CITIC SEC: Power supply has become one of the core bottlenecks in U.S. AI data center construction.

CITIC SEC released a research report stating that power supply has become one of the core bottlenecks in U.S. AI data center construction. According to Aterio, the capacity of data centers currently under construction in the U.S. has reached 71.7GW, while new grid capacity in 2027-2028 is expected to be only 25.2GW. The bank judges that the support for this round of U.S. power industry prosperity lies in the contradiction between the relatively high investment returns of data centers and the scarcity of power supply resources: investment in the power system is relatively limited, but early energization can accelerate capital recovery; according to the bank's estimates, if first-year capacity utilization rises from 50% to 75%, project ROIC can increase from 16.5% to 19.8%, providing an economic basis for customers to procure high-efficiency power supply solutions. Considering the long cycle of grid expansion and the high difficulty of fully off-grid operation, the bank expects grid renovation and behind-the-meter power generation to advance in parallel, with hybrid power supply gradually becoming mainstream, and the grid equipment, power transmission and transformation EPC, and behind-the-meter power generation industry chains are expected to benefit together. CITIC SEC's main views are as follows: Power supply has become a core bottleneck in U.S. AI data center construction, and capital expenditure on computing infrastructure is spreading to the power supply industry chain. According to Aterio's official website, as of early September 2026, the U.S. already had 52.4GW of data center capacity, with 71.7GW under construction; according to SemiAnalysis, new U.S. grid capacity in 2027-2028 is expected to be only 25.2GW, just 35% of the capacity currently under construction, corresponding to a gap of about 46.5GW. As the scale of individual data centers rises toward the GW level, project requirements for power supply capacity, reliability, and delivery speed increase simultaneously, but grid planning, equipment production, and engineering construction cycles are all significantly longer than data center construction cycles. The bank believes that the constraint on data center construction has shifted from server and GPU supply to power access, and North America will simultaneously advance transmission and distribution grid renovation, cross-regional transmission, and park-side behind-the-meter power generation construction, with power equipment and engineering services expected to become an important incremental direction for AI infrastructure investment. Demand side: time is money, efficiency is life. Based on the requirements for Tier IV data centers on Uptime Institute's official website, and according to Tom's Hardware, alpha-matica's official website, and announcements and official websites of companies such as Nvidia, ETN, WDC, and GEV, the bank estimates that the total capital expenditure for a 1GW AI data center is about US$49.9 billion, of which main power supply, backup power supply, and grid systems together account for about 17%; annual operating costs are about US$8.4 billion, with electricity costs accounting for about 8%. Investment in the power system is relatively limited, but the timing of power supply determines when computing capacity begins operation. Under the leasing price and utilization assumptions set in the report, annual revenue after full ramp-up can reach US$18.9 billion, with full-lifecycle ROIC of about 16.5%. If the power supply solution enables the project to come online 3 months earlier, the corresponding incremental revenue is about US$4.7 billion, close to the scale of capital expenditure on the main power supply system; if first-year capacity utilization rises from 50% to 75%, project ROIC can rise to 19.8%, an increase of 3.3ppts. The bank believes that data center customers have a high willingness to pay for shortening grid connection and energization cycles, and transformers, switchgear, gas turbines, fuel cells, and engineering services that can accelerate delivery are expected to benefit. Supply side: the pace of grid expansion is difficult to match data center construction demand, and the industry is entering a hybrid power supply stage. According to SemiAnalysis, U.S. data centers' new capacity demand in 2027 is about 39GW, while available grid capacity is only 24.1GW, with about 14.8GW needing to be supplemented by behind-the-meter power generation; in 2028, demand rises to 54GW, while new grid supply is only about 11.1GW, further expanding the behind-the-meter power supply gap to about 42.6GW. On the grid side, Wood Mackenzie estimates that the shortfall in power transformers and substation equipment in 2026 will be about 15% and 8%, respectively; ETN disclosed that transformer lead times were once close to 2 years, and power transmission and transformation construction is also constrained by a shortage of skilled workers. On the park side, fuel cells, aeroderivative gas turbines, and reciprocating gas engines can