Guosheng Securities: AIDC Power Management Ultimate Solution SST Industry Chain Upstream Materials and Components Welcome Development Opportunities.

Guosheng Securities released a research report stating that the popularization of Solid State Transformers (SST) will drive demand for wide-bandgap semiconductors such as Silicon Carbide (SiC) and Gallium Nitride (GaN). Among these, SiC, with its high voltage resistance and excellent thermal management capabilities, is mainly used in the input end (AC-DC), while GaN, with its extremely high electron mobility, is mainly used in the output end (DC-DC). At the same time, high-performance soft magnetic materials such as nanocrystalline/amorphous with high magnetic permeability and low hysteresis loss have become the ideal choice for SST magnetic cores. According to predictions from the Chinese big data center, the global solid-state transformer market is expected to grow rapidly at an average compound annual growth rate of 25% to 35% in the next 5-10 years, with magnetic materials and power semiconductors benefiting simultaneously. Guosheng Securities' main points are as follows: - Data center power supply systems are facing profound changes brought about by the AI computing power explosion. - As chip power consumption continues to rise, the power density per cabinet is increasing from traditional less than 60kW to 150kW or even higher levels, posing unprecedented requirements for efficiency, reliability, and space utilization of the power supply system. In typical AI data center projects, the value proportion of power supply facilities is close to 50%, making it a strategic investment direction that cannot be ignored. While mainstream UPS and HVDC solutions are continuously optimized in terms of reliability, their multi-level energy conversion architecture still faces efficiency bottlenecks and occupies large areas, making it difficult to meet future high-density AI data center requirements. In this context, Solid State Transformers (SST), with their system efficiency of over 98% and occupying less than 50% of the space compared to traditional solutions, are moving from a technological concept to an industrial application and are expected to become the core solution for the next generation of IDC power supply systems. Of special note, NVIDIA recently released an 800V DC white paper at the OCP summit, which clearly demonstrates the key role of SST in its next-generation power supply architecture, indicating that SST technology has gained high recognition from global AI computing leaders. Why has SST become the industrial focus? The technical advantages of SST are mainly reflected in high efficiency and miniaturization. 1) In terms of efficiency, SST replaces traditional low-frequency transformers with high-frequency power electronics conversion, increasing system efficiency to over 98%, a significant improvement compared to traditional HVDC's 95.1% and Panama Electric's 97.5%. For gigawatt-scale data centers with huge annual power consumption, every 1% increase in efficiency translates to saving millions of yuan in electricity costs annually. 2) In terms of miniaturization, SST adopts high-frequency magnetic materials and modular design, where the volume of the magnetic core is inversely proportional to the operating frequency, significantly reducing the size of transformers for the same power level. Additionally, through the highly integrated isolation, rectification, and inversion functions of the three-level structure of the input level-isolation level-output level, a significant amount of data center space is saved. SST, like a router for electricity, is proactive and flexible, adapting to green energy requirements, opening up greater imagination for its long-term development. 1) Proactive flexibility: SST is essentially a "software-defined" power router that achieves real-time control through a digital signal processor (DSP) or field-programmable gate array (FPGA), seamlessly integrating grids, dynamically compensating for reactive power, suppressing harmonics, and possessing fault self-healing capabilities, fundamentally improving the intelligence and resilience of power supply systems. 2) Green energy adaptability: China requires that the proportion of green energy in new national hub data centers exceeds 80%, posing a key challenge of how to economically and effectively access renewable energy. With its wide voltage input and multi-port compatibility, SST can directly connect to DC green energy sources such as solar and wind power, avoiding losses from traditional AC-DC conversions. Research shows that new power grids using SST architecture can increase the acceptance capacity of renewable energy by over 50% compared to traditional solutions. Additionally, the dual-active bridge topology used by SST supports bidirectional energy flow, allowing for energy storage during low electricity usage periods and feedback to the grid during peak times to earn price differentials, reducing data center operating costs. Recommendations to watch for SST systems: Beijing Sifang Automation (SST overall efficiency increased to 98.5%, applied in multiple national-level demonstration projects), China XD Electric (subsidiary 2.4MW SST successfully put into operation), Hainan Jinpan Smart Technology (developing a 10kV/2.4MW prototype), Newonder Special Electric (currently researching and developing transformers compatible with SST); SST materials: Hengdian Group DMEGC Magnetics (the world's largest ferrite materials enterprise), POCO Holding (new generation soft magnetic materials with frequencies exceeding 10MHz), Qingdao Yunlu Advanced Materials Technology (products covering a wide frequency range from 50Hz to 100MHz, a global leader in amorphous alloys). Risks include: technical route iteration, downstream demand fluctuations, and worsening competition.