6981 — Murata Manufacturing Co., Ltd.

Thesis

Stance: cautious bull / BUY (scale-in, half-size initial). Conviction: Medium. Murata is the highest-quality way to own the MLCC oligopoly into an AI-server-led upcycle — the #1 global MLCC maker (~34% share), owner of the high-end of every cyclical bucket (auto, AI compute, premium phones), with a fortress balance sheet (¥598B net cash) and the deepest moat in passives. The stock has run 3.3× off the FY25 low on AI-thesis re-rating and is no longer cheap, but earnings are inflecting (Q3 FY26 op income +12.6% YoY; FY27 guide +35% / +34.8% op income), capacity is tightening, and Murata is reportedly considering price increases on AI-server MLCCs (Feb-2026 Digitimes). The single most important fact: forward P/E 28.4× embeds a sharp recovery that the second-derivative data confirms is already underway (Sep-25 op margin 21.3% = strongest in two years).

What has to be true. AI server MLCC volumes must track the company's stated 3.3× FY25→FY30 forecast (~27% CAGR). If that growth materializes and is paired with mid-cycle pricing discipline (no new capacity flooding), op margin recovers to 22%+ (FY22 peak was 23.4%) and EPS roughly doubles vs the FY24 trough. This is a margin-recovery + mix-shift thesis, not a pure top-line compounder thesis — organic revenue CAGR is only ~3%, and AI server is +27% CAGR but only ~10% of revenue today, not enough to drag the full top line to 8–12%.

The honest catch. The stock is consensus and the sell-side mean PT is ¥5,203 vs spot ¥6,733 — analysts sit 23–28% below the current price. Either targets are stale (likely, set before the Dec-2025 IR Day re-rating and Feb-2026 capacity reports) or the market has run ahead of fundamentals. Treat Murata as a quality compounder being re-rated, not a cyclical bottom-fish — entry timing matters. FundamentEdge hard-rule pass count: 3 of 5 (fails on modest absolute revenue growth and ambiguous estimate-revision direction; both real, neither disqualifying). Stock sits at all-time-high ¥6,740 after the run, so the recommended action is scale-in entry, not full size now.

Snapshot

One-liner. Murata makes the tiny ceramic capacitors that keep modern electronics from frying themselves — a smartphone has hundreds, an EV has thousands, an Nvidia GB200 baseboard now eats 15,000–25,000 — and it is the largest, highest-quality, highest-margin maker in a 5-player global oligopoly. The closest analogue to TSMC in the passives industry.

Identity. Murata Manufacturing Co., Ltd. — Ticker 6981 (TSE Prime) / MRAAY (ADR) / MRAAY (yf surfaces as 6981.T). GICS Technology — Electronic Components. HQ Nagaokakyo, Kyoto, Japan. Founded 1944; Tokyo Stock Exchange listing 1963. Employees 74,302 (FY2025). FY ends March 31 (FY26 = year ending March 2026).

Valuation snapshot (2026-05-20).

Metric	Value
Share price	¥6,733 (¥6,735 per profile)
Market cap	¥12.26T (~$78B at ¥157/$)
Enterprise value	¥10.62T (~¥1.64T net cash reduces EV)
P/E TTM	52.8×
Forward P/E (NTM)	28.4×
EV/EBITDA TTM	~20.8×
EV/Revenue	~6.1×
P/B	4.51×
P/FCF (TTM)	~107× — depressed by capex
FCF yield	~0.9% (TTM)
Dividend yield	1.14%
52-week range	¥2,018 – ¥6,740 (spot at ATH; +234% off the low)

12-month target ¥7,500–8,000 (my range; sell-side high ¥7,300; sell-side mean ¥5,203 = the downside, not the base). Expected return +12 to +20% over 12 months (base case), pre-dividends.

Segment mix. ~45% MLCCs, ~22% communication modules, ~10% batteries, ~23% inductors/EMI/sensors/timing. Exports ~90% of production from Japan; ~75% of revenue ships via China assembly hubs for global onward shipment. Geographic revenue (FY2025): China 51%, Japan ~15%, rest of Asia 18%, Americas 9%, EMEA 7%.

Business

The franchise. Murata develops, manufactures and sells ceramic-based passive components (MLCCs, inductors, EMI filters), RF modules (SAW filters, front-end modules, connectivity), lithium-ion coin/cylindrical batteries, sensors, and timing devices. Manufacturing model — high-mix, high-volume, fab-like capital intensity. Pricing is set in annual frame contracts with allocation in tight markets; margin is a function of (i) end-market mix (auto + AI server >> consumer), (ii) capacity utilization, and (iii) yen FX. Recurring is mechanical: products are designed in once and re-ordered for the life of the platform (5–10 years), but volumes are cyclical.

First principles — the MLCC technology

Every electronic circuit needs decoupling capacitors that smooth voltage spikes, store charge for instant delivery, and filter noise. A modern GB200 GPU draws 1,000+ amps in nanosecond transients — the faster the load swing, the closer and smaller the capacitor must be. An MLCC is a sandwich of nickel electrodes and ~0.5µm BaTiO3 ceramic dielectric layers (500–1,000 of them), tin-plated copper terminations. Physics: C = εA/d (BaTiO3 ε ~3,000). To pack more capacitance into a 0402 body (0.4×0.2mm, ~1/100th of a grain of rice), make the dielectric thinner and stack more layers — modern high-cap MLCCs stack 1,000+ layers at 0.5µm pitch.

