3436 — SUMCO Corporation
Thesis
Verdict: PASS at ¥3,994 (profile, 2026-05-30). The inflection is already in the tape, with essentially no margin of safety. SUMCO is the world's #2 silicon-wafer maker (~21-24% share, behind Shin-Etsu Handotai) and a near-perfect pure-play: essentially 100% of revenue is semiconductor silicon. That purity is the whole investment character — an unhedged, high-operating-leverage bet on the wafer cycle and on wafer pricing, with none of the diversification that cushions its larger rival. The bull case (leading-edge 300mm for AI logic and HBM/DRAM tightening) is real but node-specific and not yet in the P&L.
Two corrections anchor the honest read, both material:
- There is no SUMCO "Chitose" fab. Chitose (Hokkaido) is Rapidus, the unrelated 2nm logic foundry. SUMCO's domestic 300mm expansion was in Saga (Imari/Yoshinogari), and it was shelved on 6 April 2026 in favour of upgrading existing equipment, with the METI subsidy cut from up to ¥75B to ¥19.3B. A company genuinely convinced of a durable, imminent recovery does not throttle its own expansion — this is management's own tell.
- The "81× forward P/E" is a data artifact. yfinance carries a corrupted positive forward EPS (+¥49.23). Real consensus has SUMCO loss-making again in FY2026 and profitable only in FY2027 (~¥23B). The stock capitalizes a forecast loss, not a recovery the Street already underwrites.
The honest valuation read. Strip out the mirage multiple and the reality is starker: the stock has no positive forward earnings to anchor any multiple; consensus models another loss in FY2026 and profit only in FY2027, yet the equity has run +233% over 52 weeks to ~56% above the ¥2,562 mean PT, at 2.45× book, right after a 64% Q1 miss (Q1 FY2026, reported 12 May 2026). SUMCO is the higher-torque vehicle for an investor who is early-and-right on a still-undiscounted 2027 inflection — it is not that today. Shin-Etsu (4063.T) is the more defensible way to own the same theme: duopoly wafer leadership wrapped in net cash, through-cycle profitability, and lower earnings volatility, at a genuine forward multiple. Per dollar invested, SUMCO gives ~4-5× the wafer exposure of Shin-Etsu — concentrated torque, concentrated risk.
(Note: this page consolidates a single /profile fragment. No /deep-dive, /dcf, or /checklist has run — there is no formal buy/sell call or price-target zone yet. The PASS framing is the profile's analytical read, not a checklist verdict.)
Snapshot
One-liner: The world's #2 silicon-wafer maker and the purest listed pure-play on the wafer cycle — ~100% semiconductor silicon, ~85%+ 300mm, riding the AI advanced-node tailwind but still loss-making, with the recovery already priced and no margin of safety.
- Ticker / exchange: 3436 on the Tokyo Stock Exchange (Prime); GICS Semiconductor Materials & Equipment. Yahoo carries it as 3436.T
- Legal name: SUMCO Corporation; HQ Minato-ku, Tokyo; formed 1999 as the Sumitomo Metal Industries / Mitsubishi Materials silicon JV ("Sumitomo Mitsubishi Silicon"), renamed SUMCO in 2005; listed on the TSE late 2005 (raised ~$1.5B); sumcosi.com/english
- Fiscal year: December (FY2025 = year ended 31 Dec 2025). Shin-Etsu runs a March year-end, so the two benchmarks sit on offset calendars and SUMCO's reported trough is ~nine months fresher than Shin-Etsu's last full year.
- Spot price: ¥3,994 (2026-05-29 close); +131% YTD, +233% over 52 weeks, at the 52-week high (52-wk range ¥937–¥4,047)
- Market cap: ¥1,397B (~$8.8B) | EV ~¥1.7T (~$10.9B) | net debt ~¥264-278B (~$1.7B)
- Valuation: P/E (TTM) n/a — loss-making (trailing EPS −¥33.48) | Forward P/E no meaningful figure (yfinance's 81× rests on a corrupted positive forward EPS; real consensus models a FY2026 loss) | EV/EBITDA 16.0× (soft — EBITDA flattered by ¥116B depreciation on near-zero operating profit) | P/B 2.45× | Dividend yield 0.6% (¥20/share, funded from the balance sheet against a net loss)
- Price vs analyst mean PT: ~+56% above the ¥2,562 mean (16 analysts, range ¥1,300–¥4,200); the rally has blown past consensus, and even the upgraded targets sit well below spot
- Ownership: insiders ~3.9% (no controlling founder family); institutions ~62.6%, foreign-institution-heavy; float ~98%
- Conviction: PASS at spot (profile read) — the 2027 inflection is in the tape with no margin of safety; Shin-Etsu the more defensible vehicle for the same theme
Business
Business model. SUMCO grows ultra-pure single-crystal silicon ingots and slices, polishes, and finishes them into the wafers on which every chipmaker builds logic, DRAM/HBM, and NAND — predominantly 300mm, with a shrinking 200mm-and-below tail, sold to IDMs, foundries, and memory makers, largely under multi-year long-term agreements (LTAs). Revenue is volume × price of wafers shipped (one-time product sales, not recurring) plus a smaller stream of processing and reclaim services.
