MLCC Peer Swarm — Best-Positioned Into the AI/EV/Smartphone Up-Cycle
Date: 2026-05-20 | Universe: Murata 6981.T, Yageo 2327.TW, Taiyo Yuden 6976.T, TDK 6762.T, Holy Stone 3026.TW, Walsin 2492.TW, Prosperity Dielectrics 6173.TWO
1. Verdict
Rank by best-positioned right now (quality of cycle exposure × valuation cushion × balance sheet × management): (1) Murata, (2) TDK, (3) Yageo, (4) Walsin, (5) Taiyo Yuden, (6) Prosperity Dielectrics, (7) Holy Stone. The headline why: Murata owns the high end of every cyclical bucket with a fortress balance sheet and the cleanest pricing-power signal (Feb-2026 capacity doubling plus a confirmed Apr-1 price hike of 15-35% on high-end MLCC), and it is the only name where the AI-server thesis is paired with new SKU optionality (48V VPD power modules). TDK is the highest-quality governance in the swarm (ISS QualityScore 1, ROIC-and-TSR PSU hurdles) but the MLCC slice is only ~10% of revenue — owning TDK for MLCC means owning the ATL Apple-battery business by accident; cheaper than Murata on EV/EBITDA but with the wrong product mix for the thesis. Highest-conviction long: Murata, scaled in on any pullback from all-time-high ¥6,733. Best risk-adjusted buy on pullback: Walsin at TWD 200-225 zone (highest cyclical beta, cleanest entry math if it gives 15-25% back). Avoid at current prices: Holy Stone (TWD 433.5, 66x trailing P/E on a Tier-2/3 distributor-mix hybrid that has 5x'd off the low on one good quarter, single stale analyst PT at TWD 160). Pair trade: long Murata / short Holy Stone or PDC to harvest the parabolic vs quality dislocation. Upstream alternative: Sakai Chemical (4078.T) — flagged by three sub-agents as the under-priced barium-titanate powder bottleneck (~3-5% of Murata's market cap, sole-supplier dynamics on sub-100nm dielectric). Not researched in this swarm; queued as the cleanest non-consensus way to play the MLCC cycle.
2. Snapshot
| Ticker | Spot | Fwd P/E | EV/EBITDA | FCF Yield | Mean PT vs Spot | Cycle Position | Verdict | Entry Zone |
|---|---|---|---|---|---|---|---|---|
| Murata 6981.T | ¥6,735 | 28.4x | 21x | 0.9% | -23% (¥5,203) | Mid-cycle recovery; FY27 OP guide +35% | BUY (scale-in) | ¥6,000-6,500 add; full at ¥5,500 pullback |
| TDK 6762.T | ¥3,179 | 31.8x | 11.3x | 1.9% | -10% (¥2,869) | Multi-cycle peak (iPhone 17 + nearline HDD + AI passives) | WATCH | ¥2,400-2,500 starter; full at ¥2,000-2,200 |
| Yageo 2327.TW | NT$572 | 26.8x | 25.5x | 2.2% | -26% (NT$423) | Mid-cycle; mix-shift to AI tantalum + auto | WATCH | NT$480-520 starter; full at NT$420-460 |
| Walsin 2492.TW | TWD 262 | 15.1x | 36.9x | 4.0% | n/a (no consensus) | Mid-cycle; price-follower not setter | WATCH | TWD 200-225 starter; full at TWD 175 |
| Taiyo Yuden 6976.T | ¥8,231 | 43.9x | 15.4x | ~0% | -43% (¥5,739) | Mid-cycle; highest-beta MLCC pure-play | WATCH | ¥6,000-6,500 starter; full at ¥5,000-5,500 |
| PDC 6173.TWO | TWD 175 | 30x+ (no consensus) | 23.7x | 2.6% | n/a (no consensus) | Early-cycle confirmation pending June 11 print | PASS | TWD 95-110 starter; conviction at ¥55-75 |
| Holy Stone 3026.TW | TWD 433.5 | 33.2x (1 analyst) | 46.5x | 1.1% | -63% (TWD 160 stale) | Mid-cycle on one print; parabolic | PASS | TWD 280-320 watch zone only |
Spot prices are as of the sub-agent timestamps (2026-05-20 and 2026-05-21). Holy Stone, PDC, Walsin, TDK, and Taiyo Yuden moved within the 30-min sub-agent window; treat as snapshots. Yageo spot NT$572 is from the profile (deep-dive notes spot NT$572 also). Holy Stone deep-dive references TWD 433.5; checklist references TWD 428-433.5. PDC ATH was TWD 175 hit May 15; spot ~TWD 173-175 May 21.
