Everbright Securities has released a report indicating that the core cause of electricity shortages in the United States is the continuously strengthening capital expenditure expectations for data centers. Based on this, GridStrategies has significantly revised upward its forecast for US summer peak load growth from 64GW in 2024 to 166GW in 2025 for the period 2025-2030. As of mid-October 2025, the planned capacity of US data center reserve projects since January 2023 has reached 245GW. As data centers continue to become operational, peak loads will rise accordingly, making it difficult to meet demand relying solely on stable power sources. The electricity shortage issue in the US is driving increased demand for power system reliability, with gas turbines, electrical equipment, and energy storage sectors expected to benefit substantially.

Key viewpoints from Everbright Securities are as follows:

Regarding the causes of US electricity shortages, Everbright Securities believes the core reason is persistently strong capital expenditure expectations for data centers. This has led GridStrategies to sharply increase its peak load growth forecast. A mismatch exists between data center capital expenditure expectations and actual demand, as well as between actual demand and infrastructure capacity, creating uncertainty in the actual implementation timeline for data centers.

Flexible power sources like energy storage and Solid Oxide Fuel Cells (SOFC) can effectively compensate for load gaps. Future new power capacity additions in the US will be dominated by natural gas, with EIA projecting additions of 7GW, 7GW, 16GW, 8GW, and 7GW from 2026 to 2030 under current project plans. Meanwhile, other stable power sources show essentially no growth, and coal-fired units face significant retirement pressure. With data center reserve project capacity reaching 245GW, peak loads will increase as these centers come online, making reliance on stable power sources alone insufficient.

Calculations of US electricity shortage severity under different data center implementation scenarios indicate that without considering flexible power sources, load gaps in 2030 would be 2GW, 35GW, 65GW, and 157GW for implementation paces of 90GW over 5 years, 245GW over 10 years, 245GW over 8 years, and 245GW over 5 years, respectively. When flexible power sources are considered, load demand can be met under all four scenarios.

Load growth shows distinct regional characteristics, primarily concentrated in areas with dense data center construction. Future US peak electricity load growth is mainly focused on transmission regions like ERCOT and PJM, driven predominantly by data centers. PJM and ERCOT are key data center construction regions. As of January 2026, planned data center capacity in Virginia (PJM) and Texas (ERCOT) reached 34.19GW and 26.60GW, respectively. The former benefits from low-latency network advantages, while the latter leverages abundant natural gas resources for power.

In PJM, surging data center demand is projected to push summer peak load from 156GW in 2026 to 222GW in 2036. Combined with reduced effective capacity, PJM's power reserve margin fell to 18.6% in the 2025/26 period, below the 20% safety threshold. This caused capacity prices to surge from $28.92/MW-day in 2024/25 to $269.92/MW-day in 2025/26. PJM is implementing policy adjustments to accelerate power plant construction while prioritizing stable power source development. By June 2025, PJM had 46GW of generation projects under interconnection agreements, with 11.8GW of primarily gas-fired projects entering fast-track approval.

ERCOT's abundant natural gas resources, lower electricity prices, and faster grid connection approvals make it a primary location for newly planned data centers. As of November 8, 2025, ERCOT received 226GW of large load interconnection requests, including 164GW from data centers. This is expected to drive summer peak load from 87GW in 2025 to 138GW in 2030. ERCOT forecasts its power system reliability will become insufficient by 2028. In response, ERCOT is accelerating the construction of reliability-enhancing power sources like energy storage and gas power. By the end of 2025, ERCOT had 441GW of generation projects in the interconnection queue, including 175GW of energy storage and 54GW of gas-fired power.

Regarding investment targets: 1) Gas Turbines: Amid high market demand, overseas leading gas turbine manufacturers face capacity constraints. Chinese companies are well-positioned to increase their market share. Companies to watch include Dongfang Electric Corporation (600875.SH, 01072) and Shanghai Electric (601727.SH, 02727). 2) Electrical Equipment: Growing US grid infrastructure demand highlights potential in supply-constrained transformer segments. Companies to watch include Jinpan International Ltd (688676.SH), Sieyuan Electric Co., Ltd. (002028.SZ), and IGO Limited (002922.SZ). AI power architecture upgrades can effectively improve power efficiency, favoring the adoption of HVDC and SST solutions. Companies to monitor include Shenghong Co., Ltd. (300693.SZ), Sifang Co., Ltd. (601126.SH), Megmeet Electrical Co., Ltd. (002851.SZ), and Hope Electric Co., Ltd. (603063.SH). 3) Energy Storage: Effective for enhancing short-term grid reliability. Companies to watch include Sungrow Power Supply Co., Ltd. (300274.SZ) and Canadian Solar Inc. (688472.SH).

Risk analysis includes potential delays in AI data center construction, US trade risks for exports, and raw material price volatility.