The commercial backbone is LTAs: a large share of 300mm and 200mm volume is locked under multi-year, take-or-pay-style contracts at committed prices. Management confirmed for both Q4 2025 and Q1 2026 that LTA prices held even through the trough, protecting ASPs while spot and legacy demand sagged. The flip side, visible in 2024-25: in a downturn customers defer LTA deliveries, which is precisely what gutted utilization and earnings.
The economics are brutally fixed-cost-heavy, so margins swing with utilization. Gross margin ran from 32.5% at the FY2022 peak to 13.3% in FY2025. The tell of a freshly built-out cyclical at trough utilization: FY2025 depreciation of ¥115.6B exceeded operating profit entirely, so EBITDA margin held at a healthy-looking 27.4% while the operating line was barely above zero and the net line was a loss. A roughly ¥1.1-1.2B operating-profit hit per ¥1 of yen appreciation makes it FX-sensitive on top of cyclical.
Segment mix (single reported segment). SUMCO reports as one segment; no audited revenue split by diameter is disclosed. The following are management commentary and third-party estimates, not segment accounts:
- 300mm wafers — est. ~85%+ of revenue. The dominant line and the only one exposed to the AI cycle. Management says leading-edge 300mm (for 2nm/3nm logic and HBM/DRAM) is roughly 20% of 300mm revenue; the larger remainder is mature 300mm where demand and pricing recovery lag.
- 200mm wafers — mature and structurally shrinking. Pressured by the customer shift to 300mm and by Chinese supply. SUMCO is exiting: it will end wafer production at its Miyazaki plant by end-2026, converting the site to ingot-only and redeploying staff to 300mm.
- 150mm and smaller — in wind-down. Transferred to other Japanese sites and to Indonesia.
Customers & concentration. SUMCO does not name customers or disclose concentration percentages (no SEC filing applies; TDnet/EDINET does not break it out). The customer base is the global leading-edge fab roster, inferred by industry structure:
- Logic / foundry (inferred): TSMC, Intel, GlobalFoundries, Samsung Foundry — highest-margin leading-edge epitaxial 300mm pull.
- DRAM (inferred): Samsung, SK Hynix, Micron — HBM stacks feeding AI GPUs, the active bright spot.
- NAND (inferred): Kioxia, SK Hynix, Samsung; historically YMTC (China) — recovering for AI inference/storage.
Buyer power is structurally high — a handful of mega-fabs dictate volumes and, in downturns, defer LTA deliveries (the proximate cause of the FY2024-25 collapse). The sharpest single-customer flag: SUMCO's CEO acknowledged that business with its major memory client (believed to be YMTC) "declined substantially" as China localized wafer supply. That is the textbook bad case — the same Chinese demand that once anchored SUMCO's NAND volume is now the low-cost competitor displacing it. Concentration is qualitatively material but not quantified in public disclosure.
Operations footprint. Manufacturing is concentrated in Japan (seven domestic sites), with overseas production via FST in Taiwan, an Oregon fab (US), and Indonesia for small-diameter:
| Site | Role | Status |
|---|---|---|
| Imari / Saga (Nagahama, Kubara, Saga) | Core 300mm hub — ingot growth + 300mm wafers | Operating; FY2025+ capex now goes to equipment upgrades, not a new building |
| Yonezawa (Yamagata) | Ingot growth + polished-wafer processing | Operating |
| Miyazaki (SUMCO TECHXIV) | Small-diameter wafers | Wafer production ends by end-2026; converts to ingot-only |
| Indonesia | Small-diameter wafer production | Receiving transferred volume |
| SUMCO USA (Oregon) | 150/200/300mm polished + epitaxial for North America | Operating |
| Formosa Sumco Technology (FST), Taiwan | JV 300mm/200mm production + sales base | Operating; being deconsolidated (see JVs) |
| Saga (Yoshinogari) new fab | Planned >¥400B greenfield 300mm expansion | SHELVED 6 Apr 2026 — subsidy cut ¥75B → ¥19.3B |
Joint ventures & partnerships.
