how + PGM working capital).
Financials
All figures in JPY unless noted. FY ends 30 June. FY26E is the 2-analyst consensus mean (sparse coverage — interpret with caution).
Income statement & margins (¥M)
| Metric | FY22 | FY23 | FY24 | FY25 | FY26E |
|---|---|---|---|---|---|
| Revenue | 45,321 | 48,115 | 47,527 | 57,379 | ~62,000 |
| Revenue growth YoY | +12% | +6.2% | -1.2% | +20.7% | ~+8% |
| Gross profit | 16,152 | 15,380 | 14,670 | 14,188 | n/a |
| Gross margin | 35.6% | 32.0% | 30.9% | 24.7% | recovers if PGM stabilizes |
| Operating income | 13,056 | 11,485 | 9,814 | 9,538 | ~10,500 |
| Operating margin | 28.8% | 23.9% | 20.7% | 16.6% | ~17% |
| EBITDA | 14,188 | 13,723 | 12,858 | 11,748 | n/a |
| Net income | 9,142 | 9,406 | 7,410 | 6,468 | ~9,800 |
| Net margin | 20.2% | 19.5% | 15.6% | 11.3% | ~16% |
| Diluted EPS (¥) | 435 | 447 | 302 | 262 | ~400 |
Read the margin trend carefully. Revenue rose 20.7% in FY25 but operating income fell. This is the precious-metal-price pass-through effect: when ruthenium spot doubled, headline revenue inflated (the metal embedded in each shipment costs more) without proportional margin gain, because hedging is imperfect on Ru and processing fees are renegotiated on contract cycles. The "real" volume + processing-fee growth is closer to high-single-digit, well below the 20% headline. The gross-margin compression from 35.6% to 24.7% is mostly metal-price pass-through, not operational decay — normalize GP-per-revenue by the PGM index to see the underlying trend.
Discrepancy flag — revenue CAGR. The checklist states a 4-year revenue CAGR of 8.2% (¥45.3B → ¥57.4B over 3 years), with real volume CAGR likely ~5%. The deep-dive's FY22 YoY is shown as +12%. Both are kept; the point both agree on is that headline growth overstates real volume growth materially.
Incremental margin analysis (FY25 vs FY24)
- ΔRevenue: +¥9,852M
- ΔGross Profit: −¥482M
- Incremental gross margin: NEGATIVE — every additional revenue yen lost gross profit because the added "revenue" was largely PGM price pass-through with no associated processing fee
- ΔOp Income: −¥276M → incremental operating margin: NEGATIVE
This confirms the structural reading: FY25 revenue growth is largely metal-price illusion and the underlying operating leverage is currently masked. Tracking the FY26 incremental on revenue × volume mix is the key analytical exercise — if PGM prices stabilize and Sano volumes ramp, incrementals should swing strongly positive in FY26–27.
Cash flow & balance sheet (¥M)
| Metric | FY22 | FY23 | FY24 | FY25 |
|---|---|---|---|---|
| Operating cash flow | -1,225 | -1,460 | ~4,000 (est) | 921 |
| Capex | -1,758 | -3,200 | -4,500 | -4,914 |
| Free cash flow | -2,983 | -4,660 | -500 (est) | -3,993 |
| FCF margin | -6.6% | -9.7% | -1.0% | -7.0% |
| Net debt (cash) | net cash | net cash | -8,000 (est) | ~+15,174 |
| Net debt / EBITDA | net cash | net cash | net cash | ~1.3x |
| ROIC (rough) | ~22% | ~17% | ~12% | ~9% |
Free cash flow has been persistently negative for four consecutive years — consuming cash to fund the Sano plant capex and to build PGM working-capital inventory at higher metal prices. Net debt flipped from a net cash position pre-2024 to ~¥15B at FY25: total debt ¥21.7B against ¥6.5B cash. Current ratio 3.9x (working-capital heavy). This is a meaningful balance-sheet shift for a previously fortress-balance-sheet name and an underdiscussed risk. Capex stepped from ~4% to ~9% of revenue with the Sano build — the elevated-investment phase.
ROIC has compressed from ~22% to ~9% over four years — partly margin compression, partly capital base growing faster than profit. WACC for a Japanese small-cap of this profile is ~7–8%. ROIC > WACC barely, and the spread is narrowing — the most underdiscussed risk. If Sano doesn't ramp and Ru content doesn't step up, ROIC could fall through WACC. No aggressive-accounting flags — the deterioration is operational + commodity-cycle, not accounting-related.
