Group 1 - The core viewpoint of the article highlights the challenges faced by the AIDC power supply system, including stability, cost control, and carbon emission management, amidst the rapid growth of global intelligent computing centers and increasing energy consumption demands [1] - It is projected that AIDC energy consumption will reach 77.7 billion kilowatt-hours by 2025, indicating a significant rise in energy needs [1] - The article suggests constructing a diversified energy network and promoting new technologies such as solid-state transformers (SST) to address these challenges [1] Group 2 - By 2030, the domestic AIDC installed capacity is expected to reach 17.7 GW, with the SST market space estimated at approximately 13.27 billion yuan, reflecting a compound annual growth rate of 64.9% [1] - The Chuangye Board New Energy ETF Guotai (159387) tracks the Innovation Energy Index (399266), which focuses on high-growth and innovative companies within the new energy sector, particularly in solar energy, wind energy, electric vehicles, and related services [1] - The Innovation Energy Index has a daily price fluctuation limit of 20%, showcasing its potential for significant market movements [1]

Core Insights - The rapid development of global intelligent computing centers is leading to an explosive growth in energy demand, with China's total intelligent computing scale expected to reach 780,000 Pfops by July 2025, ranking second in the world [1] - The expansion of computing power is causing a significant increase in energy consumption, with data center electricity usage projected to reach between 405.1 billion to 530.1 billion kilowatt-hours from 2024 to 2030, and AIDC energy consumption expected to be 77.7 billion kilowatt-hours in 2025 [1] Group 1: Power Supply Challenges - The power supply system faces three major challenges: 1) Stability: The existing power supply system struggles to adapt to the load fluctuations of intelligent computing centers, which can reach a volatility of 50% [2] 2) Cost Control: Electricity costs account for 57% of operational expenses, significantly surpassing depreciation, rent, and labor costs [2] 3) Carbon Emission Management: New policies require over 80% of green electricity for new data centers, yet 63% of current data centers have a PUE above 1.2 [2] Group 2: Energy Solutions and Efficiency - To overcome power constraints, a diversified energy network comprising solar, wind, storage, and nuclear energy is necessary [3] - Enhancing computing flexibility through dynamic GPU frequency adjustments and task migration between data centers, along with promoting technologies like liquid cooling and waste heat utilization, can lower PUE and improve energy efficiency [3] Group 3: Advancements in Power Supply Architecture - The power supply architecture is evolving from UPS to high-voltage direct current (HVDC), Panama power sources, and solid-state transformers (SST) [4] - SST solutions can achieve system efficiencies of 98.5%, with a single power cabinet outputting 1MW while significantly reducing space requirements, making it well-suited for next-generation intelligent computing centers [4] - The domestic AIDC installed capacity is projected to reach 17.7 GW by 2030, with the SST market space estimated at approximately 13.27 billion yuan, and a compound annual growth rate of 64.9% from 2024 to 2030 [4] Group 4: Investment Recommendations - Companies to focus on include: 1) SST technology leaders: Sifang Co., China West Electric, Jinpan Technology, and TBEA [4] 2) 800V HVDC systems: Zhongheng Electric, Kehua Data, and Hewei Electric [4] 3) AI server power supplies: Magpow, Oulu Tong, and Aike Saibo [4] 4) Solid-state circuit breakers: Taiyong Changzheng and Liangxin Co. [4] - Additionally, potential targets include New Special Electric, New Wind Light, Shenghong Co., and Shuangjie Electric, as well as companies involved in power semiconductors and upstream materials like Yunlu Co., Sanan Optoelectronics, and Inno-Sci [4]