3. Ranked Positioning 1-7
1. Murata (6981.T) — the quality anchor with a real catalyst. ~34% global MLCC share with the deepest moat in high-cap, ultra-miniature (008004), and auto-grade AEC-Q200 — exactly the tiers where the AI/EV cycle has its highest pricing leverage. Q2 FY26 op margin recovered to 21.3% (strongest in two years) and Apr-2026 EPS beat by 25%. The Feb-2026 Digitimes report of capacity doubling + confirmed Apr-1 industry-leading price hikes is the rare explicit pricing-power signal — Murata is leading the cycle, not following. Fortress balance sheet (¥598B net cash) and 48V VPD power modules for AI servers (¥50B by FY27 with a named US hyperscaler) give it the only meaningful new product line in the swarm.
2. TDK (6762.T) — governance-quality leader, MLCC by accident. ATL Apple silicon-anode batteries (55% of revenue, 89% of segment OP) plus nearline-HDD duopoly (+380% segment OP YoY) plus AI server aluminum-electrolytic plus 100V+ automotive MLCC — four cycles inflecting at once. ISS QualityScore 1 across all four pillars (rare for any Japanese mega-cap), majority-outside board since 2024, PSU hurdles include ROIC at 8.1% (currently behind at 6.7% — meaning real payout-at-risk). The catch: MLCC is only ~10% of revenue and the company has stopped being primarily an MLCC story — owning TDK for the MLCC thesis is using a sledgehammer to crack a walnut.
3. Yageo (2327.TW) — the consolidator, fairly priced for what it owns. Global #1 tantalum capacitor (>46% share via KEMET, AI-server-designed-in for NVIDIA Hopper/Blackwell rails), #1 chip resistor, #3 MLCC. Operating margin expansion 19.0% to 25.2% over 8 quarters is real; Q1 2026 +9.2% EPS beat continues a 6-quarter beat streak. Pierre Chen owns ~20% personally (cleanest founder alignment in the swarm), and KEMET (2020), Chilisin (2021), Nexensos (2023), Telemecanique (2023), Shibaura (2025) are a respectable roll-up record. The brief's "Nexensos pending" framing was wrong — it closed in March 2023 and is generating revenue; the Yageo sub-agent corrected this. Tradeoff: Yageo has rallied +207% off the low to NT$572 and trades 35% above mean analyst PT — fair but not cheap.
4. Walsin (2492.TW) — second-leg cyclical participation, awaiting pullback. Taiwan's #2 MLCC pure-play, +245% off the September 2024 low. Operating leverage is real (China utilization 70%, recovering toward 85%+) but Walsin's pricing posture in the May-2026 round is the structural tell — Murata pushed 15-35% on high-end Apr 1, Yageo 10-20%, Walsin is still negotiating case-by-case on loss-making SKUs only. Walsin is a fast-follower, not a price-setter, and forward 15x P/E implies 3.7x EPS growth in 12 months — steep without leading pricing. Family-controlled (Chiao group, 41.7% insider) with ISS QualityScore 10 (worst decile) per the sub-agent's read. Right name for cyclical torque, wrong entry price at TWD 262.
5. Taiyo Yuden (6976.T) — the highest-beta pure play, worst entry math. 71% of revenue is MLCC — the only major where MLCC is the dominant economic line, and the most levered to the cycle. 4 consecutive earnings beats (last one +25%), Q2 FY26 OP margin recovered from 1% trough to 6%+ run-rate, and Taiyo Yuden led the May-2026 6-13% commodity MLCC price hikes (the canonical "cycle confirmed" signal because pricing returns to the high-end first and the tail last). But the stock has overshot — ¥8,231 is +44% above the 16-analyst mean PT of ¥5,739, forward P/E 43.9x prices peak EPS within 18 months, and capital allocation timing in the prior cycle was textbook bad (bought back at FY3/22 peak, over-built capex into FY3/24 trough). Highest beta, but worst entry-priced of the swarm.
6. Prosperity Dielectrics (6173.TWO) — right business, wrong price, structurally one-sided catalyst. The genuinely interesting asset: #2 global ceramic dielectric powder maker by volume (3-supplier oligopoly with Sakai and Nippon Chemical) bolted onto a sub-scale finished MLCC business. PSA group ally of Walsin since 2005 (Walsin is shareholder + powder customer, not pure customer as the brief framed). 47% insider alignment, fortress balance sheet, 18-29% FCF margin through the cycle. But TWD 175 is the 52-week high after a 5x in 12 months and a double in three weeks, on FY25 numbers that grew net income only 12% YoY. June 11 Q1 2026 earnings is the catalyst, and it is asymmetric to the downside given 53x trailing P/E. The Q4 2025 capex jump (TWD 92M vs TWD 11M prior year, 8x) is the only forward signal in disclosed data — promising but one quarter, not a cycle. Real business, wait for the print.
