China: China seeking energy-efficient solutions to power AI computing infrastructure.

【Voice_over】 China's AI data centers are finding innovative solutions to one of the industry's biggest hurdles: the most efficient way to pay for and supply the required energy. In Inner Mongolia, a weather model fuses satellite and radar data for real-time monitoring, allowing energy to be dispatched, or stored, as needed. The manager says the system reduces power and infrastructure costs. 【Sound_bite】 Zheng Zihao, general manager, AI computing center, Envision Group: "Tianji predicts renewable fluctuations via weather. Tianshu then dispatches storage and charging based on generation curves." 【Voice_over】 Elsewhere in the region, the cool climate is a draw. One new center is using immersion liquid cooling, which has dropped its power usage effectiveness rating, or PUE, to around one. 【Sound_bite】 Ding Zhenlei, computing products director, UCloud: "Computing must follow electricity. We build training clusters in places like Ulanqab, where natural cooling and green power are abundant." 【Voice_over】 IEA projects global data center electricity use will nearly double by 2030. [Restrictions: No access Chinese mainland]