The ceramic powder is the moat. Barium titanate particles must be sub-100nm, spherical, monodisperse — Murata makes its own via hydrothermal synthesis with grain-size control no Chinese commodity player matches (~50% of dielectric cost in-housed). Co-firing the stack at 1,300°C in reducing atmosphere without delamination or short is the crown jewel; yields on 008004-size high-cap parts can be <50% in trial runs. The 8-step process (powder synthesis → tape casting → electrode printing → stacking/lamination → dicing → co-firing → termination → test & tape-and-reel) has hardest steps at electrode printing (sub-µm precision), co-firing (organic residue → carbon contamination → leakage), and tape casting (any defect propagates through all 1,000 layers). Every shrink (0402 → 0201 → 008004 → 006003) requires reformulated powder, retooled tape-casting, retuned co-firing — a two-year retool cycle at ¥150–200B capex per generation. Chinese makers do 0603/0402, struggle at 0201, don't ship 008004 — that is the floor under Murata's pricing power. Metrics that matter: capacitance density (industry leader at ≥10µF X7R 0402), smallest body (008004 = 0.25×0.125mm in mass production; 006003 in R&D), AEC-Q200 across 6.3V–630V, temperature class X7R (-55 to +125°C) / X8R (-55 to +150°C) for auto.

Segments

• A — MLCCs (~45% of revenue, the franchise, >60% of op profit). Sizes 008004 to 3225; capacitance 1pF to 100µF; voltage 4V to 630V. Pricing sub-cent at the commodity end (X7R 0402 1µF ~$0.001) to $0.50+ for ultra-high-cap auto-grade / 008004. AI-server baseboard MLCC content ~$30–60 today, headed to $100+. Ships ~80 billion MLCCs/month at peak. Per-platform attach: ~700 in a flagship smartphone, 10,000+ in an EV, 15,000–25,000 in an Nvidia GB200 baseboard — server attach is the secular acceleration story. Estimated GM 45–55% in high-cap/008004/auto tiers, 25–35% commodity.
• B — Communication Modules (~22%, the swing factor). RF front-end (SAW filters, BAW via licensing, FEM), Wi-Fi/BT, NFC, GNSS, mmWave; module ASPs $0.50–$5. This is where the SAW-filter problem lives: Q3 FY26 booked a ¥48.9B goodwill impairment as silicon-based BAW filters (Qorvo, Skyworks, Broadcom) eat the premium-band socket. Heaviest Apple exposure.
• C — Batteries (~10%). Li-ion coin/cylindrical cells (ex-Sony, 2017). Marginal profitability — strategic option value, not a profit driver.
• D — Inductors, EMI filters, Sensors, Timing (~23%). Diversified; inductors/EMI ride the same AI-server/auto trend; sensors include MEMS. No sub-segment >10% of revenue.
• E (new) — AI Server Power Modules (~0% today, ¥50B target by FY27). 48V Vertical Power Delivery (VPD) modules co-developed with at least one US hyperscaler (unnamed). Mass production starts FY26 H2. Estimated GM >40%; the only meaningful new product line in five years — could 3–5× the ¥50B target if a second hyperscaler adopts.

Customers & concentration

Murata does not disclose customer names; shares are estimates. Apple ~20% (18–23% range) — iPhone/iPad/Watch MLCCs + RF modules + BAW via direct + Foxconn; the only single name almost certainly >10% (historically disclosed as "a single customer exceeding 10% of net sales"). Samsung Electronics ~7–8% (also a competitor via Samsung Electro-Mechanics). Hon Hai/Foxconn high-single-digit EMS pass-through. Nvidia/hyperscalers mid-single-digit and rising fastest (indirect via Foxconn/Quanta/Wistron/Inventec, so the demand signal lags hyperscaler capex 1–2 quarters). Tesla/Toyota/VW/BYD collectively ~25–30% via Tier 1s (Denso, Bosch, Continental). Top 5 ~40–45% of revenue. Apple concentration is the dominant single-customer risk.

Moat & value-chain position

Four-pillar moat: (1) Process IP — ~50-year head start in barium titanate chemistry, co-firing yield, miniaturisation; sub-100nm grain control is not commodity-replicable. (2) Scale + capex barrier — ¥150–200B and 2 years to add high-end capacity; top 5 control ~75% share. (3) Qualification lock-in — auto/aerospace slots need 2–5 years of qualification; once designed in, switching is rare. (4) Vertical integration — in-house ceramic powder, dielectric film, electrode paste. Murata sits at the highest-margin layer because it is the upstream bottleneck — vertical integration removes most supplier exposure. The one adjacent supplier name is Sumitomo Metal Mining (5713.T) for nickel/palladium electrode paste, but MLCC is a single-digit revenue contribution there — no clean upstream long hides in the supply chain. Business-quality 3-test: (a) 5-year lock-up = yes, still #1 in 2031; (b) unique engine = ceramic process IP × Japanese manufacturing discipline × ~50-year head start × fortress balance sheet; (c) blank-check disruptor = could match commodity tier in 18 months with $20B but not 008004 in <5 years. Quality verdict: high-quality / durable oligopoly compounder; the challenge is valuation, not quality.

Financials

All figures ¥B unless noted; FY ends March 31.

Core four

1. Organic revenue growth. FY22 +25%, FY23 -7%, FY24 -3%, FY25 +6%, FY26E +5%, FY27 guide ~+6%. Underlying organic growth is modest (~3% 5yr CAGR); the leverage is in margins, not top line. TTM +12% (yfinance).
2. Margins (the story). Gross 42% → 40% → 39% → 41% → 42% (TTM). Op 23% → 18% → 13% → 16% → ~18% (FY26 ex-impairment) → 22%+ FY27 guide.
3. Capital intensity. Capex 16% of revenue (FY22) → 18% (FY23) → 15% (FY24) → 12% (FY25); FCF/revenue 4–7% — modest.
4. Capital deployment. Buybacks modest, dividends ~¥100B/yr, no major M&A; net cash grew to ¥598B (defensive but under-deployed).

