The cycle is confirmed. The stocks have front-run it. Here is what to own, what to watch, and what to avoid.
The MLCC up-cycle is real. Capacity utilization hit 87-88% in the second half of 2025 and is approaching the 90% tight supply/demand threshold that triggered the 2017-18 and 2020-21 pricing cycles. Murata, Samsung Electro-Mechanics, Taiyo Yuden, and Yageo all raised prices in spring 2026. AI server MLCC content per rack nearly triples from Blackwell to Rubin ($1,530 to $4,320, per Morgan Stanley’s NVL72 BOM). The 800V EV architecture transition is adding a second independent demand driver at 17.3% CAGR. Barium titanate powder supply is consolidating around a handful of qualified producers, with Murata internalizing and TDK partnering via JVs.
The problem is not the thesis. The problem is the price. Five of the ten names in this universe trade at or above their most recent sell-side price targets. Korean and Taiwanese small-caps show 4-9x stock moves over the trailing twelve months against earnings revisions of only 3-10%. The operating cycle is mid-cycle. The equity cycle is late-cycle. That gap is the central tension in this research.
1. Murata (6981.T) is the only name worth scaling into at current prices. The world’s largest MLCC manufacturer holds approximately 32% global share and commands the high end of every product tier that matters: ultra-high-CV MLCCs for AI servers, AEC-Q200 automotive-grade, and the new 48V VPD power modules for hyperscaler AI racks. Capacitor backlog surged 89.5% to ¥269.2B in Q3 FY3/26, with computers/servers up 28.4%. Management guided FY3/27 operating profit at ¥380B (+34.8% YoY) on only 7.1% revenue growth, which makes sense only through the lens of strategic-to-commodity mix shift. Murata led the spring 2026 price hikes at 15-35% on high-end MLCC while commodity peers managed 5-13%. Balance sheet: ¥653B cash, essentially zero debt, 85% equity ratio. BofA maintains Buy with a ¥6,200 price objective; JPM has Overweight. The fortress here is real.
2. TDK (6762.T) has the highest governance quality in the basket but is an MLCC play by accident. MLCC accounts for roughly 10% of TDK’s revenue; the dominant economic engine is ATL rechargeable batteries (55% of revenue, 89% of segment operating profit via Apple iPhone silicon-anode cells). ISS QualityScore 1 across all four pillars. The TDK-NCI Advanced Materials JV (April 2026, TDK 51% / Nippon Chemical 49%) is the mirror of Murata’s MF Material JV and locks TDK into a named barium titanate supply partnership. BofA Buy at ¥3,500; JPM Overweight at ¥2,700. The stock at ¥3,179 trades near or above both targets. Entry on a -10-15% pullback.
3. Yageo (2327.TW) is the broadest passive-component portfolio with real founder alignment, but the AI MLCC narrative is overstated. Pierre Chen holds approximately 20% personally. Global #1 in chip resistors and tantalum capacitors (via KEMET), #3 in MLCC. Q1 2026 revenue grew 22.7% YoY with 25.2% operating margin (record). The tantalum AI thesis is legitimate: AI now accounts for more than 30% of total tantalum revenue, with the highest book-to-bill across all products. But the MLCC AI story is smaller than the headline suggests. At 1 million Rubin racks over the product lifetime, Yageo’s share of AI rack MLCC would represent approximately 3-6% of annual revenue. Goldman Sachs maintains Buy with a price target of NT$346; the stock trades at NT$572, 65% above. Goldman uses a 15x PB/ROE upcycle multiple; by that methodology the stock is already overvalued on Goldman’s own numbers.
4. Walsin (2492.TW) is the best risk-adjusted pullback play in the basket. Taiwan’s #2 MLCC pure-play with real operating leverage (China utilization 70% recovering to 85%+, Q4 2025 gross margin +130bps YoY). But Walsin is a pricing follower, not a setter: in the spring 2026 round, Murata pushed 15-35%, Yageo 10-20%, and Walsin negotiated case-by-case on loss-making SKUs only. That caps the upside vs Tier-1. Chiao family conglomerate with ISS QualityScore 10 (worst decile). Entry ladder: TWD 225 starter / 200 add / 175 conviction. Hard stop -25% from average cost.
5. Taiyo Yuden (6976.T) is the highest-beta MLCC pure-play with the worst entry math. 71% of revenue is MLCC, the most concentrated exposure in the basket. Taiyo Yuden led the commodity-tier price hikes at 6-13%, the canonical “cycle confirmed” signal since pricing returns to the high end first and the tail last. AI server MLCC sales are growing at 30-35% annually per management’s sell-side meeting. But the stock has run 279% off the May 2025 low to ¥8,231, sits 43% above the 16-analyst mean price target of ¥5,739, and trades at 44x forward P/E. BofA rates it Underperform with a ¥3,800 target (the most bearish sell-side call in the basket). JPM simultaneously upgraded to Overweight at ¥3,600. This is the sharpest sell-side split in the universe. Entry only on a hard pullback to ¥6,000-6,500.
