Nvidia Rack Disassembly Reveals New Growth Opportunity, MLCC Value Surges 182%
Supply bottlenecks in AI infrastructure have expanded to fundamental hardware components like multilayer ceramic capacitors (MLCCs), crucial for stabilizing power and filtering noise in AI servers. Both Goldman Sachs and Morgan Stanley highlight MLCCs as entering a historic "volume-price dual increase" supercycle driven by AI. Goldman forecasts the AI server MLCC market to surge over fourfold from ~$1.4B in FY2025 to ~$5.8B in FY2030, a 34% CAGR.
The core driver is a structural supply-demand imbalance. While AI server demand is projected to grow ~4.3x by 2030, industry capacity expands at only ~10% annually, constrained by internal production of equipment and materials. This is compounded by strong demand from electric vehicles. The shortage is evident, with lead times for high-end MLCCs exceeding 20 weeks.
The price cycle has officially begun. Japanese leaders Murata and Taiyo Yuden have raised prices by 15-35% for AI server and automotive MLCCs since April, citing material costs. Japan's April export data confirms the trend, with MLCC export value up 28% year-over-year.
Profit leverage is significant: Goldman estimates a mere 5% price increase could boost Murata's FY2027 operating profit by ~13% and Taiyo Yuden's by up to 37%.
Morgan Stanley's teardown of Nvidia's upcoming Vera Rubin AI rack reveals another catalyst: the MLCC value per rack has skyrocketed 182% from the previous generation to ~$4,320, highlighting the component's growing importance.
With demand set to massively outstrip constrained supply, and price increases just starting, analysts position MLCCs at the beginning of a major, prolonged upcycle.
marsbit06/01 09:06
