AI servers are restructuring MLCC demand, giving Yageo and Murata pricing power and better profitability from a richer high-end mix. Per Korean media, AI servers need five to 13 times more high-capacitance MLCCs than conventional servers, pushing high-end lead times to five to over 10 months while standard parts sit at 14-18 weeks. Murata's high-capacitance lead times reached 30 weeks in July. New capacity has slipped into 2027, and makers are raising prices and signing long-term deals. x.com 

Murata, Samsung Electro-Mechanics and Taiyo Yuden benefit from a narrow shortage of the high-capacitance, low-inductance MLCCs AI accelerators need, even as commodity capacitor pricing stays soft. MLCC content per rack rises from about $1,530 in GB300 to $4,320 in VR200. Taiyo Yuden's book-to-bill is about 1.7x, Murata's high-end utilization about 95%, and SEMCO raised prices 30%. Murata has the strongest position, SEMCO the clearest earnings conversion, Taiyo Yuden the cleanest exposure. x.com  … Show more

Japanese materials and component suppliers will inevitably cede share to China because they are too cautious about expanding capacity. Sinocera said its key raw materials for high-end MLCCs passed qualification at some customers; those customers may keep using Chinese suppliers through a downcycle, increasing pressure on Japanese firms. Taiwanese stocks are more attractive than Japanese materials and components plays; unlike Chinese firms, Taiwanese companies expand capacity and return capital. … Show more

Murata Manufacturing may be a hidden AI infrastructure winner, with MLCCs a possibly overlooked bottleneck beyond GPUs and HBM, per a deep dive by @damnang2. AI servers need tens of thousands of MLCCs; per the piece a GB200 NVL72 rack may use around 440,000, raising the value of Murata's high-end capacitors. Its moat is materials control, reliability, qualification and PDN design-data lock-in. A second re-rating may follow if datacenter revenue, pricing power and AI demand keep compounding. … Show more

Silicon capacitors are in tight supply and could take off the way MLCC makers did, with Murata, Samsung Electro-Mechanics and AP Memory the most exposed. Per @jukan05, Samsung Electro-Mechanics is demolishing a parking lot to build a new fab because ABF substrate demand is so strong and it lacks greenfield space. x.com 

Murata, Samsung Electro-Mechanics, Taiyo Yuden, Yageo and TDK look priced too hot even though MLCC fundamentals are booming. Taiyo Yuden cautioned that with 30-week lead times, ODMs are buffering inventory and part of the order book is panic buying. Samsung Electro-Mechanics' CFO said low-to-mid tier margins could face pressure in 18-24 months as Chinese rivals expand. Morgan Stanley and Citi analysts warned that at forward P/Es above 40x, any late-2026 capex slowdown could hit multiples hard. … Show more

Taiyo Yuden is the wrong way to buy AI server exposure, because Murata and Samsung Electro-Mechanics supply all the MLCCs for AI servers. Murata is cutting prices on low-end MLCCs, where Taiyo, Yageo and Walsin are exposed, to fend off Chinese competition while raising high-end prices, and its management has been clear it won't lose low-end share. Signals from Chinese players and Murata were already clear by June. x.com  x.com 

MLCCs and power semiconductors are an overlooked picks-and-shovels play on the still-early AI infrastructure buildout. MLCC demand from AI servers is projected to grow 5x by 2027, with lead times jumping from about 2 weeks to 36 weeks. A move to 800V DC could lift power-semiconductor content from about $9.5K toward $115K per rack. x.com 

MLCC makers such as Murata look due for a selloff while MLCC materials suppliers should benefit. Murata and others trade at very extreme multiples even after price hikes, while materials makers do not, and as MLCC capacity expands, materials should follow with a lag. x.com  x.com 

Musashi Seimitsu is the only maker mass-producing hybrid supercapacitors for AI data centers, and that supply gap could become a structural constraint like CoWoS in 2023. Nvidia built them into power delivery from GB300, about 300 per NVL72 rack, implying demand on the order of ten million units in 2026 versus Musashi capacity of 6.5 million by Q3 2026; Ming-Chi Kuo notes the count may be cut. Rubin should use more, Skeleton and Jianghai trail, and the 800V HVDC shift needs requalification. … Show more