A recent analysis highlights that structural electricity shortages in the United States are creating long-term, high-certainty investment opportunities, with gas turbines, power equipment, and energy storage emerging as key beneficiaries. 1) Gas Turbines: Leading international gas turbine manufacturers face capacity constraints amid strong market demand, creating opportunities for Chinese companies to increase their market share. Companies such as Dongfang Electric and Shanghai Electric are well-positioned. 2) Power Equipment: Growing U.S. grid infrastructure needs are expected to benefit segments with supply limitations, particularly transformers. Firms like Jinpan International, Siyuan Electric, and Yageo are noteworthy. Additionally, upgrades in AI power architecture, including HVDC and SST solutions, are set to enhance power efficiency. Companies including Sinexcel, SF Electric, Megmeet, and Hope Electric may see increased demand. 3) Energy Storage: Energy storage systems offer a near-term solution to improve grid reliability. Companies such as Sungrow Power and CSI Solar are identified as potential beneficiaries.

The core driver of U.S. electricity shortages is the rising capital expenditure in data centers. Grid Strategies has significantly raised its forecast for U.S. summer peak load growth from 64 GW in 2024 to 166 GW in 2025. Mismatches between projected data center investment, actual demand, and infrastructure readiness create uncertainty in the pace of data center deployment. Flexible power sources like energy storage and solid oxide fuel cells can help address load gaps.

Future U.S. power capacity additions will be dominated by natural gas. The EIA projects annual gas-fired capacity additions of 7 GW, 7 GW, 16 GW, 8 GW, and 7 GW from 2026 to 2030 under current plans, while other stable power sources see little growth and coal plants face retirement pressures. As of mid-October 2025, planned U.S. data center capacity since January 2023 has reached 245 GW. As these centers come online, peak loads will rise, making it difficult for conventional power sources alone to meet demand.

Under different data center deployment scenarios—90 GW over 5 years, 245 GW over 10 years, 245 GW over 8 years, or 245 GW over 5 years—projected U.S. electricity shortfalls by 2030 could reach 2 GW, 35 GW, 65 GW, or 157 GW, respectively, without accounting for flexible power resources. With such resources, these gaps could be mitigated.

Load growth is highly regional, concentrated in areas with dense data center development, primarily within the PJM and ERCOT transmission regions. As of January 2026, planned data center capacity in Virginia (PJM) and Texas (ERCOT) stood at 34.19 GW and 26.60 GW, respectively. Virginia offers low-latency network advantages, while Texas benefits from abundant natural gas resources.

In PJM, surging data center demand is expected to push summer peak load from 156 GW in 2026 to 222 GW by 2036. Coupled with declining effective capacity, PJM's reserve margin fell to 18.6% in 2025/2026, below the 20% safety threshold, driving capacity prices up sharply. PJM is implementing policy adjustments to accelerate generation projects, prioritizing stable power sources. As of June 2025, 46 GW of generation projects had signed interconnection agreements, with 11.8 GW of gas-fired projects in a fast-track approval process.

ERCOT's low power prices, abundant natural gas, and faster interconnection approvals make it a preferred location for new data centers. By November 8, 2025, ERCOT had received 226 GW of large-load interconnection requests, including 164 GW from data centers. This is projected to increase summer peak load from 87 GW in 2025 to 138 GW by 2030, with grid reliability expected to become insufficient by 2028. In response, ERCOT is accelerating the development of energy storage and gas-fired generation to enhance system reliability. By the end of 2025, 441 GW of generation projects were in the interconnection queue, including 175 GW of energy storage and 54 GW of gas-fired capacity.

Key risks include slower-than-expected AI data center construction, U.S. trade policy uncertainties, and volatility in raw material prices.