7. Holy Stone (3026.TW) — parabolic on the smallest float, the weakest underlying business in the swarm. Hybrid manufacture-plus-distribution model where mfg vs distribution split is not disclosed (single biggest information gap in any name here). One sell-side analyst with a stale TWD 160 PT — stock at TWD 433.5 trades 2.7x the published PT. 22% insider ownership concentrated in the Tang family (Chair + President since 1994, combined-role governance flag, no disclosed succession). Q1 2026 was good (net margin 13% vs 8.2% prior year, EPS +19% beat) but extrapolating one quarter to FY26 EPS doubling is recency bias at scale. P/E 66.6x trailing on a Tier-2/3 MLCC maker that has never traded above 25x trailing P/E in the last decade except at the prior cycle peak. Mediocre business at a peak-cycle price.
4. Incentive Alignment Screen
| Ticker | Guidance Tendency | Follow-Through Rate | Incentive Alignment | Hurdle-to-Model Gap | Key Flag |
|---|---|---|---|---|---|
| Murata 6981.T | Erratic (5 beats / 6 misses, 36pp std dev) | 70% | Yellow — 1.83% insider, ROIC in PSU, modest comp | Tight on FY27 ¥350B OP target | SAW filter "evaluating strategic options" → ¥48.9B impairment was 2-quarter telegraphed; one real weasel |
| TDK 6762.T | Conservative sandbagger (two upward revisions in FY3/26) | 83% | Green — ISS QS=1, PSU includes ROIC 8.1% (currently behind at 6.7%) + TSR + ESG | Tight — ROIC hurdle at risk, validates non-decorative | Saito personal stake ¥41M (~4 years comp); alignment institutional not personal |
| Yageo 2327.TW | Straight shooter trending conservative (9 beats / 11 misses last 11 quarters) | ~80% | Green — Pierre Chen ~20% personal, professional bench externally sourced | N/A (Taiwan disclosure regime; restricted stock with performance targets introduced 2024) | Founder share-shuffling Oct 2024-Apr 2025 (estate planning likely, but pattern-matches asset-shuffling); audit + comp + nomination committees all chaired by ex-Vice-Chair Cheng-Ling Lee |
| Walsin 2492.TW | Cannot verify (Taiwan disclosure opacity) | n/a | Yellow — 41.7% insider but Chiao family wealth in Walsin Lihwa parent, not Walsin Tech | N/A — no public hurdle disclosure | Pricing posture (case-by-case only, no broad hike) is the structural admission of weak pricing power; ISS QualityScore 10 |
| Taiyo Yuden 6976.T | Erratic at cycle turn → sandbagger in recovery (3 consecutive beats post-trough at +28%, +71%, +45%) | 75-85% | Yellow — 1.67% insider (Japan-normal), TSR-vs-TOPIX in LT comp is positive | Cannot verify hurdle calibration | FY3/22 buyback at peak + FY3/24 capex into trough = textbook bad cyclical capital allocation; balance sheet levered ¥106B over 3 years |
| PDC 6173.TWO | Cannot verify in English (TWSE MOPS Chinese-only) | n/a | Green by ownership (47% insider, founder + Walsin/PSA strategic) | N/A | Disclosure gap is the central flag — board composition, exec bios, RPT pricing on Walsin powder sales all unverifiable; no formal sell-side anchor |
| Holy Stone 3026.TW | Cannot verify (single stale analyst PT, no consensus) | n/a | Yellow — 22% Tang family but Chair+President combined role, no disclosed succession | N/A | 84.5% dividend payout in a capex year (NT$3B Hokkaido + Yilan commitment); declining R&D intensity 3.5% (FY23) → 2.0% (FY25) |
Flags resolved: Murata's SAW filter weasel and Yageo's share-shuffling are both worth monitoring but neither is a thesis-breaker. Walsin's case-by-case pricing posture is the most informative single data point in the screen — it reframes Walsin as a follower, not a participant, in the high-end pricing recovery. TDK's PSU ROIC hurdle being currently behind glide path is the only example of a hurdle with genuine payout-at-risk, which is governance gold. Holy Stone, PDC, and Walsin all share the Taiwan/TPEx disclosure gap that means foreign investors are working with less than the local information set.