Income statement & margins (annual)

Metric	FY22	FY23	FY24	FY25	LTM/TTM	FY26E
Revenue (¥B)	1,812.5	1,686.8	1,640.2	1,743.4	~1,762	~1,830
YoY %	+25.4%	-6.9%	-2.8%	+6.3%	+12% (TTM yf)	+5.0%
Gross profit (¥B)	768.2	673.5	637.0	717.7	~746	n/a
Gross margin %	42.4%	39.9%	38.8%	41.2%	42.3%	~42%
Op income reported (¥B)	424.1	298.2	215.4	279.7	~286	~270 (after impairment)
Op margin %	23.4%	17.7%	13.1%	16.0%	16.2%	~15% (reported)
Net income (¥B)	314.1	244.0	180.8	233.8	~228	~245
Net margin %	17.3%	14.5%	11.0%	13.4%	12.8%	13.4%
EPS (¥)	—	128.6	95.7	125.1	127.7	~135

FY24 was the cycle trough — gross margin collapsed 360bps on inventory correction. FY26 guide includes the ¥48.9B SAW filter goodwill impairment that masks ~¥87.7B underlying Q3 operating profit. Management guides 34.8% op income growth for FY2027 on AI-server ramp; medium-term plan (FY24–FY27) targets ¥2.5T sales and >¥500B op income by FY30, with FY27 op income guide ~¥350B (midpoint).

Second-derivative check (YoY revenue acceleration, last 5 quarters)

	Q-4 (Sep-24)	Q-3 (Dec-24)	Q-2 (Mar-25)	Q-1 (Jun-25)	Q (Sep-25)
Revenue (¥B)	461.8	448.0	411.9	416.2	486.6
Revenue YoY %	+8.3%	+3.3%	-0.9%	-10.3%	+5.4%
Op income (¥B)	91.8	76.0	45.5	61.6	103.5
Op margin %	19.9%	17.0%	11.1%	14.8%	21.3%
2nd derivative (Δ growth rate)	—	-5.0pp	-4.2pp	-9.4pp	+15.7pp

Jun-25 was the local low (-10% YoY); Sep-25 is the inflection — revenue +5.4% YoY, op margin 21.3% (strongest in two years). The Apr-2026 print beat (EPS ¥42.17 vs ¥33.69 consensus, +25%) confirmed the trajectory; the Feb-2026 print missed -62% purely on the ¥48.9B SAW impairment (ex-impairment, in line).

Incremental margin (operating leverage)

YoY pair	ΔRevenue (¥B)	ΔGross Profit (¥B)	Incremental GM	ΔEBIT (¥B)	Incremental EBIT %
Sep-25 vs Sep-24	+25	+9	37%	+12	47%
Jun-25 vs Jun-24	-48	-28	58%	-34	71% (on negative growth)
Mar-25 vs Mar-24	-4	-7	175% (noise)	-8	n/m
Dec-24 vs Dec-23	+15	+9	60%	+9	60%

When Murata grows, incremental gross margin runs 37–60% and incremental EBIT 45–60% — high operating leverage; when revenue shrinks the de-leverage is ~70% drop-through. This is why margin swings ~1,000bp peak-to-trough on ±10% revenue. Operating leverage is the single biggest variable in the FY27 outcome — +¥80–100B revenue at 40–50% incremental EBIT = +¥35–50B op income, consistent with the +35% FY27 guide.

Cash flow & balance sheet

Metric	FY22	FY23	FY24	FY25	TTM
OCF (¥B)	421 (~370)	278 (~340)	490 (~370)	452 (~425)	425
Capex (¥B)	290	310	250	210	186
FCF (¥B)	131 (~80)	-32 (~30)	240 (~120)	242 (~115 yf)	239 (yf says 115; varies by definition)
FCF margin %	7.2% (4.4%)	-1.9% (1.8%)	14.6% (7.3%)	13.9% (6.6%)	6.5–13.6%
Net cash	+740	+630	+730	+600	net cash ¥598B
ROIC	~17%	~12%	~8%	~10%	~11%
ROE	14.7%	10.7%	7.4%	9.1%	8.8%

(OCF/FCF figures differ between deep-dive and profile fragments — both kept above; the discrepancy is definitional, OCF ~¥425B vs capex ~¥186–210B TTM.) Balance sheet is fortress: ¥654B cash, ¥56B debt, ¥598B net cash — effectively zero financial risk. The ¥598B net cash on a ¥12.3T cap = ~5% of market cap, under-deployed. ROIC vs WACC: estimated WACC 6–7% (low JPY cost of debt, equity beta 1.1, risk-free ~1.5%, ERP 5–6%); ROIC 17% (FY22) → 8% (FY24 trough) → 11% (TTM) → ~13–14% (FY27 implied). Spread +4 to +7pp through cycle — creating value, compressed at trough. R&D ¥149B / 8.6% of sales (FY25) — heaviest among MLCC peers.

Q3 FY2026 detail (Feb-2026)

Sales ¥467.5B (+4.3% YoY). Operating profit ¥37.9B (halved) on the ¥48.9B SAW goodwill impairment; ex-impairment OP ~¥87.7B (in line). Full-year sales guide raised ¥60B to ¥1.8T; OP guide cut ¥10B to ¥270B.

