AI走向“大象起舞”:深度剖析PJM电网高频数据,透视AI算力需求
ZHONGTAI SECURITIES·2025-11-27 04:59

Investment Rating - The report maintains an "Overweight" rating for the industry [2] Core Insights - High-frequency power data confirms that the demand for AI computing power is accelerating [3][4] - The AI landscape is evolving from "1 to N," with a positive outlook on Google's performance [4][6] Summary by Sections Industry Overview - The industry comprises 130 listed companies with a total market capitalization of 1,802.515 billion [3] - The circulating market capitalization stands at 1,665.920 billion [3] Power Demand Analysis - In the Virginia DOM region, the average monthly load increment for 2025 is approximately 3 GW, an increase of 0.98 GW compared to 2024 [5] - The average load increment for September to November shows significant year-on-year growth of 73%, 53.2%, and 56.4% respectively [5][28] - The Ohio AEP region shows an average load increment of about 1.34 GW for 2025, with a year-on-year increase of 158%, 223%, and 180% for the same months [5][29] Price Trends - Nighttime price differences have significantly increased, with the price difference for the ARCOLA node reaching 7.94 USD/MWh in October 2025, a 197% increase year-on-year [5][50] - In November 2025, the price reached 13.11 USD/MWh, marking a 680% increase year-on-year [5][50] AI Application Insights - The report identifies four major barriers to AI application: weak cost scale effects, subscription model limitations, high ROI requirements, and the need for a data feedback loop [6][65] - Google is highlighted as having a robust AI ecosystem, leveraging its proprietary TPU chips to reduce computing costs significantly [6][69] Investment Recommendations - Suggested stocks include Alphabet (GOOGL.O), Cipher Mining (CIFR.O) in the US, and Alibaba (9988.HK), Tencent (0700.HK), and others in Hong Kong [6]