- Formosa Sumco Technology (FST; TWSE: 3532) — JV with Taiwan's Formosa Plastics Group (FPG), established 1995, producing 300mm/200mm wafers in Taiwan. SUMCO held 43.2% (consolidated) via its wholly-owned subsidiary SUMCO TECHXIV. On 19 Feb 2026 SUMCO resolved to sell ~5.2%, cutting its stake to 38.0% and reclassifying FST from a consolidated subsidiary to an equity-method affiliate. This will reduce reported group revenue once effective — a structural step-down in the headline top line that is easy to misread as organic decline.
- Customer LTAs — multi-year take-or-pay-style volume and price commitments with major IDMs/foundries/memory makers; terms and counterparties not itemized publicly.
- No other material standalone JVs. SUMCO is itself the legacy product of the Sumitomo/Mitsubishi silicon merger.
Competitive position & moat. The wafer industry is a disciplined five-player oligopoly: Shin-Etsu Handotai (#1), SUMCO (#2), GlobalWafers, Siltronic, SK Siltron, with the two Japanese names as the duopoly core of 300mm. Market share (third-party estimates, ranges wide): Shin-Etsu ~27-30% (#1), SUMCO ~21-24% (#2), GlobalWafers ~17%, Siltronic ~12%, SK Siltron ~9%; top-5 control ~80-85%. SUMCO's own share is the one figure sources disagree on most (estimates span 21% to 27%); ~21-24% is the more defensible read.
Moat: (1) capital intensity — a new 300mm line costs hundreds of billions of yen; (2) qualification lock-in — each spec is co-developed and qualified per customer per node over 12-24 months; (3) process IP — defect density, flatness/nanotopography, epitaxial uniformity at the leading edge; (4) oligopoly discipline via LTAs. Porter's five forces: rivalry moderate among disciplined incumbents, intensifying from Chinese entrants in legacy grades; new entrants low at the leading edge (capital + qualification), actively breached in mature grades by state-backed Chinese capacity; supplier power moderate (polysilicon, quartz crucibles, slurry — specialized, multi-sourced); buyer power high and concentrated; substitutes essentially none for silicon at scale (SiC/GaN address only niche power/RF). The moat holds at the leading edge; it is being breached in legacy — Chinese entrants (NSIG, Shanghai Simgui, TCL Zhonghuan) run ~1M wafers/month at 10-15% below Japanese pricing.
Financials
Income statement (December fiscal years; FY2026E = sell-side consensus, NOT company guidance):
| Metric | FY2023 | FY2024 | FY2025 | FY2026E |
|---|---|---|---|---|
| Revenue | ¥425.9B | ¥396.6B (trough) | ¥409.7B (+3.3%) | ~¥429B |
| Gross margin | 25.4% | 18.3% | 13.3% | – |
| Operating income | ¥73.1B | ¥35.1B | +¥1.3B (0.3%) | small/negative |
| Net income | ¥63.9B | ¥19.9B | −¥11.8B | −¥17.7B (loss) |
| EPS | ¥182.59 | ¥56.84 | −¥33.60 | ~ −¥50 |
The arc is a clean cyclical roll-over from a ~¥200 EPS peak in FY2022 into losses. The nuance that matters: FY2025 operating profit was barely positive (+¥1.3B), but the net line was a ¥11.8B loss driven by depreciation and below-the-line items. The trough then deepened into FY2026: Q1 (reported 12 May 2026) was an operating loss of ¥5.27B and a net loss of ¥8.47B, EPS −¥24.22, missing consensus by ~64%. H1 FY2026 is guided to a ¥7.7B operating loss / ¥15.4B net loss, and the company issued no full-year FY2026 forecast — it is guiding half a year at a time. Consensus does not see profit return until FY2027 (~¥23B).