Industry landscape

MLCC market ~$15B (2025) → ~$22B (2030) at ~8% blended CAGR (MarketsandMarkets, Mordor Intelligence); high-end (auto + AI server) growing >15%. AI-server MLCC sub-segment: Murata's own forecast 3.3× FY2025→FY2030 (~27% CAGR), with the Dec-2025 IR Day raising per-baseboard count to 15–25k (from 10–20k). Murata's total addressable (MLCC + inductors + EMI + RF + batteries) ~$60B; serviceable ~$40–45B. Market share: overall MLCC ~34% (#1); automotive MLCC 40–50% (nearest TDK ~20%); consumer MLCC ~34% (Samsung Electro-Mechanics 24%, TDK 12%); high-cap (≥10µF X7R/X5R) >50% — the pricing-power tier. End-use mix (FY2025): communications ~30%, automotive ~30% (was ~20% in FY2020), industrial ~10–15%, computing/data-centre/AI ~10% and growing fastest (>30% CAGR), consumer/wellness/energy the balance.

Structure is oligopolistic — top 5 = ~69–79% (~75%) share, consolidated at the high end, fragmenting at the commodity end where Chinese makers (Fenghua 000636.SZ, Sanhuan 300408.SZ, Eyang 600732.SH) have risen from ~6% (2019) to ~10% (2025), +4pp, but remain stuck at commodity X7R 0402/0603. Cycle position: mid-cycle on the recovery side — FY24 trough (op margin 13.1%), FY25 recovery (16.0%), FY26 ex-impairment ~18–19%, FY27 guide ~22%+; prior peak FY22 ~23.4%. Up/down cycles run 18–30 months. Each cycle troughs higher because AI/EV/miniaturisation secular drivers layer on top. Why now: AI-server MLCC demand inflecting as Murata exits a 2-year capex cycle with incremental high-end capacity (Moriyama online 2026, Fukui completed Apr-2026), tight-segment supply returning pricing power (Feb-2026 capacity doubling + price-increase consideration), and SAW writedown clearing bad news — the window is now through end-FY27.

See sector page: passives-mlcc. Material-moat + AI-power-stack synthesis (Murata MLCC vs Musashi HSC, corrected upstream chain): 6981-vs-7220-pdn-stack.

Management

Overall management grade: B+ — high-quality, conservative, low-drama Japanese keiretsu-style stewardship; strong on capital preservation and governance, weak on aggressive value-creation. Exactly what you want owning a moaty oligopoly; less what you want for a pivot or turnaround.

Leadership (all internal, lifer-track, Japanese)

• Norio Nakajima — President & CEO since Jun 2020 (6 years; joined 1986, 39 years). Engineering background, ran the Communication & Sensor BU through the 5G ramp — the RF franchise he built is the one now being eroded by Qualcomm RFFE integration and BAW substitution (the ¥48.9B SAW impairment is the largest write-down on his watch; uncomfortable optics, not necessarily his fault). Re-anchored capex toward AI server + auto; steered the FY23-24 downcycle without margin emergency; took the SAW impairment surgically rather than spreading it.
• Masanori Minamide — Executive Deputy President & CFO since 2023. Corporate admin / legal-IP / management-DX background — not a traditional finance career, closer to a CAO/GC pivot. Modernised IR (IR Day expansion, granular segment disclosure) but the non-finance background showed in the Q3 FY26 impairment-plus-guide-cut announcement that confused analysts on the ¥87.7B underlying run-rate. Total comp ~¥130M (~$830K).
• Hiroshi Iwatsubo — Executive Deputy President & CTO, 20+ year tenure. Drove the 008004 / ultra-miniature roadmap and 8.6%-of-sales R&D; also oversees Medical Products. Total comp ~¥140M (~$890K).
• Nagato Omori — SEVP, Ceramic Capacitor BU head — effectively the franchise CEO (MLCC = ~45% revenue, >60% op profit) but not on the board (minor governance flag); likely next-CEO pipeline.
• Ken Tonegawa — EVP, Energy BU (ex-Sony batteries; mandate to make it profitable, slow progress).
• Eiichi Morimoto — EVP, Global Sales & Marketing (Apple/Samsung/hyperscaler relationships).
• Hiroshi Izumitani — EVP & Board Director, Communication & Sensor BU head — runs the impaired SAW business and sits on the board; awkward juxtaposition, no disclosed SAW restructuring plan within 12 months.

Skin in the game (Yellow)

Aggregate insider holdings 33.4M shares = 1.83% of outstanding (~¥225B / ~$1.4B at spot). No individual director >0.1%. Insider buys last 6 months: 0; sales: 0; net: 0. Founding Murata family (Akira Murata, founder, d. 2006) no longer holds a material direct stake — dilution since the 1963 TSE listing and generational succession. Low by US standards, normal for Japan; alignment is via career/reputational/lifer-track, not equity — fine for an oligopoly franchise, insufficient for a turnaround thesis. Japanese-disclosure caveat: no DEF 14A; no 10b5-1 / Section 16 equivalent at US granularity; "no flag" means no flag in disclosed Japanese materials (Yuho, Corporate Governance Report Jul-2025, Notice of Convocation, ≥5% Large Shareholding Reports). Absence of disclosure is not evidence of absence.

Holdings concentration & shell scan (Green)

No disclosed cross-holdings in Murata customers/suppliers/competitors; no insider board seats creating conflict; executives' wealth is pensions/savings/modest equity, Japanese norm. Structure is materially above-average clean: all operating subsidiaries 100% wholly-owned (Tohoku Murata, Wuxi Murata, PFTC Philippines, PT Murata Indonesia, Murata Electronics North America / Singapore, Murata Vietnam, Murata Italy, Murata Energy, ~70 consolidated entities) — no Cayman/BVI shells, no minority loops, no insider-affiliated JVs (Resonac MOU 2025 is a dielectric-supply collaboration, not equity), no related-party transactions beyond standard pension-trust (Master Trust / Custody Bank). No SEC/JFSA/TSE enforcement, no fiduciary-duty lawsuits, no fraudulent-conveyance claims, no bankruptcies. Disclosure asymmetry acknowledged (trust-account and family-member holdings not individually reportable).

