AIDC供电新方案有望助力SiC/GaN打开成长空间
Orient Securities·2025-11-04 08:16

Investment Rating - The report maintains a "Positive" investment rating for the electronic industry, particularly focusing on the potential growth of SiC/GaN power devices driven by new power supply solutions for AI data centers [6]. Core Insights - The demand for AI servers and data centers is expected to open up growth opportunities for SiC/GaN power devices, with new power supply solutions like HVDC and SST becoming increasingly important [3][11]. - The report highlights that the transition to 800V HVDC power supply architecture is anticipated to significantly enhance power efficiency and reduce operational costs for data centers [9][28]. Summary by Sections 1. AI Data Center Power Requirements - The power requirements for AI data centers are increasing, with single cabinet power rising from 5-8 kW to 20-50 kW, and potentially exceeding 100 kW [19]. - NVIDIA is promoting the transition to an 800V HVDC power supply architecture, which is expected to be fully implemented by 2027 [26][28]. 2. SST (Solid State Transformer) - SSTs are characterized by high efficiency and compact size, making them suitable for modern power supply needs [37]. - The adoption of SST technology is expected to become mainstream in future power supply solutions, significantly improving space utilization and power efficiency in data centers [47]. 3. HVDC and SST Impact on Power Semiconductors - The report indicates that HVDC and SST solutions will enhance the performance requirements for power semiconductors, creating a favorable environment for the penetration of SiC/GaN devices [56]. - The market for SiC/GaN devices is projected to reach $2.7 billion by 2030, driven by their application in 800V HVDC data center power systems [11][68]. 4. SiC/GaN Growth Potential - The report identifies key companies that are well-positioned to benefit from the growth in SiC/GaN devices, including industry leaders like Innoscience and Tianyu Advanced [3][14]. - The penetration rates for SiC and GaN in the power semiconductor market are expected to increase, with forecasts suggesting a rise in demand due to AI computing facilities [68].