Cash flow & balance sheet (December fiscal years):
| Metric | FY2023 | FY2024 | FY2025 |
|---|---|---|---|
| Operating cash flow | ¥96.3B | ¥69.6B | ¥100.0B |
| Capex | −¥256.9B | −¥247.2B | −¥111.0B (rolling over) |
| Free cash flow | −¥160.6B | −¥177.6B | −¥11.0B |
| Net debt | ¥68.1B | ¥258.3B | ~¥264-278B (~$1.7B) |
| Equity ratio | – | – | 51.3% |
The cash-flow story is the late-cycle squeeze: a ~¥250B/yr capex super-cycle in FY2023-24 (funding the 300mm build-out) collided with collapsing margins, burning ~¥160-178B of FCF a year and turning a net-cash balance sheet into ~¥264-278B net debt. Capex is now "past the peak" and falling (FY2026 guided ~¥50-60B), so FCF is nearly back to breakeven, but the leverage built up at the worst possible time.
Valuation (as of 2026-05-29). Price ¥3,994 (+19.3% on the day, at the 52-week high); market cap ¥1,397B (~$8.8B); EV ~¥1.7T (~$10.9B). P/E (TTM) n/a — loss-making (trailing EPS −¥33.48); forward P/E no meaningful figure (yfinance shows 81× on a corrupted positive forward EPS; real consensus models a FY2026 loss); EV/EBITDA 16.0× (soft — EBITDA flattered by ¥116B depreciation on near-zero operating profit); P/B 2.45×; dividend yield 0.6% (¥20/share, funded from the balance sheet against a net loss). At ¥3,994 the stock trades ~+56% above the ¥2,562 mean PT and above all but the single highest target.
Industry landscape
There is no chip without a wafer, and no substitute material at scale, so the wafer makers are the true floor of the semiconductor stack. But the cycle is bifurcated, not a rising tide. Leading-edge 300mm (ultra-flat, low-defect polished and epitaxial wafers for 2nm/3nm logic and HBM-grade DRAM) commands premium pricing and tight per-node qualification, and is genuinely tightening on AI demand. Commodity 200mm and legacy 300mm (analog, power, MCU, automotive) sit oversupplied, with much of that volume ceding to Chinese makers. SUMCO is leveraged to both, which is why the AI narrative and the loss-making reality coexist.
- End markets: logic/foundry (strongest, highest margin), DRAM/HBM (active AI beneficiary), NAND (recovering, but the China-localization wound), power/analog/legacy (weak, being exited).
- TAM (SEMI, 2025 actuals): global silicon-wafer area shipments 12,973 MSI (+5.8% YoY) on revenue $11.4B (−1.2% YoY). The volume-up / revenue-down divergence is the story — more area shipped, lower revenue, because weak legacy pricing offset AI-driven strength in advanced wafers. SEMI projects a new record ~15,485 MSI by 2028.
- SAM: 300mm is the addressable core (~64%+ of area volume, higher share of revenue); the Shin-Etsu + SUMCO duopoly holds >50% of global 300mm volume.
- Secular tailwind / threat: AI capex is the demand engine, but Chinese entrants (NSIG, Shanghai Simgui, TCL Zhonghuan) run ~1M wafers/month at 10-15% below Japanese pricing, already displacing incumbents in mature grades.
(No standalone sector page exists yet for semiconductor-materials wafers; if one is created, link up here.)
Management
A March 2026 board reshuffle reset the top of the house:
| Name | Title | Note |
|---|---|---|
| Jiro Ryuta | Chairman of the Board + President (Representative Director) | Promoted from EVP at the 27 Mar 2026 AGM; combined chair/president role (independence weakness) |
| Shinichi Kubozoe | Executive Vice President, Rep. Director | Finance leadership likely sits here; no separately named CFO is disclosed |
| Naruya Hirota | Executive Vice President, Rep. Director | |
| Mayuki Hashimoto | Director / Chief Executive Adviser | Long-time Chairman & CEO (~2012-2026); stepped back but retained a board seat |
| Toshihiro Awa | Former President | Retired at the Mar 2026 AGM |
Board: a Company-with-Audit-and-Supervisory-Committee structure, 13 directors, ~46% independent (6 of 13 independent outside directors, including three women) — below a majority-independent board but typical for the structure. The Nomination & Remuneration committee is advisory only. Board met 16 times in FY2025 at 100% attendance. ISS Governance QualityScore 5/10 (Audit 1, Board 4, Shareholder Rights 6, Compensation 5) — middling, weakest on shareholder rights.