Capital allocation (Green, grade B+)

Last material M&A was the 2017 Sony battery acquisition (modest price, no goodwill write-down, hasn't compounded) — Murata is fundamentally an organic operator, a poor acquirer by default. Buybacks: share count ~1.896B (FY23) → 1.832B (TTM) = ~64M shares / ~3.4% retired, executed in the FY24-25 trough at ¥2,000–4,000 vs spot ¥6,733 — disciplined timing (textbook AZO pattern). Capex throttled in the trough (¥250–290B → ¥210B FY25) not accelerated like TDK / Samsung E-M; incremental revenue per ¥1 capex ~¥0.7 (fab-like). No equity issuance in a decade; no convertibles/warrants/shelf. Dividend consistent (FY25 payout ~¥101B, ~43% ratio, yield 1.14%, no cut in 20+ years). The clearest ding: ¥598B net cash under-deployed at ~28× forward P/E — they neither aggressively buy back nor pay a special; an honest "we think it's fairly priced" signal, not a promoter.

Compensation & alignment (Yellow, grade B/B+)

CEO Nakajima total comp ~¥251M (~$1.6M) — a 10–15× discount to a US S&P 100 industrial CEO at a $78B cap. Average board director ~¥80M (~$510K). Three components: fixed base (~60–70%), annual cash bonus (tied to consolidated sales, op income, ROIC, ESG/succession/DX KPIs), and stock LTI (Restricted Stock Compensation Plan + Performance Share Plan on 3-year medium-term-plan KPIs). ROIC inclusion is a positive; SBC <0.1% of revenue; standard Japanese severance (1–2× base, no golden parachutes); no related-party perks/leases/family-payroll. Per-tranche hurdle granularity not disclosed at US level — inferred aligned (FY27 OP ≥¥350B, FY30 OP >¥500B targets require the AI thesis to play out for LTI to vest), partial verification only.

Credibility (Yellow, grade B/Mixed)

12-quarter beat/miss tape is ERRATIC — 5 beats / 6 misses over 11 reported quarters, ~36pp surprise std dev. Beats: Oct-23 +39.4%, Jul-24 +13.2%, Jul-25 +3.1%, Oct-25 +37.2%, Apr-26 +25.2%. Misses: Feb-24 -23.5%, Apr-24 -86.0% (trough + impairment), Oct-24 -15.1%, Feb-25 -4.3%, Apr-25 -17.0%, Feb-26 -62.0% (SAW impairment). Average miss (ex-impairment) -16%, average beat +24%. Much of the erratic-ness is structural operating leverage (±5% revenue → ±20% EPS), not dishonesty — but it means guidance has lower information content than TSMC's. The one real weasel: "continuing to evaluate strategic options" on the SAW business for 2–3 quarters before quantifying the impairment — management knew, the language was cover (one honesty deduction, not serial). Strategy guidance and capital-structure commitments are strong (IR Day FY2030 targets tracking, no dilution surprises); overall follow-through ~70%.

Board & governance (Green, grade A-)

Board ~10. Independent chairman Takashi Nishijima (former Yokogawa Electric chairman) — materially better than the Japanese norm where the chair is the predecessor CEO. Inside directors: Nakajima (CEO), Iwatsubo (CTO), Minamide (CFO), Izumitani (BU head). Majority-independent Audit & Supervisory Committee (Code 4.7), refreshed Apr-2026 with two new outside auditors; Personnel & Compensation Advisory Committee chaired by an outside director. Single-class shares, no poison pill, no staggered board (annual elections). AGM votes consistently >90% for management. No activist 5%+ filer, no sale/take-private signalling — consistent with the long-cycle organic-operator profile. No material red flags identified. Among MLCC peers, Murata's governance is the highest — above TDK's messier structure, Taiyo Yuden's smaller-cap governance, and comfortably above Samsung Electro-Mechanics' chaebol related-party concerns.

Catalysts & risks

Bull / catalysts

• AI server MLCC content — 15–25k MLCCs per GB200/GB300 baseboard; 3.3× FY25→FY30 (~27% CAGR); 5–10 year duration.
• EV content — 3–4× MLCC per EV vs ICE (~10,000 vs ~3,000); auto already 30% of revenue, mid-teens % CAGR, structurally permanent.
• 48V VPD power modules — new SKU, ¥50B target by FY27 (~3% of revenue at target; could be 10%+ if multi-hyperscaler).
• Ultra-miniature (008004, 006003) — design wins in iPhone flagships + AI server; margin-accretive.
• Pricing power return — Feb-2026 Digitimes: capacity doubled + price increases considered (first explicit pricing signal since 2022); 100–300bp margin upside.
• Near-term (0–12mo): FY26 full-year results July 31, 2026 (sets FY27 base; underlying ~¥320B run-rate vs ¥270B guide); GB300 launch H2 2026 reveals Murata content per board; first explicit MLCC price-increase announcement; auto restocking visible in TDK/Taiyo Yuden monthly data.
• Medium-term (1–3yr): Trainium-3 / MI400 / GB400 ramps broaden multi-hyperscaler AI MLCC adoption; second 48V VPD hyperscaler (would 3–5× the ¥50B target); India localisation ramp (Apple India); SAW filter restructuring conclusion.
• Key contracts: 48V VPD ¥50B cumulative through FY27 (US hyperscaler, mass-production FY26 H2); AI server MLCC allocation via Foxconn/Quanta/Wistron (capacity doubled Feb-2026); Moriyama Innovation Center (¥46B, online 2026); Fukui MLCC building (¥47B incl. equipment, completed Apr-2026); India leased plant (full-scale FY26).
• Add triggers: pullback to ¥5,500–6,000 on a cycle scare without thesis break; first explicit MLCC price increase; second hyperscaler signs for 48V VPD.