6. Nippon Chemical (4092.T) is the longer-dated upstream operating-leverage play. MLCC dielectric powder accounts for 25.9% of sales, double Sakai’s concentration, and the TDK-NCI Advanced Materials JV locks TDK as a named anchor customer. But FY3/26 operating profit declined 27.7% and missed the mid-term plan target. The +22% revenue growth that triggered the initial research was Q4-only channel fill; full-year was +3.4%. Watch, leaning pass, at ¥3,905. Better entries at ¥3,200-3,400.
7. Sakai Chemical (4078.T) is the near-term capital-return play in the upstream tier. 6.17% buyback plus 4.6% dividend yield on a stock trading at 0.68x book value. Claims 40-50% of the hydrothermal BaTiO3 merchant market, though third-party data from Dongguan Securities puts Sakai’s total market share at 28% (not 40-50%). Murata’s MF Material JV is structurally diluting external powder demand from the single biggest buyer. Watch at ¥3,465.
8. Samsung Electro-Mechanics (009150.KS) is real Tier-1 quality priced like a momentum stock. Global #2 MLCC share at approximately 22%, AI server MLCC volume leader (reportedly 40% of high-end AI server MLCC market), and FC-BGA substrate leader for AI processor packaging. Q1 2026 operating profit rose 40% YoY. But the stock has moved approximately 9x in twelve months to KRW 1,061,000 against a 49.8x forward P/E; pricing follower in the spring 2026 round (5-10% hike vs Murata’s 15-35%); chaebol governance discount. Analyst target dispersion is 4x (Mirae KRW 250,000 vs KB Securities KRW 1,050,000). Pass at current. Slots to rank #5 at fair-value entry around KRW 400-600K.
9. Prosperity Dielectrics (6173.TWO) is the right business at the wrong price. Hybrid model: finished MLCCs plus barium titanate dielectric powders (#2 globally by volume). PSA group ally of Walsin since 2005. Stock doubled in three weeks in May 2026. 53x trailing P/E on FY25 net income growth of only 12% YoY. Q1 2026 earnings on June 11 is the binary catalyst and is one-sided to the downside at these levels. Pass; re-engage TWD 95-110.
10. Holy Stone (3026.TW) trades at the highest multiple for the lowest-quality business. Mid-tier MLCC producer plus semiconductor distribution. 66x trailing P/E. Manufacturing-vs-distribution revenue split is not publicly disclosed, which means the most important quality metric is unknowable from English filings. Single sell-side analyst with a stale price target of TWD 160 vs spot TWD 433.5. Chair and president combined role since 1994 with no disclosed succession plan. Avoid.
The global capacitor market reached $41.23 billion in 2025 and has split into two distinct segments with different economics. Strategic capacitors (high-voltage MLCCs, automotive-grade polymer hybrids, DC-link film capacitors) represent 35-40% of market value but drive 60-70% of industry profits. They command 30-80% price premiums over commodity equivalents, carry lead times of 24-47 weeks, and are specified at the architectural design stage by reliability and supplier qualification rather than price. Only four suppliers hold production-scale capabilities for ultra-high-CV designs above 47uF: Murata, TDK, Samsung Electro-Mechanics, and Kyocera AVX. Hyperscalers have locked 60-70% of available high-CV capacity through exclusive supplier relationships.
Commodity capacitors (general-purpose MLCCs, standard aluminum electrolytics) face 10-20% pricing pressure from Chinese capacity additions. Lead times have normalized to 12-16 weeks. These components function as undifferentiated inputs purchased on price and availability.
This bifurcation explains something that looks contradictory in Murata’s financials: revenue grew only 5% in FY2026 but operating profit is guided to grow 34.8% in FY2027. The mix is shifting from the -18-20% commodity segment toward the +29% AI strategic segment. The top-line number understates the margin improvement.
Each AI server board carries 10,000-20,000 ceramic decoupling capacitors and 300-600 polymer/hybrid aluminum capacitors across all voltage rails. Peak current transients during GPU context switching approach 100A/us. This is 10-15x more ceramics than a conventional server.
Murata’s updated IR Day guidance raised the MLCC count to 15,000-25,000 per AI server baseboard, with the GB300 platform requiring approximately 30,000 MLCCs (about thirty times a mobile phone and three times an automobile). A single rack consumes as many as 440,000 units.
JPM forecasts MLCC industry units to surpass 4.6 trillion in 2026, growing to 6.1 trillion by 2030 at +6% CAGR. Overall MLCC market TAM reaches $25 billion by 2030 at +11% CAGR. The AI server MLCC mix within the total market is projected to grow from 1.1% in 2025 to 4% by 2030. MLCC content per GPU is expected to rise from 4k to 12k (3x in five years).