Alignment: insider ownership ~3.9% (yfinance aggregate) — there is no controlling founder family, as SUMCO is a merger of corporate silicon units. Institutions hold ~62.6%, foreign-institution-heavy. The founding parents (Mitsubishi Materials, Sumitomo) have exited the top 10 — SUMCO is now a widely held, foreign-institution-heavy float with no strategic anchor and no visible government/METI equity holder. No material recent insider transactions surfaced (would require TDnet/EDINET officer filings to confirm).
Governance flags: the combined chair/president role; a board that is not majority-independent; and a definitive read on cross-shareholdings and any takeover defense requires the EDINET Securities Report, which was not retrieved — treat "none" as unverified. Key-person risk: low and just reduced — the long-tenured Hashimoto stepped back in March 2026 with an orderly internal succession to Ryuta; no founder dependency.
Top holders (as of 31 Dec 2025). The register is dominated by custodial/trust nominees, so apparent concentration overstates real economic ownership:
| Holder | Type | Who they are | % |
|---|---|---|---|
| Master Trust Bank of Japan | Trust nominee | Aggregates many domestic pension/funds — not one owner | 16.67% |
| Custody Bank of Japan | Trust nominee | The other major Japanese trust nominee | 10.66% |
| BNY / JPMorgan / State Street accounts | Foreign custodial nominees | Aggregate foreign institutional holdings | ~4-5% each combined |
| Fidelity (Global Technology Pool, via BBH Lux) | Active fund | A genuine economic holder | 2.14% |
De-dup note: ranks 1-2 and the State Street / JPMorgan / BNY lines are nominees pooling dispersed beneficiaries, not direct stakes.
Catalysts & risks
Growth drivers (the bull case — real but node-specific and not yet in the P&L): leading-edge 300mm for AI logic and HBM/DRAM is tightening, inventory days are normalizing (they fell in Q1), and the exit from 200mm (Miyazaki by end-2026) concentrates the asset base on AI-grade 300mm. The supply side is rationalizing too — SUMCO and Shin-Etsu jointly invested ~¥150B in 2025 to add ultra-flat 300mm capacity for 2nm/3nm.
Management's own tell (the most important capex datapoint — a negative one): on 6 April 2026 SUMCO shelved its two planned new Saga fabs (Imari/Yoshinogari, ~¥225B planned), opting to upgrade existing equipment instead, and the METI subsidy was cut from up to ¥75B to ¥19.3B. A company genuinely convinced of a durable, imminent recovery does not throttle its own expansion. SUMCO's published outlook has been consistently more cautious than its customers, its competitors, and the sell-side — a contrast independent analysts (Morningstar, Semicon Alpha) have flagged against the +233% tape. (No standalone R&D-as-%-of-revenue is broken out; R&D is embedded in the wafer process. No M&A; the only corporate action is the FST deconsolidation.)
Risks (structured):
| Risk | Likelihood | Mitigants | Can it be closed? |
|---|---|---|---|
| Cycle-timing / recovery slips to 2027+ | High | LTAs hold price; inventory normalizing; AI demand real | No — structural cyclicality; only managed. Closes only when utilization and pricing actually recover in the P&L |
| Valuation with no earnings to anchor it | High | None | Closes only via a real FY2026-27 earnings inflection or a price reset. The stock capitalizes a forecast loss |
| Chinese legacy-grade encroachment | High (already realised) | Leading-edge moat; 200mm exit | No — structural; SUMCO is retreating up-market rather than defending legacy |
| Customer concentration / LTA deferral | Medium-High | Multi-year LTAs; 200+ customers | No — buyer power is structural; deferral risk recurs every downcycle |
| Balance-sheet leverage at the trough | Medium | Equity ratio 51.3%; capex now falling; OCF positive | Yes, gradually — net debt falls as capex rolls over and (if) margins recover |
| FX (yen strength) | Medium | – | No — hedgeable but not eliminable; ~¥1.1-1.2B OP per ¥1 |
Dilution risk: Low. SUMCO's last major raise was the October 2021 offering — ¥109.4B, 52.17M new shares, ~18% dilution — which funded the Imari 300mm expansion (this corrects the "2023 dilution" premise from the original brief). Share count stepped from ~290M to ~350M then and has been flat-to-slightly-down since; no further equity raises and no convertibles/warrants outstanding. Post-2021 capex was debt- and cash-funded, which is why the leverage sits on the balance sheet rather than the share count. With FCF near breakeven and capex falling, near-term equity dilution risk is low, though a prolonged trough would pressure the dividend before the share count.