Second-derivative check
| FY22 | FY23 | FY24 | FY25 | |
|---|---|---|---|---|
| Revenue YoY % | +12% | +6.2% | -1.2% | +20.7% |
| Δ growth rate | — | -5.8pp | -7.4pp | +21.9pp |
The +21.9pp acceleration in FY25 is mostly metal-price but also reflects real volume strength as Sano comes online; the exit run-rate into FY26 looks healthy.
Industry landscape
Furuya sits at the PGM fabricator layer of the semiconductor materials value chain: PGM mines (Anglo American Platinum / AMS.L, Norilsk Nickel, Sibanye-Stillwater / SBSW, Implats) → LBMA refiners (Anglo, Norilsk, Heraeus, Johnson Matthey, Tanaka) → PGM fabricators (Furuya★, Tanaka, Heraeus, Mitsubishi) → fab tool makers (AMAT, TEL, Lam) → foundries/IDMs (TSMC, Samsung, Intel) → fabless/systems. Furuya also overlaps the refiner layer via recycling.
The genuine bottleneck is not a single supplier but the Ru spot market itself — annual primary Ru supply is ~30 tonnes globally; bulk-Ru adoption at scale could double current demand against a near-vertical, price-inelastic supply curve. Ru spot doubled from ~$500/oz at the start of 2024 to >$1,000/oz by mid-2025 (S. African mine output cuts, Russian sanction premium, advanced-node demand perception). 81% of 2024 Ir output and ~90% of Ru output come from South Africa; ~9% Ir / ~4% Ru from Russia (per STF / SFA Oxford). SFA Oxford estimates a Ru supply shortfall of ~203k oz in 2026.
TAM building blocks (TECHCET / SEMI / industry-consultant estimates): Ru sputter targets for semis $200M (2022) → $400–600M (2028) base, $1B+ if bulk-Ru reaches multiple fabs; Ir crucibles for sapphire/SiC seed $300–400M, stable-to-modest; HDD targets ~$200M flat-to-down; DRAM electrode Ru $50–150M emerging; MRAM Ru $20–50M emerging. Total addressable Ru/Ir/Pt-specialty for semis + adjacencies roughly $700M–1.2B by 2028 — against Furuya's FY25 revenue of ¥57.4B (~$370M), this is a potentially company-transformative line. SEMI projects sub-2nm wafer-starts/month rising from <200k (2025) to >500k (2028).
See sector page: passives-mlcc
Management
Leadership
| Name | Title | Tenure | Background |
|---|---|---|---|
| Yoichi Nishikawa (西川 洋一) | Representative Director, President & CEO | Since 2017 (CEO) | Joined Furuya 1985; engineering / production background; rose through plant management; no external roles; 30+ year internal lifer |
| Furuya family member (rotating) | Non-executive Chairman | Long-standing | Founding family with PGM-trading roots from the early 1900s; specific incumbent rotates at AGM |
| Tatsuro Imai (今井 達朗) | Director, CFO | Since c.2020 | Internal promotion from corporate planning |
| Heads of Electronics / Thin Film / R&D | Operating directors | Long-tenure | Plant / technical lifers — typical of a family-controlled Japanese mid-cap |
Names sourced from FY25 Yuho; cross-check after the September 2025 AGM for board changes. Every appointment is an internal promotion — no "parachute" hires. No member of management has been associated with another listed company in an executive capacity, nor with any prior bankruptcy, regulatory action, or material litigation in public records. Clean regulatory/enforcement history in EDINET, FSA actions, and Japanese-language press.
Jurisdictional note. Furuya is a TSE Standard filer; US tooling (SEC EDGAR, DEF 14A) does not apply. Japanese equivalents are EDINET, the annual Yuho (securities report), and the Kabunushi-shokai corporate governance report. Where US-style granularity (individual officer ownership, full board comp tables, fine-grained related-party tables) is not disclosed at TSE Standard tier, that absence is itself a finding.