Barium titanate ceramic materials account for 20-25% of costs in low-capacitance MLCCs, rising to 35-45% in high-capacitance parts. The global MLCC ceramic powder market is concentrated: Sakai Chemical of Japan holds 28% share, Ferro (US) 20%, Nippon Chemical (Japan) 14%, Guoci Materials (China) 10%, Fuji Titanium (Japan) 9%.
The critical technology gap: Japanese companies achieve average BaTiO3 particle sizes of 80-100nm. Chinese companies can only achieve 120-150nm. This 40-50nm gap directly limits layer count. Japanese manufacturers stack 1,200 layers in 0.5-0.6um thin-film dielectric (Murata achieving 1,600); Chinese manufacturers average 800 layers in 1-2um film.
Two JVs are reshaping this landscape. Murata formed MF Material with Ishihara Sangyo and Fuji Titanium in September 2023 to internalize barium titanate supply. TDK formed the TDK-NCI Advanced Materials JV with Nippon Chemical in April 2026 (TDK 51% / NCI 49%). The structural implication: the merchant powder market is shrinking at the top end. Murata is internalizing. TDK is partnering with Nippon Chemical. The remaining merchant base (Samsung E-M, Taiyo Yuden, Yageo, Walsin) is the residual addressable market for independent powder makers like Sakai.
Murata is the only name where sell-side and our own research agree on a Buy at current. BofA has a ¥6,200 target and Buy rating. JPM has Overweight (though their ¥3,500 target is stale from December 2025).
Taiyo Yuden is the sharpest sell-side split: BofA rates it Underperform with a ¥3,800 target (the stock trades at ¥8,231, 117% above), while JPM upgraded to Overweight in April with a ¥3,600 target. The stock has rallied 77% while NTM EPS estimates were revised down. This is the framework’s worst Revision Velocity scenario.
Yageo’s Goldman Sachs Buy at NT$346 uses a 15x PB/ROE upcycle valuation. The stock at NT$572 trades 65% above Goldman’s own target. Both Goldman and our research agree the stock should not be chased at current.
Samsung Electro-Mechanics: JPM’s Overweight at KRW 325,000 was set in January 2026 when the stock was at KRW 276,500. The stock has since nearly quadrupled to KRW 1,061,000. The PT is functionally obsolete.
Two Chinese domestic names from Dongguan Securities: Fenghua (000636.SZ, Buy and Hold, 48x 2026E P/E) and Sanhuan Group (300408.SZ, Buy and Hold, 32x 2026E P/E). These are the domestic-substitution plays, operating in the mid-to-large-size, low-capacitance tier that Japanese and Korean manufacturers are ceding to focus on high-end products.
Four calendar-bound catalysts define the next read on this cycle. PDC (6173.TWO) reports Q1 2026 earnings on June 11; the print is one-sided to the downside at current valuation and will test whether the parabolic move has fundamental support. Yageo’s Q2 earnings and May/June monthly revenue prints arrive July 28 and will show whether the tantalum pricing hikes are flowing through. Taiyo Yuden reports Q1 FY3/27 on August 5; this is the binary read on both the MLCC cycle and (via Nippon Chemical in the same window) whether upstream powder operating leverage is real or channel fill. The Q3 2026 book-to-bill prints from Murata, Samsung Electro-Mechanics, and Taiyo Yuden, available starting mid-October, are the structural read on whether this is normal mid-cycle pricing power or a late-cycle rollover signal. Taiyo Yuden’s Q2 BB already dropped to 0.89.
The single largest correlated risk to this entire complex is AI capex digestion in 2027. A 15%+ capex guide-down from any of Amazon, Microsoft, Meta, Google, or Oracle cascades through all ten names within days. The cycle thesis embeds continued AI server BOM expansion through 2027. The right portfolio construction is one to two names at conviction sizing, not ten names at trace sizing.
Three names surfaced during this exercise were not fully profiled and remain open. Kingboard Holdings (0148.HK) scored 19/30 Watch in the Global Passives Basket and was the only name trading below consensus fair value. It sits upstream of MLCC in the BOM as a CCL and PCB producer with four announced price hikes through April 2026. Kyocera Corporation (6971.T) is the parent of Kyocera AVX, the fourth supplier in the ultra-high-CV AI MLCC oligopoly. Samsung Electronics (005930.KS) owns 23.7% of Samsung Electro-Mechanics and offers a chaebol holding-company discount path to SEMCO exposure.
Research compiled May 20-25, 2026. 10 tickers analyzed across 4 research skills per ticker (profile, deep-dive, management due diligence, pre-buy checklist), plus industry primer, peer comparison synthesis, and 7 sell-side source integrations (BofA, Goldman Sachs, JPMorgan, Capacitor Dossier, Global Passives Basket, Dongguan Securities, sell-side meeting notes). Prices are snapshots from the May 20-24 analysis window; verify before acting.