According to a research report from Soochow Securities, the explosion in AI computing power is widening the power supply gap in the United States, leading to a significant increase in electricity demand. Cloud Service Providers (CSPs) are now building their own power facilities, with gas turbines being the primary choice, followed by solar-plus-storage solutions. Chinese companies are expected to benefit from the high growth in U.S. power infrastructure construction. Key recommendations include core gas turbine manufacturer Dongfang Electric Corporation (600875.SH), with additional attention on companies like HaiLianXun (300277.SZ), Harbin Electric (01133), and Shanghai Electric (601727.SH). The report also highlights leading solar-storage companies such as Sungrow Power Supply (300274.SZ), Contemporary Amperex Technology Co. Limited (300750.SZ), and EVE Energy Co., Ltd. (300014.SZ). The main viewpoints of Soochow Securities are as follows:
The rapid growth of AI computing is significantly expanding the power supply shortfall in the United States, driving a sharp rise in electricity demand. Projections indicate that by 2030, cumulative AI computing power in the U.S. will reach 153 GW, corresponding to a peak electricity load of 963 GW. Considering a peak load factor of 55%–60%, the required power generation capacity by 2030 will be 1,751 GW, necessitating an average annual addition of 100 GW over the next five years. However, the cumulative planned new capacity in the U.S. from 2026 to 2030 is less than 200 GW, averaging only 50 GW per year—just half of the projected demand. Compounded by poor interconnectivity among the three major U.S. power grids, prolonged power outage durations, and the ongoing retirement of thermal power plants, the rigid power supply gap has become a central challenge for industry development.
For AI data centers, gas turbines are the preferred power supply method, with solar-storage as a secondary option, and CSPs increasingly building their own facilities. AI data centers have stringent requirements for power supply stability, and gas turbines are favored by CSPs due to their reliable operation and low cost per kilowatt-hour. Heavy-duty gas turbines are suitable for large-capacity scenarios, while aeroderivative turbines offer faster delivery times, making them ideal for time-sensitive projects. The International Energy Agency (IEA) estimates that 80 GW of new gas-fired power capacity in the U.S. from 2024 to 2035 will be directed toward data centers. Solar-storage solutions serve as an important supplement due to their renewable attributes and cost-effectiveness, with major providers adopting hybrid power supply models combining gas turbines and solar-storage. Although solid oxide fuel cells (SOFCs) offer rapid deployment, high efficiency, and low carbon emissions, their higher cost per kilowatt-hour limits them to small-scale distributed applications. Currently, Bloom Energy dominates the North American SOFC market, having secured over 2 GW in confirmed orders.
Chinese supply chains have achieved technological breakthroughs in gas turbine manufacturing and core components, positioning them to benefit from the overflow of overseas orders. The global gas turbine industry is experiencing both volume and price growth, with new orders for turbines above 10 MW reaching 58 GW in 2024 and expected to exceed 90 GW in 2025. The average price per MW is approximately 1.5 million yuan, corresponding to a market value of over 130 billion yuan. GE, Mitsubishi, and Siemens collectively hold 85% of the market share, with production capacity booked through 2029–2030 and overseas product prices rising by more than 30%. Chinese manufacturers have achieved localization of F-class heavy-duty gas turbines, with Dongfang Electric, Shanghai Electric, and Harbin Electric serving as core OEMs. Companies like Yingliu Co., Ltd. have made breakthroughs in turbine hot-section components, while Jereh Group, as an authorized Siemens integrator, has entered the North American aeroderivative gas turbine market. The localization rate of core components continues to increase, providing significant opportunities for order spillover amid strong global industry momentum.
Investment recommendations highlight that the AI computing boom is exacerbating power shortages in the U.S., driving substantial demand for new power capacity. CSPs are increasingly building their own facilities, prioritizing gas turbines followed by solar-storage solutions. Chinese companies stand to benefit from the robust growth in U.S. power infrastructure construction. Key recommendations include core gas turbine manufacturer Dongfang Electric, with additional attention on HaiLianXun, Harbin Electric, and Shanghai Electric. Leading solar-storage companies such as Sungrow Power Supply, Contemporary Amperex Technology Co. Limited, and EVE Energy Co., Ltd. are also emphasized. Risks include intensified competition, unexpected policy changes, slower-than-expected adoption of low-voltage DC energy storage solutions for AI data centers, and insufficient raw material supply.
Comments