Ownership & alignment
Aggregate insider ownership: ~33.5% per yfinance — high by Japanese standards. The founding Furuya family holds the bulk (~25–30%, est. ~6–7M shares) via Furuya Trading (a closely-held family holding vehicle) plus personal accounts; directors and officers ex-family hold ~2–4% aggregate; CEO Nishikawa holds a single-digit % personal stake (est. <1%); CFO Imai de minimis. Institutional ownership ~23%, almost entirely passive (Vanguard Total International 0.93%, Vanguard Developed Markets 0.62%, Dimensional Intl Small Cap Value 0.40%, DFA Japanese Small Co 0.27%, Avantis Intl Small Cap Value, Schwab Intl Small Cap).
Net insider activity (LTM): no material reported sales; a small executive stock-purchase plan operates monthly (open-market buys). Highest-confidence signal in the whole DD: at a 5x'd-from-low share price, the family block has NOT been selling — they could trim into strength and have not, and the CEO/CFO continue to add modestly through the monthly plan. That is the single most credible alignment signal available.
Holdings concentration: the founding family's net worth is concentrated in 7826 (>95% of family net worth) — no parallel public-company holding of comparable size, no competing private operating company, no "real money is elsewhere" signal. The cleanest possible alignment shape.
Shell & cross-holdings scan — Green
Furuya Trading sits as a top-3 shareholder and functions primarily as a share-custody entity for the family, not a separate operating business; it is not a competitor, customer, or supplier of size. Related-party transactions (FY24 Yuho) are limited to small recycled-PGM purchases from Furuya Trading at market-referenced prices (low single-digit % of cost of sales), with explicit pricing methodology. No IP licensing fees, no consulting agreements, no lease agreements (Tsuchiura and Sano plants are owned by Furuya Metal directly), no management fees, no royalties to insider entities. No undisclosed related entities, no asset-migration pattern, no nominee/opaque beneficial ownership. Subsidiaries (Furuya Metal Korea / Taiwan / Shanghai sales offices + Korea recycling cooperation) are all consolidated and disclosed. Verdict: clean, textbook Japanese family-controlled holding structure.
Capital allocation — B / B- (the one Yellow)
No equity issuance in over a decade (share count essentially flat ~24.6M). No material buybacks. Progressive dividend: ¥282/share in FY25, ~31% payout on normalized earnings, never cut, never aggressive. Discrepancy flag: the deep-dive notes ¥282/share dividend vs ¥262 FY25 EPS implies a 105%+ payout ratio on the latest unnormalized year, while the normalized basis is ~31% — both figures are kept; the divergence reflects the FY25 earnings trough from PGM pass-through. Capex stepped from ¥1.8B (FY22) to ¥4.9B (FY25), the Sano plant being the bulk; whether it earns its cost of capital is the open question, answerable in 2027 on utilization data. M&A essentially none (conservative family preference).
Capital-allocation timing test: through the 5x move management did NOT buy back, issue equity, or do M&A — passive throughout. This is family-controlled inertia, not active misallocation. The genuinely missed opportunity was the FY24 trough (a buyback at ¥2,000–3,000 would have been highly accretive); management did not act and does not appear to think about cost of equity actively. Conservative-but-not-strategic: won't destroy value, won't optimize it either.
Compensation — Yellow (Japan norm)
CEO total comp ~¥50–80M/year (~$350K–550K USD-equivalent); total board comp envelope ¥350–450M in FY24 (rough Yuho estimate). Comp drivers are not disclosed at performance-metric granularity (no revenue/ROIC/FCF/TSR thresholds; a discretionary bonus tied to "company performance"). SBC minimal. No golden parachutes, no unusual perks. Japanese small-caps almost never use US-style PSU/PRSU hurdles — there is no public hurdle-vs-LT-model table to reconcile. Alignment is structural via ownership (strong), opaque via incentive design (Japan norm, not a Furuya-specific failing). Management has no contractual incentive to pursue the high-octane bull thesis aggressively — they will be conservative-steady, so the Ru bull case may take ~2 years longer to crystallize at Furuya than at a US peer with hard hurdles.