5. The Interesting Paradox
The market is rewarding the lowest-quality businesses with the steepest re-rates. Holy Stone (TWD 433.5, +470% off the low, P/E 66.6x) and PDC (TWD 175, +400% off the low, P/E 53.8x) are the two highest-multiple, smallest-float, weakest-governance names in the swarm — and they have outrun every higher-quality peer. Walsin (+245%) and Yageo (+207%) are next. Murata at +234% from the FY25 low is in the same range but on a fortress-grade business. TDK at +105% in the last 12 months is the laggard among the names that have rallied. The paradox is that quality has performed worst on a relative basis in this cycle. The market is treating MLCC the way it treats every cycle inflection — small caps move first and hardest because the float can't absorb the AI-narrative-driven flow. The implication for entry: the highest-conviction longs (Murata) sit at the worst points in their own 5-year valuation range, while the lowest-conviction names (Holy Stone, PDC) are the most overshoot-vulnerable. Pink's question isn't "which MLCC name to buy" — it is "do I chase the parabolic moves at peak risk or do I pay full price for the only name worth holding through the cycle?" Murata is the answer to the second question. The pair trade (long quality / short parabolic small-cap) is the answer to harvesting the dislocation.
The cleanest pure-play (Taiyo Yuden) being the worst-priced entry is a separate paradox worth flagging — owning the highest beta to your stated thesis means you want it at the cycle trough, not after a 4-bagger. The boring middle-of-fairway names (Walsin, Yageo) are doing what they're supposed to do (modest cyclical mix-shift recovery, sensible price moves), but the swarm work was prompted because the small caps were running, which means even disciplined research is being pulled toward the wrong end of the quality spectrum. The fix is the action ladder below.
6. Action Ladder
Murata (¥6,735) — primary long. Start 50% of target at current price (or ¥6,500 on intraday weakness), 25% on a pullback to ¥6,000-6,200, 25% reserved for either the Jul 31 FY26 results reset or a ¥7,000 breakout confirmation. Hard stop on two consecutive quarters of negative YoY revenue or an FY27 guide cut below ¥320B OP; time stop H2 2026 if FY27 op income doesn't track to the +35% YoY guide.
Walsin (TWD 262) — best risk-adjusted pullback play. Zero at current. Set GTC at TWD 225 (1.0-1.5% starter), TWD 200 (add to 2.0-2.5%), TWD 175 (high conviction at 3.0%+). Hard stop -25% from average cost. Time stop H2 2026 if Q2 print shows GM stuck below 19%.
Yageo (NT$572) — second long, pullback only. Zero at current. Starter at NT$480-520 (30% of target), base at NT$420-460 (40%), conviction at NT$340-380 (30%). Hard stop -25%. Re-evaluate at any quarter showing sequential revenue declining + management guiding for sequential weakness.
TDK (¥3,179) — quality play if MLCC slice is acceptable trade-off. Zero at current. Starter at ¥2,800-2,900 (50-day MA retest, -10%), base at ¥2,400-2,500 (Q3 results gap-up base, -22-25%), conviction at ¥2,000-2,200 (200-day MA, -30-37%). April 28, 2026 FY3/27 guidance is the structural one-sided catalyst — even a modest disappointment compresses the multiple.
Taiyo Yuden (¥8,231) — highest cyclical torque, only on hard pullback. Zero at current. Starter at ¥6,000-6,500 (-22-28%), full position at ¥5,000-5,500 (-34-40%). Stop on two consecutive earnings misses, Murata BB <0.9 sustained, AI capex revision from any top-4 hyperscaler, or JPY through ¥140/USD.
PDC (TWD 175) — speculative tracker only. Zero at current. Starter zone TWD 95-110, core at TWD 75-85, conviction at TWD 55-65. June 11 Q1 2026 earnings is the catalyst — buy into confirmation, not ahead of it. Maximum 4-5% position even at full conviction (cyclical + Taiwan TPEx liquidity + diligence gap discipline). Hard stop -25%, no averaging down on broken thesis.
Holy Stone (TWD 433.5) — pass; watch zone only. No position at current. Watch zone TWD 280-320 (-26-35%); even at the watch zone size 1-2% only. The single sell-side PT of TWD 160 is stale but it anchors the asymmetry — the bear case is well below the watch zone.
The single highest-priority action today: nothing on the MLCC names themselves. Run a /profile or /deep-dive on Sakai Chemical (4078.T), flagged by the Taiyo Yuden, TDK, and Walsin sub-agents as the under-priced barium-titanate powder bottleneck with sub-100nm sole-supplier dynamics on next-gen MLCC dielectric. Same theme exposure, ~3-5% the market cap of Murata, less consensus crowding, less retail flow distortion.
