The explosive growth in AI computing power has led to the rapid proliferation of GPU clusters and high-density data centers, with single-rack power consumption exceeding 100 kilowatts. Issues such as millisecond-level power fluctuations in GPUs, intermittent mains power outages, PDU failures, and power module malfunctions are frequent. Power supply stability, data security, and simplified maintenance have become core pain points for AI server infrastructure. Traditional lead-acid batteries, ordinary capacitors, and old-style BBUs suffer from slow response times, high internal resistance, short lifespans, and complex maintenance, failing to meet the 24/7 high-density, high-reliability operation standards. Power supply failures can easily lead to downtime, business losses, and customer churn.

Yongming has identified common pain points in the industry and launched corresponding supercapacitor solutions for four core AI server application scenarios.

I. Common Pain Points Across Four Major Industry Scenarios

1. AI Server Rack BBU Backup Power Supply Scenarios

GPU instantaneous power can surge to 150% of rated capacity. Lead-acid and lithium-ion BBUs have high internal resistance and slow response times, making them prone to bus voltage collapse, causing motherboard and GPU crashes and restarts, resulting in computing power interruptions and repair costs. Regular battery replacements continuously increase maintenance expenses.

Traditional energy storage equipment is bulky and space-consuming, compressing rack space for computing power expansion, and has a short cycle life, leading to high long-term hardware replacement costs.

2. RAID Write Cache Power Loss Protection Scenarios

During sudden power outages, RAID cache corrupted data cannot be written to the disk, causing data loss, array rebuilding, and business downtime, resulting in high compensation and manual repair costs.

Older BBUs have capacity that degrades annually, requiring calibration and replacement every 2-3 years. Reliability drops significantly under high-temperature conditions, making them unsuitable for 1U/2U compact servers, continuously consuming maintenance budgets, and resulting in high equipment return rates.

3. Enterprise-grade SSD/AI Storage Disk PLP Power Loss Protection Scenarios

Under high IOPS AI operation, power outages can easily damage the SSD's FTL mapping table, causing direct disk failure and resulting in high after-sales costs for hardware repair and data recovery.

Conventional small-capacity capacitors provide insufficient power supply duration, and batteries age rapidly due to high temperatures, leading to frequent component replacements and increasing the total lifecycle cost of the equipment.

4. AI Server/Data Center PCS Transient Buffer Scenarios

High-power current surges from the GPU can cause sudden voltage drops or overshoots on the bus, triggering system crashes and severely depleting computing resources.

Traditional solutions require peak power redundancy design, increasing BOM and cooling costs by more than 30%; cylindrical capacitors are not conducive to modular layout, compressing project profits.

II. Three Types of Actual Losses Caused by Pain Points

**Financial Losses:** Battery replacement, hardware repair, and redundant system design incur additional hardware costs.
**Manpower and Time-Constraint Losses:** Regular inspections and troubleshooting consume manpower; data recovery and array reconstruction prolong business downtime.

**Brand Losses:** Data loss and service interruptions erode customer trust; high repair rates weaken product competitiveness and impact project cooperation.

The above losses are all structural costs caused by the inherent shortcomings of traditional energy storage solutions, which Yongming's supercapacitor solutions can address specifically.

III. Yongming's Dedicated Supercapacitor Solutions for Four Major Scenarios

Yongming offers a series of supercapacitor products for four major scenarios: BBU backup power, RAID cache protection, SSD PLP power failure protection, and PCS transient buffering. Leveraging millisecond-level response, ultra-low ESR, high-rate discharge, long lifespan, and miniaturization, these products build a complete power supply guarantee system encompassing voltage stabilization, data protection, and transient buffering.

1. AI Cabinet BBU Backup Power Supply

Product: Hybrid Supercapacitor SLF 4.0V 4500F

Parameters: 2.5~4.0V, 4500F, ESR≤0.8mΩ, Continuous Discharge 200A, 1~50ms Instantaneous Compensation, Cycle Life 1 million cycles, Service Life ≥ 6 years

Application: Parallel DC bus, forming a hybrid energy storage system with the existing BBU. The capacitor handles millisecond-level voltage regulation, while the BBU provides long-term backup power.

Value: Solves the shortcomings of lead-acid batteries: slow response, large size, and short lifespan; suppresses downtime; frees up cabinet expansion space; saves battery replacement costs.