Bear / risks

• AI capex peak / digestion (Medium) — if hyperscaler capex pauses in 2027, server MLCC demand reverts; auto offset mitigates but can never fully close cyclicality. Trigger: two consecutive quarters of negative AI-server MLCC YoY in 2027.
• SAW filter erosion vs BAW (High, already realised) — ¥48.9B impairment taken Q3 FY26; premium-band socket bleeding to Qorvo/Skyworks/Broadcom; restructuring not yet disclosed. SemiAnalysis flagged this in Feb-2022 and it has now materialised.
• Smartphone weakness (High) — ~30% revenue exposure, Apple-cycle dependent; mix-shifting away but never fully closes.
• JPY appreciation (Medium) — every ¥1 vs USD ≈ ¥2.7B op profit; JPY at 140 (vs spot ~157) ≈ ¥45B headwind, ~13% of op income; not closable (~90% export business).
• Chinese commodity competition (Medium-High at commodity, Low at high-end) — Fenghua/Sanhuan/Eyang +4pp share since 2019; bounded to commodity tier, not closable structurally.
• Capex mistiming (Medium) — fortress balance sheet absorbs error; closes once FY27 utilisation prints.
• Apple concentration ~20% (Medium) — single-customer event risk; partially closable, will remain top-3.
• 3D-stacked / silicon-capacitor alternatives (research-stage TSMC trench caps, Murata silicon-cap line) — possible 5-year story.
• Power-module integration — VRMs increasingly integrate capacitors on-package, potentially reducing discrete MLCC count (offset by overall content increase).

No dilution risk (share count declining; OCF ¥425B vs capex ¥186B TTM self-funds growth/dividends/buybacks; no convertibles/warrants/shelf). No material key-person risk (Nakajima a lifer; internal-deep succession — Iwatsubo, Minamide, Omori; Japanese planned-succession culture). Behavioral traps flagged: FOMO (stock +3.3× at ATH) and recency bias (two beats bracketing one big miss) — both mitigable via scale-in sizing.

Bear case scenario: AI capex digestion hits in 2027, FY27 op income comes in ~¥280B (flat to FY26) not ¥350B, forward P/E re-rates to 22× → ~¥4,500–5,200 share price (-23 to -33%). Trigger: two consecutive quarters of negative AI-server MLCC YoY. Downside target coincidentally matches the current sell-side mean PT.

Valuation / DCF

Murata trades at the highest multiple in the MLCC peer set on every metric — a 30–60% premium. Forward P/E 28.4× vs peers ~22× and 5-year average ~22×; EV/EBITDA ~21× vs peers ~13–14×; EV/Revenue 6.1× vs ~2–3×; P/B 4.5× vs 1.5–2.5×; FCF yield 0.9% vs 2–5%; dividend yield 1.14% vs 2–3%.

Peer comps

Company	Ticker	Fwd P/E	EV/EBITDA	EV/Rev	FCF Yield
Murata	6981.T	28.4×	~21×	~6.1×	~0.9%
TDK	6762/6762	6762.T	~22×	~14×	~3×
Taiyo Yuden	6976/6976	6976.T	~24×	~13×	~2×
2327/2327	Yageo	2327/2327	2327.TW	~20×	~14×
Samsung Electro-Mechanics	009150/009150	009150.KS	~16×	~7×	~1×

Competitive-landscape detail: Murata ¥1.74T revenue / ~34% MLCC share; Samsung Electro-Mechanics ~₩9T / ~24%; TDK ¥2.1T / ~13%; Taiyo Yuden ¥320B / ~13% (pure-play); Yageo NT$310B / ~13% (post-Kemet consolidator); Chinese commodity tier combined ~CNY15B / ~10%.

Premium justified by share leadership (~34% vs #2 ~24%), high-end mix (auto + AI + 008004), fortress balance sheet, highest R&D intensity (8.6% of sales vs 5–6% peers), cleanest governance among Asian peers — but the premium is full. EV/EBITDA 21× vs TDK 14× is ~50%; requires the AI-server pricing thesis to play out. If FY27 op income hits ¥350B (guide midpoint), forward EV/EBITDA collapses to ~17× — still premium but defensible.

Historical range: forward P/E 5-year average ~22×, cycle peak (FY22 boom) 32×, cycle trough (FY24) 18×, current 28× = upper end.

DCF back-solve: at ¥6,733 the market is pricing ~6% revenue CAGR over 10 years with op margin expanding from 16% to ~21% steady-state, then 3% terminal — maps to the FY27 guide. The market is not pricing margin upside to FY22 peak (23%+) nor the AI-server pricing-power scenario — a modest cushion. Would I buy 10–15% higher (¥7,400–7,750)? No — margin of safety insufficient, and ¥7,400 is above the high sell-side PT; the buy window narrows quickly above current levels.

The disconnect: sell-side mean PT ¥5,203 = 12.6× FY26 EPS, reflects targets set before the Dec-2025 IR Day re-rating. Either targets are revised up over the next months or the stock corrects 20–25%. Watch broker-note flow into July-2026 results.

Price targets / scenarios: my 12-month base ¥7,500–8,000 (+12 to +20%); sell-side mean ¥5,203 (-23%), high ¥7,300 (+8%), low ¥3,900 (-42%) — wide ¥3,400 range = high disagreement. Bear/downside ¥4,500–5,200 (-23 to -33%), which coincidentally equals the sell-side mean.

Decision log

2026-05-20 — Pre-buy checklist verdict: BUY (scale-in, half-size initial). Conviction Medium. Murata is the highest-quality way to own the MLCC oligopoly into a real AI-server-led upcycle, with a fortress balance sheet and B+ governance; valuation full but defensible; second-derivative on earnings turned sharply positive (Sep-25 op margin 21.3%, Apr-26 EPS +25% beat). Catch: at ATH after a 3.3× run, sell-side PTs 23% below spot, July 31 earnings a high-risk binary.