7. What This Doesn't Tell You
The Chinese-language MOPS gap is the biggest blind spot for the Taiwan names (Yageo, Walsin, Holy Stone, PDC). Board composition, related-party transactions, executive comp granularity, and major-shareholder tables all live in Chinese-language filings; English-language disclosure is materially thinner than US 10-K + DEF 14A norms. The PDC sub-agent flagged this most explicitly. Walsin's PSA group powder supply relationship with PDC is a related-party transaction that should be disclosed in the Taiwan annual report but is not verifiable in English — could be neutral, could be slightly favorable to one side or the other. Holy Stone's manufacture-vs-distribution revenue split is the single most important undisclosed metric in the swarm.
The cycle-timing gap is the central macro question. The May-2026 MLCC price hikes are typically read as cycle-confirmation signals (pricing returns to the high end first, the tail last; Taiyo Yuden led 6-13% on commodity, which closes the loop). Historical MLCC up-cycles run 18-30 months from trough; this one started Q3/Q4 2024, so we are 12-18 months in. Plausible 6-18 months of cyclical tailwind remaining. But Q3 2026 book-to-bill prints from Murata, Samsung Electro-Mechanics, and Taiyo Yuden are the binary read on whether this is normal mid-cycle pricing power or a late-cycle rollover signal. Pink should set alerts for those.
The Apple battery vs MLCC separation problem at TDK is structural. ATL silicon-anode small-cell batteries are ~55% of revenue and ~89% of segment OP. Anyone framing TDK as an "MLCC peer" is mispricing the business. The cleaner framing is: TDK is an Apple battery proxy with HDD-head duopoly + AI server passive + auto MLCC + sensor optionality stapled on. If the iPhone 17 silicon-anode ramp disappoints or Samsung SDI takes share, TDK breaks regardless of what happens in MLCC. This is not a hidden risk — it's just not visible if you only look at the MLCC headline.
Sakai Chemical (4078.T) is the missing seventh name. Multiple sub-agents (Taiyo Yuden deep-dive Section 5, TDK deep-dive Section 5, Walsin checklist) independently flagged it as the upstream alpha candidate — barium titanate dielectric powder for high-cap MLCC, multi-decade process moat, only 3-4 suppliers globally produce qualified grades, ~3-5% of Murata's market cap with structurally higher long-run growth tied to high-cap mix shift. It was not researched in this swarm. That's the cleanest follow-up; same thesis exposure, much less consensus crowding, much less retail-flow distortion. A separate /profile run is the highest-leverage next step.
The Samsung Electro-Mechanics gap is also worth flagging. Korea's chaebol peer (009150.KS) is ~22% global MLCC share, the closest credible rival to Murata at high-end and the AI-server MLCC volume leader per Digitimes (~40% AI-server share). It was named in every sub-agent's competitive set but not researched in this swarm. Trades at ~16x forward P/E vs Murata's 28x — the cheapest entry to Tier-1 MLCC exposure if you can tolerate chaebol governance discount. The TDK mgmt-dd specifically flagged Samsung Electro-Mechanics as the realistic competitive threat in 100V+ automotive — not the Chinese new entrants. Worth a swarm extension.
The cycle assumption baked into every individual name's forward valuation is roughly the same: AI capex extends through 2027, auto recovery accelerates in H2 2026, and pricing power holds in the high-end tier. None of these are individually high-conviction. AI capex digestion in 2027 is a real risk (TDK deep-dive bear case, Murata deep-dive Section 15 risk #1, Taiyo Yuden bear case ¥4,500-5,500), and a single hyperscaler guide-down would cascade through the entire swarm in days. The correlated bet across all 7 names is high; the diversification benefit of owning multiple MLCC names is low; the right portfolio construction is 1-2 names at conviction sizing, not 7 names at trace sizing.
Sources
- Sub-agent outputs:
~/claude/output/profile/{6981-t,2327-tw,6976-t,6762-t,3026-tw,2492-tw,6173-two}-profile.md - Sub-agent deep-dives:
~/claude/output/deep-dive/{ticker}-deep-dive.md - Sub-agent mgmt-dd:
~/claude/output/mgmt-dd/{ticker}-mgmt-dd.md - Sub-agent checklists:
~/claude/output/checklist/{ticker}-checklist.md(Walsin:2492-tw-buy-checklist.md) - yfinance API (price, valuation, holders, recommendations) — accessed 2026-05-20 and 2026-05-21 across sub-agents
- SemiAnalysis mirror searched at
~/Dropbox/Wafflebun/KB/wiki/semianalysis/— no MLCC-dedicated coverage 2024-2026; one stale Feb-2022 bearish Murata roundup that has been partially validated (SAW filter impairment) and partially superseded (AI server demand)
Topics
- green-finance