help projects energize earlier, and GEV's aeroderivative gas turbine shipments in the first half of 2026 have already approached the full-year 2025 level. However, fully off-grid operation still requires solving issues related to power supply redundancy, load fluctuations, and long-term operation and maintenance. The bank believes that hybrid power supply combining front-of-meter and behind-the-meter is more feasible, and the grid equipment, power transmission and transformation EPC, and behind-the-meter power generation industry chains are expected to benefit together. Grid expansion: the pace of grid expansion is difficult to match data center construction demand, and equipment and construction bottlenecks are driving the simultaneous development of behind-the-meter power generation. On the equipment side, ETN's transformer delivery cycle was once close to 2 years and was extended again in 2026Q2. On the construction side, power transmission and transformation projects are constrained by a shortage of electricians and skilled workers, with the training cycle for experienced skilled workers at about 4 years. Supply constraints have gradually been reflected in corporate orders: in 2026Q2, ETN's North America electrical business backlog reached US$15.2 billion (up 33% year over year), GEV's electrification business data center orders in 2026Q2 exceeded US$5 billion, and PWR's power infrastructure backlog in 2026Q2 reached US$43.79 billion. The bank believes that order accumulation provides strong support for subsequent revenue growth in the grid equipment and grid EPC industries. Behind-the-meter power generation: fuel cells, aeroderivative gas turbines, and reciprocating gas engines are accelerating deployment. Fuel cells have the advantages of modularity, high efficiency, and relatively fast deployment; aeroderivative gas turbines have single-unit power of tens of megawatts and can complete installation and commissioning in as little as about 1 month; natural gas reciprocating engines have rapid response and multi-unit parallel capabilities, which can match AI load fluctuations and phased park expansion. On the demand side, Bloom Energy's agreement with Oracle supports procurement of up to 2.8GW; GEV's new aeroderivative gas turbine orders in 2026H1 reached 70 units (up 94% year over year); the joint venture between FTAI and Jereh obtained initial orders of US$1.465 billion; CAT also received an initial 2GW unit order for the Monarch park. On the supply side, Bloom Energy plans to increase existing factory capacity to 2GW by the end of 2026, and CAT will raise its large reciprocating engine capacity target to nearly 3 times the 2024 level, providing support for subsequent deliveries. Considering that fully off-grid operation still faces difficulties such as redundant configuration, load management, and long-term operation and maintenance, the bank expects hybrid power supply combining the grid and park power sources to become mainstream. Risk factors Risk of AI data center construction progress falling short of expectations; risk of gas turbine capacity expansion falling short of expectations; risk of fuel cell capacity expansion falling short of expectations; risk of grid construction progress falling short of expectations; risk of sharp fluctuations in fuel prices such as natural gas; risk of intensifying industry competition; risk of nuclear power policy changes exceeding expectations; risk of corporate electricity expenditure declining more than expected due to a slowdown in global macroeconomic growth, etc. Investment recommendations Current AI data centers have relatively high investment returns, but project ROIC is highly sensitive to commissioning timing and first-year capacity utilization. The bank estimates that the current full-lifecycle ROIC of AI data centers is about 15%20%; if first-year capacity utilization rises from 50% to 75%, project ROIC can increase from 16.5% to 19.8%; for a 1GW AI data center, if the project is commissioned 3 months earlier, the additional cash collection can basically cover the capital expenditure of the main power supply system. Because U.S. grid expansion and grid connection cycles are difficult to match data center construction demand, the bank believes that behind-the-meter power generation is not only a temporary solution to the power gap, but also an important means of improving data center project ROIC. At the same time, because fully off-grid operation still requires solving issues such as power supply redundancy, load fluctuations, and long-term operation and maintenance, the bank believes that hybrid power supply combining front-of-meter and behind-the-meter is more feasible, and the grid equipment, power transmission and transformation EPC, and behind-the-meter power generation industry chains are expected to benefit together. In the behind-the-meter power generation direction, it is recommended to focus on companies related to fuel cells, aeroderivative gas turbines, and reciprocating gas engines; in the grid expansion direction, it is recommended to focus on companies related to transformers and grid EPC.