In recent years, the rapid development of Artificial Intelligence (AI) technology has driven explosive growth in the AI server market. According to the 2025 China Artificial Intelligence Computing Power Development Assessment Report released by IDC and Inspur Information, the global AI server market size reached USD 125.1 billion in 2024 and is expected to grow to USD 158.7 billion in 2025.
1- AI Server Requirements for Inductors
Compared with traditional servers, AI servers have significantly different requirements in terms of performance, power density, and energy efficiency, all of which directly affect inductor specifications.
First, AI servers are typically equipped with high-performance GPUs or dedicated AI chips, which often operate in high-current environments. This creates higher requirements for an inductor’s saturation current capability. If the inductor’s saturation current is insufficient, it may cause overheating, functional failure, or even damage, severely impacting system stability.
Second, AI servers require higher power density to address challenges such as limited data center space and difficult heat dissipation. This means inductors must maintain a compact size while offering lower DC resistance (DCR) to reduce thermal loss and improve overall efficiency.
Third, AI servers have extremely stringent power conversion efficiency requirements to reduce overall energy costs in data centers. Therefore, inductors must have excellent high-frequency characteristics to match the needs of next-generation high-frequency DC-DC converters.
Finally, AI servers operate under sustained high loads, creating a relatively harsh internal temperature environment. Inductors must therefore offer good thermal stability and long-term reliability.
2- Applications of Inductors in AI Servers
As an indispensable core component in AI servers, inductors are widely used in multiple critical modules, with key functions including energy storage, signal filtering, noise suppression, and voltage regulation. Given AI servers’ high-performance, high-power-density, and high-reliability requirements, inductors play a crucial role.
2.1 Power Management Systems (DC-DC Converters, Voltage Regulation Circuits)
Core AI server components such as GPUs, CPUs, and AI accelerators demand highly stable and efficient power delivery. To achieve this, servers often employ high-efficiency DC-DC converters to provide stable voltage output, with inductors being a critical component in these converters.
In buck DC-DC converters, inductance values typically range from 0.1–0.68 μH, with operating currents around 60A, saturation currents of 60–120A, and package sizes under 12mm. Inductors store and release energy to smooth input voltage fluctuations and deliver stable output current, ensuring the server operates efficiently to meet the massive data processing and storage needs of data centers.
2.2 Signal Filtering and Noise Suppression
In primary AC-DC converters, common-mode chokes, ferrite beads, and differential-mode inductors are used to suppress high-frequency noise and ensure signal integrity. For example, in a 220V AC to DC power supply, inductors can be paired with capacitors to form low-pass filters, effectively removing high-frequency ripple to provide cleaner power for key chips. Additionally, in high-speed signal transmission lines, differential-mode or common-mode chokes are often used as EMI suppression components to eliminate noise and interference.
3- Key Factors for Inductor Selection in AI Servers
The performance of inductors in AI servers directly impacts system stability, efficiency, and reliability. Selecting the right inductor for each application is therefore critical. The selection process should consider several key parameters, including inductance, saturation current, DC resistance (DCR), and operating frequency.
3.1 Inductance
Inductance determines the inductor’s ability to store magnetic energy and affects ripple current size and energy storage capacity. In AI servers, the right inductance value can effectively smooth current fluctuations and improve power delivery stability.
◾ For 48V–12V DC-DC converters, moderate inductance values (several μH to several dozen μH) balance efficiency and transient response.
◾ For 12V–0.75V DC-DC converters, inductance values typically need to be below 1 μH.
3.2 Saturation Current
Saturation current is the DC current level at which the magnetic core enters saturation and loses its ability to store energy. Given the high power consumption of AI processors, saturation current requirements are high.
Selection should be based on load demand and derating design, prioritizing materials with high saturation points and good thermal stability (such as ferrite or soft magnetic alloys). Molded inductors can further reduce magnetic leakage and improve saturation current.
3.3 DC Resistance (DCR)
DCR is the inductor coil’s internal resistance under DC conditions. Lower DCR means less conduction loss, improving overall system efficiency. In high-power-density AI server designs, lowering DCR is key to reducing energy consumption.
Products with low DCR should be selected while ensuring size and power density remain suitable. Molded indutor designs can help balance low DCR with high power density.
3.4 Operating Frequency
AI servers often use high-frequency DC-DC converters for more efficient and compact power delivery. At high frequencies, traditional magnetic materials may suffer from increased losses and heat. Therefore, high-frequency performance is a crucial selection factor.
For high-frequency applications:
◾ Choose low-loss magnetic materials such as FeNi soft magnetic powder or hot-pressed alloy powders.
◾ Optimize structure with flat wire winding, or special packaging to reduce AC losses.
◾ Ensure the selected inductor supports the target operating frequency range.
4- Recommended Inductor Types for AI Servers
Different AI server applications call for different inductor types, each with its own performance advantages. Based on high power density, efficiency, and reliability requirements, the following types are recommended:
4.1 High-Current Power Inductors
Core components in power delivery circuits for CPUs, GPUs, and accelerator cards. Must feature:
① Excellent saturation current performance to avoid inductance drop during core saturation.
② Ability to handle continuous high currents (e.g., 80–100A) with minimal temperature rise.
③ Good high-frequency characteristics with less power loss in DC-DC converters.
④ Low-loss magnetic materials (ferrite, alloy powder) to reduce losses and improve efficiency.
4.2 Molded Inductors
Molded inductors use magnetic powder and molding method to fully integrate coil and core, reducing magnetic leakage and improving saturation current. They offer higher power density and superior EMI suppression, making them ideal for high-power DC-DC converters and CPU/GPU power rails.
4.3 TLVR Inductors
Special designs such as dual-winding integrated inductors can simplify design and reduce components in the PCB. TLVR (Trans-Inductor Voltage Regulator) designs are used for low-voltage, high-current applications with fast transient load changes, offering faster response, lower ripple, and improved efficiency while reducing output capacitance and overall cost.
5- High-Performance Inductors Driving AI Servers Towards Higher Computing Power and Lower Energy Consumption
As a core component in AI servers, inductors play a vital role in power management, signal filtering, and other critical functions.
As one of the leading magnetics supplier, Codaca offers flexible, customizable inductor solutions to support AI servers’ high-performance, low-power development.
（High-performance inductors from Codaca applied to AI servers）
Codaca’s self-developed inductor series are already widely used in AI servers, including：
◾ Customizable through-hole common-mode chokes such as TCAB series for AC-DC systems;
◾ Compact high-saturation high-current inductors such as the CSBA and CSBX series;
◾ Low-loss molded inductors such as the CSAB, CSEB, CSEC, CSHB, and CSHN series;
◾ TLVR inductors such as the CSFED series.
As AI servers’ demand for higher power density, efficiency, and reliability continues to rise, inductor performance requirements will also keep increasing. By understanding AI server inductor requirements and applying scientific selection based on application scenarios, engineers can significantly optimize system performance, enhancing both stability and energy efficiency.