2. RAID Write Cache Protection

Product: Double-Layer Supercapacitor Module SDM 13.5V 8F

Parameters: 13.5V, 8.0F, Maximum Discharge 1.5A, Operating Temperature -40℃~+70℃, Automatic Restart after Power Failure, Cycle Life ≥20 years 10,000 Cycles

Application: Connects to the backup power interface of a RAID card, providing power to the cache during power outages to complete data write-back.

Value: Eliminates the need for periodic BBU calibration and replacement; retains cached data during power outages, saving array rebuild costs.

3. SSD PLP Power Loss Protection

Product: SDN 2.7V 180F single-cell capacitor + SDM 13.5V 144F module (5 series, 4 parallel)

Parameters: Single-cell 2.7V/180F, ESR=8mΩ; Module 13.5V equivalent 144F, cycle life ≥ 200,000

Application: 5 series capacitors form a 13.5V module, 4 sets are installed in parallel in the SSD PLP energy storage bay.

Value: Compensates for insufficient power supply from small capacitors and the defects of high-temperature aging of batteries, reducing SSD repair and after-sales expenses.

4. PCS Instantaneous Buffer

Product: Square double-layer supercapacitor SDF 3.0V 330F
Parameters: 3.0V/330F, ESR < 0.8mΩ, Peak discharge 360A, Dimensions 30×20×55mm, Cycles ≥ 500,000
Applications: Parallel connection of PCS DC bus to achieve peak shaving and valley filling in 200ms to second range.
Value: Suppresses bus voltage fluctuations, reduces equipment redundancy design standards, and reduces BOM and heat dissipation costs.

Supplementary Notes:

The SLF series outperforms the Japanese Musashi 3.8V 300F product, achieving domestic substitution and avoiding the risks of overseas supply price increases and delivery delays;

RAID is equipped with long and short extension cables, eliminating the need for board modifications and rewiring;

SDM modules are standardized at the factory, allowing for flexible adaptation by increasing or decreasing the number of parallel connections as needed.

IV. YMIN Solution's Dual Core Values

Technological Value

Millisecond-level instantaneous response, covering a voltage fluctuation range of 1~50ms, significantly reducing the probability of voltage sag and data loss failures;

Ultra-low ESR supports high-current charging and discharging, reducing bus voltage fluctuation by over 40%;

Up to one million cycle life, stable operation across a wide temperature range of -40℃ to 70℃, with no performance degradation in high-temperature environments;

Compact and miniaturized design, reducing volume by 30%~70% compared to traditional solutions, adaptable to 1U/2U high-density racks;

Automatic power-off recovery, requiring no calibration or daily maintenance, simplifying overall hardware design and improving equipment reliability.

Business Value

Mitigating the risks of downtime, data corruption, and hardware damage, ensuring uninterrupted business operations;

Reducing maintenance, calibration, and battery replacement costs by over 80%, with no downtime for component replacement losses;

Freeing up existing battery installation space to add more GPUs and storage devices, increasing single-rack computing power and overall system revenue;

System redundancy and BOM costs decrease by 30%~60%, while rack utilization and single-rack computing power output increase simultaneously;

Significantly reducing the total cost of ownership (TCO) over the entire equipment lifecycle, simultaneously enhancing product competitiveness and brand reputation.

V. Frequently Asked Questions

Q1: Can supercapacitors completely replace traditional batteries in all four scenarios?

A1: Yes. Yongming's SLF/SDM/SDN/SDF series products are tailored to each of the four scenarios, relying on low ESR, millisecond response, and high-rate discharge characteristics to fully replace traditional BBU and battery solutions, adapting to various power supply and protection needs of AI servers.

Q2: The initial purchase cost of supercapacitors is higher than that of batteries, so why is it more cost-effective in the long run?

A2: Supercapacitors boast a lifespan of up to one million cycles, 200 times that of lead-acid batteries. They require no periodic calibration, no mandatory replacement cycles, and zero daily maintenance. While the initial investment is slightly higher, actual tests show that they can save 50%~80% of the overall cost over their entire lifespan, while also mitigating hidden losses from downtime and data loss.

Q3: Are our products compatible with the limited space of 1U/2U compact servers?

A3: Fully compatible. All products feature a standardized, compact design, with RAID and PLP extension cables included. The PCS uses a flat, rectangular structure, significantly reducing size and weight, allowing for easy installation in high-density servers.

Q4: Why is ESR (Equivalent Series Resistance) crucial for AI servers?

A4: ESR directly determines instantaneous voltage regulation capability. The lower the resistance, the smaller the voltage drop when GPU power surges. Traditional energy storage systems typically have an ESR of 5-10mΩ, while Yongming SDF series supercapacitors have an ESR of