The Performance Competition of MLCC in the AI Data Center Era: Why Are Murata and Taiyo Yuden Significantly Ahead?

AI data centers are driving MLCC into a new cycle of technological upgrades.
In the past, servers mainly used 12V power supplies; now, they are evolving toward 48V rack power, and in the future, even into the 800V HVDC high-voltage direct current era. Meanwhile, AI platforms like NVIDIA GB200 and GB300 are continuously increasing in power consumption, with GPU core voltages dropping to 0.6V-0.8V, yet single GPUs are drawing over 1000A of current.
For MLCC, the main challenges come from three directions.
First is high voltage. 48V power supplies require higher withstand voltages, greater reliability, better heat resistance, and mechanical stress tolerance, leading to rapid growth in demand for 100V or higher voltage MLCCs.
Second is transient response. AI GPU load changes occur on the nanosecond scale, and the power supply network must have extremely low ESL (equivalent series inductance) and very low impedance; otherwise, voltage drops, performance degradation, or system instability may occur.
Third is space constraints. The PCB area around GPUs is becoming increasingly tight, and engineers want to place more decoupling capacitors as close to the GPU as possible, so MLCCs must achieve small size, high capacitance, and high volumetric efficiency simultaneously.
In response to these needs, the industry is moving toward high-voltage MLCCs, ultra-low ESL MLCCs, and ultra-high capacitance MLCCs.
Among them, Murata Manufacturing and Taiyo Yuden are the most representative companies.
Taiyo Yuden has launched the LWDC low-ESL series MLCCs, which significantly reduce ESL through a reverse electrode structure, making them particularly suitable for AI GPU power supply scenarios. They are also expanding into 100V+ high-voltage MLCCs and high-capacitance MLCCs, actively promoting embedded MLCC technology.
Murata continues to set industry records, maintaining a leading position in small size, high capacitance, and high reliability products.
Their competitive advantage lies in materials.
MLCC may belong to the manufacturing industry, but high-end MLCCs are closer to the materials science sector.
The core technology chain includes:
BaTiO₃ dielectric powder → slurry formulation → thin-layering → lamination → sintering → MLCC
The most difficult and highest threshold stage is the dielectric powder. MLCCs mainly use barium titanate (BaTiO₃) as the dielectric material.
Differences among manufacturers’ BaTiO₃ mainly show in:
Particle size control
Particle size distribution
Rare earth doping systems
Core-shell structures
Grain growth control
These capabilities collectively determine the ultimate performance ceiling.
This is also why, for the same MLCC, Murata can achieve 100μF, Taiyo Yuden can reach 50μF, while most other manufacturers struggle to even reach 22μF.
The reason is that Murata and Taiyo Yuden can make the dielectric layers thinner and stack more layers.
For fixed-size MLCCs, increasing capacitance can only rely on three factors:
Higher dielectric constant
Thinner dielectric layers
More stacking layers
The problem is, as dielectric layers become thinner, material requirements increase exponentially.
If BaTiO₃ particles are too large, when the dielectric layer thickness drops to 0.5μm or less, only two or three grains may remain in each layer.
At this point, issues like leakage, breakdown, and lifespan will rapidly worsen.
One of Murata and Taiyo Yuden’s biggest advantages is their ability to produce extremely fine and highly uniform BaTiO₃ particles, enabling continued thinning of dielectric layers.
Particle size is only the first step. Distribution is often even more important.
If particle sizes vary too much, abnormal grains, voids, and stress concentrations can form after sintering, ultimately reducing reliability and yield.
High-end MLCC manufacturers generally possess the industry’s most advanced particle size distribution control capabilities.
Beyond that is the core-shell technology. High-end MLCCs require special rare earth-doped layers coating the BaTiO₃ core.
The core provides high dielectric constant, while the shell controls leakage current, enhances insulation, and extends lifespan.
This part is often one of Murata and Taiyo Yuden’s most critical technological secrets.
Even with the same powder, sintering processes can cause significant performance differences. Temperature curves, oxygen partial pressure control, soak times, and cooling rates during sintering all influence grain growth.
Truly leading manufacturers can produce ultra-fine powder and maintain small, uniform, and stable grain structures after sintering.
This is also why high-capacitance MLCCs are so difficult to manufacture.
Achieving 100μF requires the stable stacking of hundreds or even thousands of ultra-thin dielectric layers. Any tiny defect in a layer can cause the entire product to fail.
Therefore, high-capacitance products are fundamentally a comprehensive competition in materials science, process control, and yield management.
From the industry landscape, the high-end MLCC market currently roughly follows this hierarchy:
Murata Manufacturing — Industry leader, comprehensive advantage in materials, processes, and products.
Taiyo Yuden — The closest competitor to Murata, maintaining a long-term lead in high-end MLCCs.
TDK — Strong technical strength, continuously catching up with the top tier.
Samsung Electro-Mechanics — Outstanding manufacturing capability, expanding in the AI server market.
Yageo, Fenghua Advanced Technology, and other manufacturers are also continuously catching up.
In the future, the MLCC most needed by AI servers will no longer be mass-market consumer electronics products.
Instead, products must simultaneously possess:
High voltage
High capacitance
Ultra-low ESL
Small size
High reliability
All five capabilities ultimately trace back to the same source: decades of accumulated BaTiO₃ powder technology, core-shell structure design, ultra-thin dielectric manufacturing, and sintering process experience.
This is why, in the AI data center era, the real gap is not just in MLCCs themselves, but in the underlying materials science.
Disclaimer: I hold assets mentioned in this article. The views expressed are biased and do not constitute investment advice. DYOR.