Recent developments:
- 29 May 2026: +19.3% to ¥3,994 (52-week high). Catalyst was sector-wide, not company-specific — multiple brokerages revised AI-server silicon-wafer and MLCC demand forecasts up, citing improving H2-2026 order visibility (Nikkei +2.5%, SUMCO and Taiyo Yuden leading).
- 21 May 2026: UBS upgraded Sell → Neutral, PT ¥1,050 → ¥3,100 — explicitly a sentiment / downside-removal call; UBS still said near-term LTA price increases "remain unlikely."
- 12 May 2026: Q1 FY2026 miss (operating loss ¥5.27B, net loss ¥8.47B, EPS −¥24.22, ~64% below consensus); H1 loss guidance. Stock dipped ~3% then resumed the rally.
- 6 April 2026: Shelved the two new Saga fabs; subsidy cut to ¥19.3B.
- 19 February 2026: FST deconsolidation resolved (43.2% → 38.0%).
- 10 February 2026: FY2025 results; dividend cut to ¥20; Miyazaki wafer-production wind-down confirmed.
- Next earnings: ~12 August 2026 (H1 FY2026), per the Investing.com calendar; SUMCO confirms exact dates ~2 weeks prior.
Analyst sentiment: 16 analysts, consensus tilted Buy; mean PT ¥2,562, range ¥1,300–¥4,200. At ¥3,994 the stock trades ~56% above the mean target and above all but the single highest. UBS (Neutral, ¥3,100) and Nomura (Buy, ¥2,100, on channel-check demand into Dec-2026) are the recent movers. Consensus is being revised up underneath the price, so the PT gap reads partly as stale rather than purely as a fundamental ceiling — but even the upgraded targets sit well below spot.
Valuation / DCF
No formal DCF has been run (/dcf has not been invoked). The profile-level read:
SUMCO's headline "81× forward" is a mirage built on a corrupted data input (yfinance +¥49.23 forward EPS). Strip it out and the reality is starker — the stock has no positive forward earnings to anchor any multiple; consensus models another loss in FY2026 (~¥17.7B, EPS ~−¥50) and profit only in FY2027 (~¥23B), yet the equity has run +233% over 52 weeks to ~56% above the ¥2,562 mean PT, at 2.45× book, right after a 64% Q1 miss. SUMCO is the higher-torque vehicle for an investor who is early-and-right on a still-undiscounted 2027 inflection. It is not that today — the inflection is already in the tape, with essentially no margin of safety. Shin-Etsu is the more defensible way to own the same theme: duopoly wafer leadership wrapped in net cash, through-cycle profitability, and lower earnings volatility, at a genuine forward multiple (23.5×).
Benchmark — SUMCO vs Shin-Etsu Chemical (4063.T): the wafer duopoly, opposite risk profiles. Shin-Etsu is the leader, not SUMCO, and the two are not equivalent investments despite "owning the wafer duopoly":
| SUMCO (3436.T) | Shin-Etsu Chemical (4063.T) | |
|---|---|---|
| What you buy | ~100% silicon wafers (pure-play) | Diversified chemicals giant; wafers are ~one-fifth |
| Wafer rank | #2 (~21-24%) | #1 via Shin-Etsu Handotai (~27-30%) |
| FY-end | December | March |
| Price / cap | ¥3,994 / ~$8.8B | ¥7,758 / ~$90.5B |
| Net margin (latest FY) | net loss | 18.4% |
| ROE | −3.5% | 10.7% |
| Balance sheet | net debt ~¥264-278B (~$1.7B) | net cash ~¥1.4T (~$8.9B) |
| Forward P/E | n/a (forecast loss) | 23.5× |
| Dividend | cut to ¥20; FY26 year-end undecided | 1.44% yield, growing |
| 52-week move | +233% | +58% |
Shin-Etsu's segment mix (FY ended Mar 2025): Infrastructure Materials (PVC + chlor-alkali; Shintech is the world's largest PVC producer) 40.7% of sales; Electronics Materials (silicon wafers + rare-earth magnets + photoresist + photomask blanks + synthetic quartz) 36.5% of sales and the largest profit pool at ~35% segment margin; Functional Materials (silicones) 17.5%; Processing 5.3%. Shin-Etsu never discloses a clean wafer-only line — wafers are folded inside Electronics Materials, estimated at ~18-25% of group revenue (analyst estimate, not a company figure). So per dollar invested, SUMCO gives roughly 4-5× the wafer exposure of Shin-Etsu.
