Zhongtai: Time Constraints on Computing Power Commissioning, SOFC Industrialization Accelerates Its Voyage

Zhongtai released a research report stating that the global AI arms race is driving a surge in computing power capital expenditure, and power availability has become the primary bottleneck in AIDC construction. Although traditional gas turbines have lower unit costs, the delivery of main units and the construction cycle of station sites cannot match the time window for AI commissioning, making self-provided power sources shift from optional to mandatory; SOFC, with standardized modular stacking, can be rapidly replicated at hundreds of MW scale, bypassing grid connection and transmission bottlenecks, and is upgrading from an alternative technology to a mainstream option for AI power supply. Zhongtai's main viewpoints are as follows: The global AI arms race drives a surge in computing power capital expenditure, and power availability has become the primary bottleneck in AIDC construction. With the rapid development of global artificial intelligence, tech giants are engaging in an arms race around computing power, data center capital expenditure has risen sharply, and electricity demand has surged accordingly; however, the aging U.S. power grid, grid connection queues, and slow transmission expansion, combined with continuously lengthening delivery cycles for large power generation equipment, mean that AIDC waits for grid connection and electrical equipment arrival often measured in years. Although traditional gas turbines have lower unit costs, the delivery of main units and the construction cycle of station sites cannot match the time window for AI commissioning. Electricity has become the first constraint on AIDC commissioning, and self-provided power sources have shifted from optional to mandatory. Whoever can supply power faster can lock in computing power capacity first, and thus is expected to secure a position in the AI era. The data center launch window is narrowing, and SOFC solutions with immediate delivery capacity are the first to benefit. Gas turbines, with mature technology and large-scale production capacity, have become the preferred route for AIDC on-site self-provided power, but product delivery cycles reach more than 36 months, and orders through 2028 are locked in. Other on-site energy forms such as reciprocating internal combustion engines, aeroderivative gas turbines, and retired aviation engine conversions have limited capacity, making it difficult to fill the demand gap; while SOFC, with standardized modular stacking, can be rapidly replicated at hundreds of MW scale, bypassing grid connection and transmission bottlenecks, and technically can also match the future 800VDC AIDC power architecture. SOFC is upgrading from an alternative technology to a mainstream option for AI power supply. Focus on the BE company industrial chain; suppliers with better positioning are expected to benefit significantly. The global SOFC player landscape is relatively concentrated. U.S.-based Bloom Energy has cumulatively deployed about 1.8GW and has a backlog of about US$20 billion, making it the most core integrator in the SOFC industrial chain. Based on Bloom Energy successively winning GW-level North American AIDC power supply orders such as up to 1GW from AEP and up to 2.8GW from Oracle, it is expected to become a core catalyst for the SOFC industrial chain in the future. At present, many companies have clearly formed direct or indirect product supply for BE's SOFC systems. In the future, as BE's 2027 procurement scale gradually becomes clear, more companies with positioning advantages are expected to enter BE's supply chain system, and thereby significantly benefit from the rapid increase in SOFC demand. Investment recommendations Focus on Bloom Energy, Chaozhou Three-Circle(Group), Shanghai Vital Deeptech, Chunhui Instrument, Hubei Zhenhua Chemical, Shenzhen JingQuanHua Electronics, Shenzhen Uniconn Technology, JOHNSON ELEC H, Weichai Power, Anhui Estone Materials Technology, Delta Electronics, Ceres Power, Fuelcell Energy, Doosan Group, Jiangsu Yunyi Electric, Shenzhen Kaizhong Precision Technology, Hunan Corun New Energy, ZYNP Corporation, Moon Environment Technology, etc. Risk warnings: risk of AI computing power demand and capital expenditure falling short of expectations; technology route competition risk; BE order execution and revenue conversion risk; U.S. policy and tax credit risk; geopolitical relations and tariff uncertainty risk; risk of untimely updating of information used in the research report, etc.