With the rapid development of artificial intelligence, the energy consumption of large-scale data centers continues to grow. Understanding the spatial distribution of large-scale data centers is important for optimizing power system planning, increasing renewable energy penetration, and improving energy efficiency. However, detailed information such as location and distribution of large-scale data centers is rarely available, limiting research on their energy consumption and environmental impacts. Therefore, we present a spatial distribution and environmental attributes dataset of China’s large-scale data centers in 2024. This study integrates Points of Interest from Amap with spatial features extracted from Sentinel-2 imagery and employs the Random Forest model to generate a spatial probability surface for large-scale data centers. The dataset contains the latitude and longitude of 1,005 large-scale data centers in China, urban environmental attributes (climate zone, elevation, annual average temperature, and precipitation), as well as model-generated probability surfaces and satellite image examples. This dataset can be used for estimating energy consumption of artificial intelligence infrastructure, planning and optimizing energy system, and evaluating and enhancing urban energy system resilience.
Yang et al. (Mon,) studied this question.