FundamentEdge hard rules: 3 of 5 pass. PASS — second derivative (positive, two-quarter inflection); valuation-is-not-a-thesis (buying on content growth + pricing, not low P/E); quality-can-be-a-risk (margins defensible, 3-test holds). FAIL — revenue growth primacy (~3% organic CAGR, not 8–12%; margin/mix thesis not top-line); estimate revision direction (price re-rated but sell-side mean PT fell to ¥5,203, recommendation distribution slightly degraded 5→4 strong buys). Both real, neither disqualifying — proceeding with downgraded conviction.

Why buying: growth compounder + AI-cycle exposure (primary); portfolio construction as the MLCC quality anchor (secondary). Not undervalued, not turnaround, not income.

Position plan: 2–3% of portfolio as the MLCC anchor. Scale in over 30–60 days — Tranche 1 (50%) now at ¥6,733 or intraday weakness to ¥6,500; Tranche 2 (25%) on pullback to ¥6,000–6,200 OR confirmed breakout above ¥7,000; Tranche 3 (25%) reserved for post-July-2026 earnings reaction. Holding period 18–36 months. Max loss tolerance -25% from average cost.

Exit / re-evaluate triggers: target ¥7,500–8,000 (12-month); two consecutive quarters of negative YoY revenue; FY27 op income guide cut below ¥320B; JPY breaks 140 vs USD sustained; Chinese commodity makers ship a credible 008004 part.

Management DD verdict (2026-05-20): grade B+. I would trust this management with capital — competent stewards of a high-quality oligopoly (trough buybacks, clean impairment-taking, no dilution, no bad M&A, above-Japanese-norm independent governance), but not a team for a pivot/turnaround/value-unlocking restructuring. The ¥598B under-deployed net cash is the clearest evidence; the SAW "evaluating strategic options" → impairment is the one honesty deduction — read forward guidance on weak business lines with skepticism.

Role in the MLCC peer swarm: the quality anchor / risk-adjusted compromise. For more cyclical torque or a value entry, Taiyo Yuden or Samsung Electro-Mechanics are likely better; for highest-quality lowest-drama exposure, Murata is it.

SemiAnalysis cross-check. Only direct SA mention is Feb-2022 ("Semiconductor Roundup 2/10/2022"), where Dylan Patel called Murata a short on RFFE share loss + MLCC weakness — directionally right through the FY23-24 trough and validated on SAW by the ¥48.9B impairment. No SA coverage of Murata or MLCCs in the 2024–2026 mirror. My stance diverges (cautious-bull vs SA-bear): the RFFE/SAW caution stands and is carried as a High-likelihood risk, but the AI-server demand reset SA hadn't priced changes the calculus.

Sources

Consolidated 2026-05-30 from four research fragments (all 6981.T / Murata Manufacturing — no wrong-entity content):

• 6981-t-deep-dive.md (Register D full write-up, 2026-05-20)
• 6981-t-mgmt-dd.md (management due diligence, 2026-05-20)
• 6981-t-checklist.md (pre-buy checklist, 2026-05-20)
• 6981-t-profile.md (company profile, 2026-05-20)

External sources cited across fragments:

• yfinance API (live price, financials, holders, recommendations, insider data) — accessed 2026-05-20
• Murata IR — IR Day Dec-2025 deck, FY25 Yuho, Q3 FY26 kessan tanshin
• Murata Corporate Governance Report Jul-2025
• Digitimes — Murata MLCC capacity doubling + price increase consideration (Feb-2026)
• Digitimes — Murata 2026 AI server production
• TrendForce — Murata 48V VPD AI server power modules (Dec-2025)
• SemiAnalysis — Semiconductor Roundup 2/10/2022 (historical short call, partially validated)
• Evertiq — Fukui MLCC building completion Apr-2026
• Passive-Components.eu — China MLCC share at 10%
• MarketsandMarkets — MLCC market size & forecast
• Mordor Intelligence — Automotive MLCC market
• Murata Q3 FY26 earnings call recap (BigGo)
• Morningstar Murata coverage

---

Consolidation queue (merged 2026-05-30)

The four fragments below were folded into this canonical page on 2026-05-30 and remain live pending Pink's archive confirm.

• [ ] 6981-t-deep-dive.md
• [ ] 6981-t-mgmt-dd.md
• [ ] 6981-t-checklist.md
• [ ] 6981-t-profile.md

Source updates (auto-maintained)

Drop/2. Frontend (Jun 1, 26) - GS Memory

Goldman Sachs expects DRAM/NAND/HBM supply/demand to tighten through 2028, driven by AI server build-out that now represents ~50% of DRAM demand — up from 16% in 2017 — with hyperscaler capex commitments underpinning visibility; Murata is not directly covered but appears as a supply-chain mention.

Relevant to your thesis: GS's confirmation that AI server demand is structural and multi-year (HBM TAM $56B→$168B by 2028) corroborates the Murata bull case that AI-server MLCC content (15,000–25,000 units/baseboard) is a durable volume driver, not a pull-forward.

Source: dropfile://2. Frontend/Memory/GS Memory.pdf

Intake (May 23, 26) - mops-diligence-2026-05-23

The May 2026 Taiwan MLCC peer diligence confirms Murata's structural advantage: English-language disclosure for all four Taiwan names (Walsin, PDC, Holy Stone, Prosperity Dielectrics) is materially thinner than US norms, and governance discounts of 10–20% are recommended for the PSA group names.

Relevant to your thesis: Reinforces Murata as the highest-quality way to own the MLCC oligopoly — peer governance opacity and undisclosed RPT pricing terms widen the quality gap versus the Taiwan alternatives.

