AI data Server

In AI data servers, capacitors are mainly used for power management, filtering, and energy storage to ensure stability and efficiency. High-performance capacitors can reduce power supply noise, improve power quality, support fast charging and discharging, and meet the needs of AI computing for high bandwidth and low latency. In the future, as the demand for higher energy efficiency and smaller components increases, capacitors will develop in the direction of high frequency, high temperature, low ESR (equivalent series resistance), and longer life. The application of new materials and design technologies will further enhance the performance and reliability of capacitors in AI data servers.

Server power supply ACDC/DCDC
BBU backup power
Server Motherboard
Storage SSD/Storage Control
Switch
Gateway Device
Server power supply ACDC/DCDC

企业微信截图_17653462064006

BBU backup power

企业微信截图_17653464582230

Server Motherboard

企业微信截图_17653465323169

Storage SSD/Storage Control

企业微信截图_17653467924416

Switch

企业微信截图_17653471068963

Gateway Device

企业微信截图_17653469482769

Navitas
NVIDIA
Navitas

案例_01(1)

 

YMIN & Navitas GaN Collaborative 8.5KW Server Power Supply Solution

Performance Enhancement

Volume Efficiency: At the same rated voltage (e.g., 500Vdc), the capacitance of the standard φ30*85mm product is increased to 1400μF (compared to 820μF for conventional products, an improvement of approximately 70%).

Size Optimization: Achieves an overall module size reduction of over 55%, or increases the system's base capacitance by 63% without changing the size.

High Temperature Lifespan: At 105℃, the load lifespan at rated voltage exceeds 10,000 hours, with capacity degradation ≤8%.

Ripple Current: Thanks to its low ESR design, the rated ripple current that can be withstood at 105℃ is increased to 19Arms @100kHz, meeting high-frequency switching requirements.

Saves layout space, improves system lifespan and reliability, ensures stable operation under high-temperature conditions, and meets stringent application requirements.

NVIDIA

案例_05(1)

 

YMIN SLF modular capacitors and NVIDIA AI server GB300 BBU backup power solution

Reduced Size and Weight, Improved Charge/Discharge Efficiency: The volume is reduced by 50%-70% and the weight by 50%-60% compared to the original solution. Charge/discharge efficiency is improved by more than 5 times, supporting fast charge/discharge from 10 seconds to 10 minutes.

Powerful Power Output and Long Lifespan: A single module supports 15-21kW power output, perfectly matching the peak power consumption requirements of the GB300, and has a cycle life of over 1 million cycles, ensuring reliable operation for more than 6 years.

Innovative Energy Storage Architecture: Utilizing supercapacitors for millisecond-level instantaneous power compensation, the BBU handles minute-level long-term energy backup, solving the pain points of traditional solutions such as response latency, short cycle life, and high maintenance costs.