Credibility — High (~80% follow-through)
Guidance tendency conservative / straight-shooter: revenue guidance has tracked actuals within ±10% (FY22 beat +8%, FY23 beat +2%, FY24 miss -5%, FY25 beat +10%); operating-income guidance less precise (FY22 beat +9%, FY23 miss -8%, FY24 miss -11%, FY25 beat +6%) — partly PGM-price distortion, partly cautious guiding. No pattern of overpromising and missing. Operational milestones hit (Sano commissioned on time in 2024; capex elevated as guided). The FY24–25 margin-normalization call was wrong (didn't anticipate the Ru spot rally fast enough to adjust pass-through pricing) — excusable but a "not perfectly nimble" data point. Weasel-language frequency low. Mid-term plan target: ¥70B revenue and ¥15B OP by FY27 (in progress, on track per current run-rate).
Governance — Green
TSE Standard code-compliant; ~7 directors; ≥2 outside/independent directors. "Company with Audit & Supervisory Committee" structure with at least one CPA-eligible outside director. Outside directors are professionally qualified (former bankers/lawyers/academics) but not high-firepower active overseers. No dual-class shares, no formal poison pill (control implicit via the family block). No contested shareholder proposals, no activist, no litigation/enforcement history.
Overall management grade: Green-Yellow / B+. A high-trust family-controlled story; the two Yellows (opaque incentive design, capital-allocation inertia) are structural to Japanese small-cap norms rather than Furuya-specific. No red flags identified. Yes — I would trust these people with my capital. The open question is not trustworthiness but nimbleness: are they nimble enough to capture the Ru-at-A14 opportunity if it materializes? The record is mixed — operationally strong, capital-allocation-passive. Size accordingly: a long-term-trust-the-family hold, not a high-conviction execution bet.
Key-person risk: CEO Nishikawa is replaceable; the family chairmanship provides continuity. The deeper concentration is in the 20–30 process chemists and metallurgists who hold the Ru/Ir powder-metallurgy know-how — real tribal-knowledge risk, not disclosed.
Catalysts & risks
Bull case / catalysts
Near-term (0–12 months): FY26 Q1 print (mid-November 2026 JST); TSMC / Samsung / Intel earnings commentary on advanced-node (A14 / 18A) materials decisions; Ru spot price action through the Northern-Hemisphere / South African mining season (June–September).
Medium-term (1–3 years): A14 / 14A node Ru material-of-record decisions; Sano plant capacity utilization; HBM4 DRAM Ru-electrode commercial volume; MRAM volume inflection at Samsung foundry.
Structural tailwinds (3–5 year): GAA logic ramp at TSMC N2 / Samsung SF2 / Intel 18A→14A (2026–2028); HBM4 / DRAM scaling (Ru electrodes in capacitor stacks); MRAM commercialization (Ru in MTJ stacks); SiC and GaN seed-crystal growth (Ir crucible demand, EV power); Japanese METI advanced-materials grants (Sano received a METI subsidy of order ¥1–2B). The Sano plant ramp and the Ru bulk-metal step-up (5–10x per-wafer content if Ru moves from liner to bulk interconnect at 2nm/A14) are the two transformative levers. PGM recycling grows as customer Ru-reclaim economics improve at high spot prices.
Leading indicators: Sano plant utilization (proxies semi orders); Furuya recycling-business growth (proxies Ru market tightness); spot Ru against a $1,200–1,500/oz threshold.
Bear case / risks
- Ru loses the A14 / 14A material-of-record decision to Mo or air-gap-only. Likelihood medium. Mitigant: multi-customer qualification + alternate use cases (HBM4 electrode, MRAM, SiC seed). Not fully closable — structural to a single-material thesis. What invalidates the bull thesis: any major foundry publicly choosing Mo over Ru for the A14 lowest-metal layer; or Ru spot sustained below $400/oz; or Sano utilization <50% at end-FY26.
- PGM price volatility crushes reported margins even if volumes grow. Likelihood high. Mitigant: Pt/Ir hedging on LBMA, pass-through clauses, price-renegotiation cycles. Manageable, not closable — Ru lacks liquid futures.
- Sano plant earns below cost of capital if ramp is slow. Likelihood medium. Closes when utilization hits 70%+; answerable in late 2027.
- HDD secular decline (~5–10%/yr unit decline). Likelihood med-high. Mitigant: HAMR adoption extends high-margin Ru-target life; pivoting Thin Film capacity toward semis at Sano. Closing — semis growing into the gap. (Note STF counter-view below.)
- Russian / South African PGM supply disruption. Likelihood medium. Recycling provides secondary supply, but mine supply is structurally concentrated — not closable. Anglo American Platinum and Norilsk offtakes are sanction-exposed.
