At the Qinghai clean energy and green computing scheduling center in Xining city, real-time data on the screen monitors power generation from different sources, including wind, solar and hydroelectric power, as well as the energy consumption metrics of computing centers.

Leveraging its unique geographical advantages, Xining, capital of Qinghai province in northwest China, is promoting the synergy between green electricity and intelligent computing.

In recent years, the rapid development of generative AI has led to a meteoric rise in applications utilizing large AI models, resulting in an explosive demand for computing power.

As of June 2024, China had more than 8.3 million operational computing racks, with a computing capacity reaching 246 EFLOPS. This substantial increase in computing capacity has been accompanied by a significant rise in energy consumption. According to the China Academy of Information and Communications Technology, data centers, including computing centers, accounted for about 1.6 percent of the total electricity consumption in China in 2023.

Qinghai province, situated on the Qinghai-Xizang Plateau, has natural conditions and resource advantages that make it a leader in clean energy and an ideal location for developing green computing.

Fan Kewei, general manager of the Qinghai clean energy and green computing scheduling center, explained that computing power usage is closely tied to specific application scenarios. Compared to traditional data storage and general computing, the rapidly growing intelligent computing power is more suitable for integration with green electricity.

"Intelligent computing is not always active. Tasks like large-scale rendering and data training are typically initiated only when needed, and can be done when there is an abundance of renewable energy and lower electricity prices," Fan said.

Currently, Qinghai's intelligent computing capacity has reached 1.23 EFLOPS, and the province has established the nation's first 100 percent clean energy traceable big data center. Furthermore, its large data centers maintain an average power usage effectiveness (PUE) below 1.2, placing them at the forefront nationally.

Digitalization and intelligence are critical in the construction of a new power system, with ample computing power poised to provide essential support.

The features of randomness, intermittency and volatility associated with renewable energy make balancing electricity supply and demand increasingly complex as more renewable sources connect to the grid. AI applications, represented by large models, can perform computational analysis on meteorological conditions, historical renewable energy generation data, and electricity demand, facilitating accurate predictions of renewable energy generation and consumption, thereby enhancing the safety and stability of the power system.

Using big data, cloud computing  and blockchain technologies, the province also developed the electricity-carbon calculation model to achieve a comprehensive view of carbon emissions in key industries, parks and enterprises, helping to realize the goal of carbon reduction and green development.