Source: intakefile://mops-diligence-2026-05-23.md

Intake (May 21, 26) - 2327.TW-filings

Yageo's Q1 2026 gross margin of 38.1% (+240bps YoY) and AI now at ~15% of group revenue signals the MLCC upcycle is broadening across the oligopoly, not concentrated in Murata alone.

Relevant to your thesis: Peer margin expansion confirms the AI-server demand pull that underpins Murata's FY27 op margin recovery guide, but also signals that Yageo is closing the quality gap faster than the oligopoly-pricing-discipline thesis assumes.

Source: intakefile://2327.TW-filings.md

Drop/z Misc (Aug 24, 25) - Virgin_Galactic_Expectations_vs_Post_Merger_Reality_17560128...

This article is a student finance exercise analyzing Virgin Galactic's 2019 SPAC valuation using DCF and comparable company methods; it contains no substantive mention of Murata (6981) — the ticker match appears to be a false positive.

Relevant to your thesis: Tangential — flagged for review.

Source: dropfile://z Misc/Virgin_Galactic_Expectations_vs_Post_Merger_Reality_1756012888.pdf

Drop/Bottleneck (May 24, 26) - 东莞证券_MLCC行业深度报告：供需矛盾加剧，高阶MLCC价格有望上扬_260325 (1)

Murata executives confirmed AI server MLCC order inquiries at 2× current capacity, with a March 2026 decision to raise prices 15–35% on AI server and high-end auto-grade products starting April; Murata and Samsung Electro-Mechanics are both at full capacity with delivery times lengthening.

Relevant to your thesis: Directly confirms the Feb-2026 Digitimes price-increase signal already cited in the thesis, and adds the 2× oversubscription data point that validates the supply-demand tightening underpinning the margin-recovery call.

Source: dropfile://Bottleneck/MLCC/东莞证券_MLCC行业深度报告：供需矛盾加剧，高阶MLCC价格有望上扬_260325 (1).pdf

Drop/Bottleneck (May 21, 26) - global_passives_basket_comparison

The basket comparison ranks Murata #1 of seven at 21/30 (Buy), with Moat the only top score (5/5); Crowding and Valuation both score 3/5, with fwd P/E cited at ~40× on FY27 EPS guide (~¥155) and consensus sitting ~18% below the current price.

Relevant to your thesis: The "honest catch" in the wiki (sell-side below spot, multiple at top of historical range) is independently confirmed — and the ~40× forward P/E figure implies the article uses a purer FY27 window than the wiki's blended 28.4×, suggesting the valuation is less compressed than it looks.

Source: dropfile://Bottleneck/MLCC/global_passives_basket_comparison.pdf

Drop/Bottleneck (May 21, 26) - Murata vs Yageo (5.21.2026) (1)

FY2026 actuals confirm revenue +5.0% / op profit +0.8% with ¥43.8B SAW impairment taken; FY2027 guidance reaffirmed at +7.1% revenue / +34.8% op profit; backlog surged +89.5% to ¥269.2B with server/compute +28.4%; ¥150B buyback announced; cash ¥653B.

Relevant to your thesis: Confirms the margin-recovery/mix-shift thesis — the +34.8% op profit guide on modest revenue growth is explicitly explained by strategic MLCC pricing (+29%) offsetting commodity decline, and the backlog surge validates the supply-constraint moat.

Source: dropfile://Bottleneck/MLCC/Murata vs Yageo (5.21.2026) (1).pdf

Drop/Bottleneck (May 24, 26) - J.P. Morgan-MLCC Industry：Growing likelihood of tight supply...

J.P. Morgan upgraded Murata to Overweight (April 3, 2026), citing MLCC industry capacity utilization at 87–88% approaching the 90% tight-supply threshold, GB300 MLCC content up 50–60% vs. GB200 (with another 50–60% jump under Vera Rubin), and an expectation that price declines slow in 2H FY2026 and stabilize in FY2027 — with ¥800B operating profit potential if stabilization extends through FY2028.

Relevant to your thesis: Directly validates the two most load-bearing bull points — capacity tightening is now JPM-confirmed and on a quantified timeline, and the AI server content ramp (4k → 12k MLCCs per GPU over five years) materially corroborates Murata's own 3.3× FY25–FY30 AI server volume forecast.

Source: dropfile://Bottleneck/MLCC/J.P. Morgan-MLCC Industry：Growing likelihood of tight supply demand； Murata Manufacturing and Taiyo Yuden up to Overweight-260403.pdf

Drop/Bottleneck (May 21, 26) - BofA Securities-Component Signals：MLCC production： Maintain...

The article body is unreadable — the content field contains only an encoded/garbled string (4FzBZFn3tZcUdXaQaO8...), not actual text. I can't extract what BofA said without fabricating it, which violates the hard rule against guessing.

Please re-drop the article with readable body text and I'll write the block immediately.

Source: dropfile://Bottleneck/MLCC/BofA Securities-Component Signals：MLCC production： Maintain Buy on Murata and TDK on solid earnings support-260519.pdf

Drop/Bottleneck (May 24, 26) - 2026 04 Capacitor Dossier

The dossier names Murata alongside Samsung Electro-Mechanics and TDK as Japanese/Korean suppliers pursuing "cautious capacity expansion focused on high-margin automotive and AI segments," while Chinese manufacturers maintain aggressive commodity pricing — characterizing the result as a "polarized global supply landscape."

Relevant to your thesis: Corroborates the bifurcated pricing dynamic the bull case depends on — Murata disciplined on high-end capacity, Chinese players absorbing commodity margin pressure, not encroaching on the profitable tiers.

Source: dropfile://Bottleneck/MLCC/2026 04 Capacitor Dossier.pdf