- Customer concentration creep as semis grow. Likelihood low-med. Stated policy <10% per customer; could become real by 2028 if TSMC/Samsung order books grow large.
- Yen weakness/strength swings. Medium; natural USD revenue from semis is a partial offset; not closable.
Dilution risk: share count flat for a decade; no convertibles or warrants. FCF negative four years; debt has grown (¥21.7B total vs ¥6.5B cash). Base case: more debt, not equity (family aversion to dilution). Watch: if FY26 FCF doesn't turn positive and Ru spot stays >$1,000/oz, balance-sheet pressure builds — a dividend cut becomes the first lever, equity issuance the last (a dividend cut would itself signal stress).
Bear-case downside: ¥5,000–6,500 (~30–50% from current) if multiple risks materialize together. Bear-case probability ~25% (STF evidence nudges the bull-case probability up modestly, see below).
Portfolio correlation: Japanese small-cap, semi-thematic, USD/JPY-sensitive; ~0.4–0.5 correlation to a Japanese-semi basket (Tokyo Electron, Disco, Lasertec, Sumitomo Electric) — meaningful but not redundant. Worth checking against existing book.
Holder-base structure (a risk in itself)
Outside the founding family, the entire institutional top-10 is passive index or factor-quant — no thesis-driven active institutional holder. The 5x rally was driven by retail and Japanese semi-thematic ETFs, not institutional accumulation. Cuts two ways: no large active long to defend on a drawdown, but no "smart money has already bought" overhang either. Short interest not publicly reported on the TSE listing; anecdotally modest. No 13D activist filings.
STF Research counter-view (May 4, 2026)
STF Research ("Furuya Metal: Selling the World's Scarcest Metals") is structurally more bullish and surfaces two angles this analysis underweighted: (a) HDD/cold-data Ru content is rising, not falling — HAMR/MAMR and increasing platter counts add Ru per drive; Johnson Matthey identifies data-center HDD investment as the core driver of ruthenium's 2025 deep deficit, and Furuya's own deck confirms HDD demand "continues to be strong in data centers." (b) The AI / co-packaged-optics (CPO) chain drives Ir-crucible demand via InP laser-substrate growth (NVIDIA reportedly pre-allocated EML laser capacity past 2027) and Faraday-rotation crystals (YIG/BiIG) for optical isolators — a direct AI read on the iridium leg, not just legacy LED. STF also notes Applied Materials' Endura Volta Ruthenium CVD compresses liner thickness from ~30Å to ~20Å (33% reduction) and cuts line resistance up to 25%, and asserts "RuCo has been adopted by every major logic chipmaker for the 2nm node." Furuya's own FMBI quotations show Ru moving from ~$440 (mid-2024) to $920 (year-end-2025). Soft contradiction: STF treats 2nm RuCo liner adoption as decided; this analysis treats A14/14A material-of-record as still open — both agree bulk Ru at 1.4nm is roadmap-pointed but not committed. Net effect: nudge bull-case probability from ~25% to ~30%; treat HDD as a growing (not declining) Ru-content segment; add the InP-CPO → Ir-crucible chain as a discrete demand leg. STF does not change the entry-price calculus — the WATCH recommendation stands.
Valuation / DCF
Current multiples (2026-05-15, ¥9,640)
Forward P/E 24.1x, P/E (TTM) 31.3x, EV/EBITDA ~24x on FY25 reported EBITDA of ¥11.7B (the yfinance 159x reading is a data artifact), P/B 3.5x, EV/Sales 17.2x (distorted by metal-price headline revenue), dividend yield 2.92% (~31% normalized payout), FCF yield negative (FCF negative four years).
vs own history: pre-2024 the stock typically traded 12–18x forward P/E; current 24x is at the high end, post a 5x rally. vs peers: Mitsubishi Materials 12–15x, Johnson Matthey 10–12x, broader Japanese semi-materials peers 15–22x — Furuya is at a premium reflecting the Ru-thematic story. vs market: Topix forward P/E ~14–16x. The valuation implies the market is paying ~10x EBITDA for the current book plus another ~14x of multiple for future Ru content step-up — i.e., roughly that EBITDA grows 50–80% over 3–5 years from ¥11.7B to ¥18–21B, requiring Ru bulk-metal adoption by ~2028 and/or Sano ramping to capacity at current margins. Both plausible, neither certain.
