China’s big AI power play: How the world’s largest grid is giving Beijing a massive edge | Explained

(FILES) Workers assemble power distribution cabinets at the State Grid Xiaoshan Electric Power Xinmei Electric Company in Hangzhou in eastern China's Zhejiang province on October 19, 2020. (Photo by AFP)

As artificial intelligence reshapes global power structures, the United States may lead in cutting edge AI models and advanced computer chips, but China is advancing rapidly in an area that could redefine the race. It has built the largest power grid the world has ever seen, giving Beijing a major advantage in energy hungry technologies such as AI.

A report in The Wall Street Journal notes that between 2010 and 2024, China expanded its power generation more than the rest of the world combined. Last year alone, China produced more than twice the electricity generated by the United States. According to the WSJ report, this gives Chinese data centers access to electricity at less than half the price paid by many American facilities, sharply lowering the cost of running AI systems.

China’s push for power dominance is transforming remote regions such as Inner Mongolia. Once known for its vast open plains, the region is now filled with wind turbines and long stretches of transmission lines. Officials call it a new “cloud valley of the grasslands,” with more than 100 data centers already operating or under construction.

Morgan Stanley estimates that China will spend about 560 billion dollars on grid upgrades in the five years through 2030, a significant jump from previous investment cycles. Goldman Sachs predicts that by 2030 China will have nearly 400 gigawatts of excess power capacity. This is roughly three times the world’s expected data center power demand that year.

Meanwhile, American tech companies are expressing concern about the strain on the United States power system. Microsoft chief executive Satya Nadella, among others, has warned that running the massive number of AI chips now being purchased will require enormous energy. According to the Wall Street Journal, several companies want Washington to cut regulatory delays and support modernization of the national power grid.

Morgan Stanley forecasts that US data centers could face a major electricity shortfall within three years, a scenario that would pose a “daunting challenge” for America’s AI ambitions.

China’s cheaper energy has already helped domestic developers. AI firms such as DeepSeek have been able to build high quality models at lower costs than many US competitors. This advantage is tied to a national strategy that began in 2021 when Beijing launched the “East Data, West Computing” plan. The initiative directs western regions, which have abundant energy resources, to power AI and cloud computing demand rising in the more populated east.

Both China and the United States face the same core problem. AI development consumes vast amounts of electricity, and the demand keeps rising. Every chatbot query and every model training cycle requires substantial computing power. The challenge for both countries is predicting how much energy future AI systems will need.

By 2030, China’s data centers are expected to consume as much electricity each year as the entire nation of France. The United States power use will be even higher. According to the International Energy Agency, American data centers accounted for 45 percent of global data center power consumption last year, compared to China’s 25 percent.

China’s overall power generating capacity now stands at 3.75 terawatts, more than double the US total. Its expansion continues at a rapid pace with 34 nuclear reactors under construction and nearly 200 more planned or proposed, according to the World Nuclear Association. In Tibet, China is building what will become the world’s largest hydropower project, capable of producing three times the output of the Three Gorges Dam.

These developments show how China is reshaping the global AI competition through massive investments in electricity, which has become the most important resource for the next era of technology.