AI Enables First National Census of Wind and Solar Power in China; Study by Peking University and Alibaba DAMO Academy Published in Nature

In a landmark study published in Nature on May 20, 2026, researchers from Peking University and Alibaba DAMO Academy have utilized artificial intelligence to create the first high-resolution, nationwide map of wind and solar power facilities in China. This breakthrough addresses the critical challenge of accurately inventorying the country's rapidly expanding renewable energy infrastructure, which is essential for optimizing grid integration and achieving China's carbon neutrality goals.

The team processed 7.56 terabytes of 0.5-meter resolution open-source satellite imagery covering all of China using a custom AI model developed by DAMO Academy. Overcoming challenges such as massive data volume and diverse terrain, the AI successfully located and identified 319,000 photovoltaic installations and 91,600 wind turbines across 1,915 counties. “This is the first time we have a large-scale, high-precision inventory of wind and solar facilities nationwide,” said Professor Liu Yu from Peking University’s School of Earth and Space Sciences. “It provides a bird’s-eye view of the national renewable energy landscape, offering a solid foundation for grid optimization, environmental assessment, and more.”

Leveraging this dataset, the research team conducted an in-depth analysis of “wind-solar complementarity”—the idea that wind and solar power can balance each other’s intermittency due to their different generation patterns. For instance, solar output peaks during the day, while wind generation is often more stable at night. The study found that integrating wind and solar across broader geographic scales significantly reduces the need for curtailment (the intentional reduction of output to avoid overloading the grid).

Specifically, the team analyzed renewable energy integration at four spatial scales: within provinces, between neighboring provinces, across broader regions, and nationwide. They discovered that as the coordination scale expands, the temporal complementarity between wind and solar generation and electricity demand becomes stronger, substantially improving utilization efficiency. Some cross-regional combinations showed higher theoretical complementarity than geographically closer pairs, highlighting the importance of large-scale spatial coordination.

Under conditions of high grid flexibility, the study estimates that nationwide inter-provincial coordination could unlock an additional 100 billion kilowatt-hours (100 TWh) of green electricity consumption—not through new generation capacity, but by reducing curtailment of existing wind and solar farms. This improved coordination also alleviates pressure on energy storage and grid regulation. The findings provide crucial theoretical guidance for policymakers designing inter-provincial power trading mechanisms and infrastructure investments to support large-scale renewable integration.

Yu Chaohui, an algorithm expert at Alibaba DAMO Academy, noted, “With the help of AI models, we have built a new data foundation for academia and industry. This will likely drive systematic research on wind and solar power planning, further supporting the construction of new power systems and accelerating the achievement of carbon peak and carbon neutrality goals.”

The study underscores the transformative potential of AI in renewable energy planning and offers a replicable framework for other countries seeking to optimize their clean energy transitions.