The development of artificial intelligence has driven huge electricity demand, with the world competing to buy Chinese transformers.

Mars Financial News, April 13 — Recently, in many parts of the country, transformer manufacturing companies have been operating at full capacity, with overseas orders continuously overflowing, and some companies’ production schedules have been booked until 2027. Data shows that the power consumption load of a super-large AI data center has exceeded 1 gigawatt (1 gigawatt equals 1 billion watts), equivalent to the peak summer electricity demand of a medium-sized city. Additionally, AI large models are accelerating from the “training” phase toward the “inference” phase, which means their electricity consumption is shifting from one-time investment to continuous consumption. Professor Ding Zhaohui from the School of Electrical and Electronic Engineering at North China Electric Power University explained, “In the past, training a model was a one-time process. Now, various industries are using large models, so the electricity consumption naturally increases. The energy consumption during the ‘inference’ phase of large models is becoming more and more significant, and the demand for electricity from AI data centers is also growing.” (Economic Daily)