SiC & GaN Wafers: Power Solutions for AI Data Center PSUs

With the rapid development of artificial intelligence (AI) technology, the demand for computational power in data centers has been growing exponentially. This surge in demand not only imposes higher requirements on data processing capabilities but also presents significant challenges in power consumption. As AI chips continue to enhance their performance, their power consumption is also increasing. To address this challenge, data centers require power supply units (PSUs) with higher power capabilities while also adhering to spatial constraints, such as fitting within server rack dimensions. Consequently, adopting power solutions with higher power density has become crucial. In recent years, power device manufacturers have introduced PSU solutions based on silicon carbide (SiC) and gallium nitride (GaN) devices. PAM-XIAMEN can supply SiC and GaN wafers for research and production in AI data center PSUs, please contact our sales team at [email protected] for detailed specifications.

1. Advantages of Wide Bandgap Semiconductors

In traditional data center power designs, silicon-based MOSFETs are commonly used in AC/DC and DC/DC converters. However, as switching frequencies increase, the switching losses of traditional silicon devices rise sharply, making them unsuitable for high-power-density power supplies.

In contrast, SiC semiconductor, with a bandgap of 2.26 eV, enables SiC devices to operate at higher voltages and temperatures. This significantly improves power conversion efficiency and reduces energy losses. Additionally, GaN, with an even wider bandgap of 3.39 eV, supports higher switching frequencies and lower conduction losses. These characteristics allow data center power supplies to achieve greater power density and efficiency.

2. Applications of Silicon Carbide and Gallium Nitride in Data Centers

In AI data centers, SiC and GaN each possess unique advantages and application areas. Due to its extremely low reverse recovery loss, SiC wafer can effectively reduce energy consumption and is widely used in the power factor correction (PFC) section of AI server power supplies. Currently, many companies are replacing silicon diodes with SiC diodes and silicon MOSFETs with SiC MOSFETs to improve power supply efficiency.

GaN, on the other hand, benefits from its smaller gate capacitance, lower output capacitance, lower on-resistance, and minimal reverse recovery charge, which result in lower switching and conduction losses. As a result, GaN is primarily applied in the PFC and high-voltage DC/DC (direct current-to-direct current) sections of server power supplies, where GaN MOSFETs replace silicon MOSFETs to further enhance system efficiency.

Against the backdrop of rapidly evolving AI technology, wide bandgap semiconductor materials such as SiC and GaN, with their superior performance, are becoming essential tools for improving data center power efficiency, reducing power consumption, and shrinking device size. In the future, as these technologies continue to advance, wide bandgap semiconductors will play an increasingly vital role in AI data centers, driving the industry towards higher efficiency and greater sustainability.

Whether you need wide bandgap semiconductor wafers for research or for industrial applications, please contact us email at [email protected] and [email protected].