Why it matters: a wafer-price upturn moves SUMCO's entire P&L violently (concentrated torque, concentrated risk), while it moves only a fraction of Shin-Etsu's earnings, cushioned by three other profit engines. Shin-Etsu's decisive structural edge is the balance sheet: with ~$8.9B net cash and 18% net margins through the trough, it can fund counter-cyclical 2nm/3nm wafer capex and out-invest a net-debt, dividend-cut SUMCO precisely when the pure-play is weakest. Shin-Etsu Handotai also holds the technology lead in 300mm prime, ultra-low-defect, and SOI wafers.
Decision log
2026-05-30 — Profile (price data as of 2026-05-29 close). PASS at ¥3,994 (analytical read, not a checklist verdict). Established that SUMCO is the purest listed pure-play on the wafer cycle (~100% silicon, ~85%+ 300mm), the world's #2 wafer maker at ~21-24% share. Two material corrections to the originating brief: (1) no SUMCO "Chitose" fab — that is Rapidus; SUMCO's Saga (Imari/Yoshinogari) expansion was shelved 6 Apr 2026, METI subsidy cut ¥75B → ¥19.3B (management's own caution tell); (2) the "81× forward P/E" is a yfinance data artifact (corrupted +¥49.23 forward EPS) — real consensus models a FY2026 loss (~−¥50 EPS) and profit only FY2027 (~¥23B). The equity has run +233% over 52 weeks to ~56% above the ¥2,562 mean PT, at 2.45× book, right after a 64% Q1 miss (Q1 FY2026 op loss ¥5.27B / net loss ¥8.47B / EPS −¥24.22, reported 12 May 2026). The inflection is already in the tape with essentially no margin of safety. Shin-Etsu (4063.T) framed as the more defensible vehicle for the same theme — duopoly leadership wrapped in net cash, through-cycle profit, genuine forward multiple — though SUMCO is the higher-torque vehicle for an investor early-and-right on a still-undiscounted 2027 inflection.
(No /deep-dive, /dcf, or /checklist has run. No formal entry/exit zones or position plan exist yet — add on the next research pass.)
Sources
Fragments folded into this canonical page (consolidated 2026-06-02; original archived to _migration-archive/2026-06-02/3436/): 3436-t-profile.md (/profile, 2026-05-30; price data as of 2026-05-29 close).
Data caveats (honesty log, carried from the profile):
- Forward EPS: yfinance carries +¥49.23 (the source of the bogus 81× P/E). Independent consensus (SimplyWallSt, MarketScreener) shows a FY2026 net loss (~¥17.7B, EPS ~−¥50), profit only FY2027. The corrupted figure is not used on this page's forward column.
- FY2025 operating result: small operating profit (+¥1.3B), net loss (−¥11.8B). The live feed's "−7.3% operating margin" is a trailing-twelve-month artifact (includes the Q1 FY2026 operating loss), not the fiscal-year figure.
- Market share: SUMCO ~21-24% (sources span 21-27%); leadership ranking (Shin-Etsu #1, SUMCO #2) is solid, point estimates are not.
- Segment / geographic splits, customer names, wafer-% of Shin-Etsu: estimates or inference, not company disclosure — flagged inline.
- Net debt ranges ¥264-278B by balance-sheet date and lease/ST-investment definition.
- "Chitose" and "2023 dilution" premises from the originating brief were both factually wrong and are corrected above (Saga not Chitose; Oct-2021 not 2023).
SemiAnalysis cross-check: the local SA mirror has no dedicated SUMCO or Shin-Etsu coverage — only tangential 2022 pieces on Disco and Intel tooling that mention wafers in passing. No contradiction to flag. SA is one input, never authoritative; here it simply has no view on these names.
Related vault pages: ai-infrastructure
Key external sources: SUMCO IR / FY2025 results presentation (10 Feb 2026) + FST and management disclosures (TDnet); SEMI 2025 wafer shipment data; Shin-Etsu Annual Report 2025 Financial Section; Simpson Thacher (2021 offering); TrendForce / Nikkei / MLex (Saga fab + subsidy); Investing.com / TipRanks (FY2025 + Q1 FY2026 results, UBS/Nomura ratings); SimplyWallSt / MarketScreener (forward consensus); Mordor / BusinessResearchInsights / semiconductorinsight (market share); yfinance + Google Finance (prices, verified 2026-05-29).