Implied equity-required IRR (Gordon-growth back-out) at 24x forward and 31% payout is ~7–9% assuming 5% dividend growth — roughly Japanese small-cap fair compensation; the market is pricing normal returns + thematic optionality on the Ru step-up. No deep margin of safety, but not crazy.
Scenario framework (12-month)
Base case (¥9,500–11,500). Sano ramps modestly, PGM prices stabilize, gross margin recovers toward 28–30%, FY27E EBITDA ¥14B at 22x = ¥308B equity / ¥12,500; discount back to 12m: ¥10,000–11,000.
Bull case (¥14,000+). Probability ~25% (~30% after STF evidence). TSMC formally adopts Ru as material-of-record for the A14 lowest-metal layer; Furuya wins primary qualification; FY27E EBITDA ¥18B at 25x = ¥450B / ¥18,300.
Bear case (¥6,500). Probability ~25%. PGM prices roll over (Ru -30%), Sano ramp slow, A14 chooses Mo for contact fill; FY26 EBITDA stays ¥11–12B at 15x = ¥175B / ¥7,100. (Deeper bear variant: FY27 EBITDA ¥9–10B at 12–13x = ¥120B / ¥4,900 → downside target ¥5,000–6,500, ~30–50% from current.)
Probability-weighted target: ~¥10,300 (deep-dive). Profile/analyst average target ¥10,000 (high ¥11,500 / low ¥8,500, 2 analysts, consensus "Buy"). Current price ¥9,640 → expected return ~7%. Not a high-asymmetry trade at current price — a high-quality watchlist name. Would NOT buy at +10–15% (¥10,600–11,100): margin of safety already thin at ¥9,640.
Technical read
52w range ¥2,151–¥11,010, current ¥9,640 (~12% off high). Classic Stage 2 advance, 5x off the 2024 low; 50- and 200-day MAs well below price (uptrend intact); higher highs/higher lows; daily RSI ~50–55 (neutral, cooled from overbought). Support ¥8,500–9,000, then ¥7,500 (50-day MA area), major ¥5,000; resistance ¥10,500–11,000. Multi-month base/handle under the ¥11,000 prior high — constructive but not yet a breakout. Buying at ¥9,640 today is neither a breakout nor a discount — mid-range with no edge. Trade structure: scale-in, not all-at-once.
Decision log
2026-05-15 — Net-new swarm research initiated (profile, deep-dive, mgmt-dd, checklist all generated this date, Register D). Price reference ¥9,640.
2026-05-15 — Management DD verdict: Green-Yellow / B+. "Yes — I would trust these people with my capital." Clean forensic ledger: no shell games, no related-party self-dealing, no broken-promise pattern, no anti-takeover entrenchment, no dilutive raises. Family net-worth-concentrated in 7826 and not selling into the 5x rally — single most credible alignment signal. Red flags: none. Two Yellows (opaque incentive design, capital-allocation inertia) are structural to Japanese small-cap norms. The investment question is competence/pace (nimbleness on the Ru-at-A14 opportunity), not character.
2026-05-15 — Pre-buy checklist verdict: WATCH — partial scale-in possible but not a chase. Conviction medium. FundamentEdge hard rules: marginal pass — fails the 8–12% durable revenue-growth gate cleanly (a 5–8% organic grower with Ru optionality), passes the others (second derivative, valuation-not-a-thesis, quality-as-risk, estimate-revision direction mildly positive). Behavioral audit: FOMO acknowledged (stock 5x'd) and mitigated by scale-in plan and explicit bear case (~25% probability). Technicals: no — mid-range, no edge. Recommendation: build a starter only if Japan-semi-thematic exposure is missing from the book; otherwise wait for ¥7,500–8,000 pullback or A14 material-of-record confirmation before sizing up. Re-trigger on either event.
Position plan (if entering): conviction medium; target size 1.0–1.5% of equity portfolio; scale in — 30% at ¥9,500–9,700 (only if underweight Japan-semis), 30% at ¥8,000–8,500, 40% at ¥7,000–7,500 (on Ru spot weakness); bump to full size only on confirmed A14 Ru material-of-record news. Max acceptable loss −40% from average entry.
Exit / re-evaluate triggers: sell if Ru is publicly disqualified at A14/14A in favor of Mo; sell if Sano utilization stays <50% at end-FY26; trim if stock crosses ¥14,000+ (pricing the bull case before confirmation); re-evaluate if Ru spot rolls below $400/oz on supply normalization; dividend cut would signal balance-sheet stress. Add on: TSMC/Samsung public Ru-at-A14 commitment; HBM4 Ru-electrode commercial-volume confirmation; operating margin recovering to 22%+ on stable PGM prices.
2026-05-15 — SemiAnalysis cross-check (all reports): SA mirror has no Furuya-specific piece; IEDM 2022 coverage (semianalysis/2023/iedm2022p1.md) is on-thesis and confirms TSMC N3 Ru liner, Intel 10nm/Intel-7 Ru, and IBM-Samsung subtractive Ru at 18nm pitch, but frames timing cautiously ("the material that everyone hopes takes over"). No contradiction — SA's caution supports the WATCH-not-CHASE call over the chase-the-rally retail tape.
2026-05-15 — STF Research cross-check (deep-dive): STF's May 4, 2026 post is structurally bullish; surfaces HDD/cold-data rising-Ru-content and AI-CPO → InP → Ir-crucible angles. Action items: treat HDD as growing not declining for FY26–FY28 volume modeling; add InP-CPO Ir leg; nudge bull-case probability ~25%→~30%; WATCH recommendation stands (STF doesn't change entry-price calculus).
Net stance as of consolidation (2026-05-30): WATCH. Real story, high-trust management, genuine moat, credible A14 Ru-content path — but stock has 5x'd, ~24x forward P/E with minimal margin of safety, FundamentEdge growth-primacy fails, mid-range technicals. Not a chase at ¥9,640.
Sources
Consolidated 2026-05-30 from four research fragments (all Register D, generated 2026-05-15), folded into this canonical page:
7826-profile.md— company profile7826-deep-dive.md— full deep-dive (incl. STF Research cross-check)7826-mgmt-dd.md— management due diligence (NongAap "Dark Arts" forensic lens)7826-checklist.md— pre-buy checklist (FundamentEdge hard rules)
External / primary sources cited across fragments:
- yfinance market data, financials, holders, and major-holders data (2026-05-15 pull)
- Furuya FY22–FY25 results presentations and Yuho summaries, https://www.furuyametals.co.jp/ir/ and /ir/library/ (mid-term plan FY24–FY27 on same page)
- TSE Corporate Governance Report; EDINET (FSA) Japanese-language registry summaries
- SemiAnalysis "TSMC 3nm FinFlex + IBM VTFET + Ru Interconnects" (IEDM 2022 coverage, mirrored 2023 at
~/Dropbox/Wafflebun/KB/wiki/semianalysis/2023/iedm2022p1.md) - STF Research, "Furuya Metal: Selling the World's Scarcest Metals," May 4, 2026 (paid tier, early-access for founding members), https://stfbutnou.substack.com/p/furuya-metal-selling-the-worlds-scarcest — citing Johnson Matthey, SFA Oxford / WPIC, SEMI, Applied Materials Endura Volta Ruthenium CVD
- TECHCET, SEMI public summaries on Ru-target TAM; IEDM 2022 conference papers (IBM-Samsung subtractive Ru; AMAT barrierless tungsten)
- LBMA PGM price history (2024–2026 spot moves); Nikkei industry coverage of PGM markets and Japanese semiconductor-materials supply chains
- Frameworks applied as analytical lenses: NongAap "Dark Arts" governance forensics; FundamentEdge / Brett Caughran methodology
Filings note. /filings not run (sec-monitor is US-EDGAR oriented; Furuya files via EDINET/Yuho). Follow-up via EDINET FY25 Yuho recommended to firm up: (1) segment-level operating margins, (2) PGM inventory carrying value at year-end, (3) precise major-shareholder breakdown beyond yfinance, (4) post-September-2025-AGM board changes.
Prior topic backlinks from fragments (for relink pass): ai-infrastructure, optical-components, japan-semi, supply-chain-security, critical-minerals.
Consolidation queue (merged 2026-05-30)
Folded in from these files; they stay live pending Pink's archive confirm.
- [ ]
7826-mgmt-dd.md - [ ]
7826-checklist.md - [ ]
7826-profile.md - [ ]
7826-deep-dive.md