Published : 15/06/2026
Following on from the previous article, tokens, the new oil of the AI era, require a corresponding consumption of electricity and computing power for every single token generated. When it comes down to it, the race for AI development is ultimately a energy battle over computing power and electricity.
Encouragingly, China possesses its own unique advantages in both power supply and computing capabilities, placing it at the forefront of the global stage.
World's No.1 in electricity supply
Let's talk about electricity first.
Research data shows that electricity costs account for as much as 60 to 70% of the operating costs of large AI models. Therefore, some have described it as a few years ago, AI competed on chips, but now it competes on electricity metres.
The essence of competition among "token factories" is, in fact, who has more stable and cheaper electricity.
China's advantages in electricity need no elaboration. The data makes it clear:
In 2025, China's total social electricity consumption will exceed 10 trillion kilowatt-hours, twice that of the United States, and more than the combined total of the four economies of the European Union, Russia, India, and Japan.
China's power supply system is exceptionally stable, with large-scale blackouts being rare. Even during peak electricity usage and under extreme weather conditions, the power grid can generally maintain stable operation.
In addition, China has been strategically developing renewable and clean energy for many years, including hydropower, wind power, solar power, and nuclear power, and currently about 40% of electricity used comes from green power.
In terms of electricity tariffs, whether for domestic or industrial use, electricity prices in the Chinese Mainland are at a relatively low level, which is particularly evident when compared with developed countries in Europe and America.
The aforementioned factors are undoubtedly positive for China's large-scale, high-efficiency production of tokens.
American business magnate Elon Musk also agrees on this point, having repeatedly stated that the bottleneck for AI computing power has shifted from chips to electricity. And China's lead in total electricity output will be directly translated into an advantage in AI.
Read more: China's UHV project: The world-leading "Electricity Highway"
Read more: How does China become "champion" of clean energy?
2nd-largest computing power scale globally
Now let's look at computing power.
China's current computing power scale is firmly ranked second in the world.
Looking at the data, as of the end of March 2026, the country's total scale of intelligent computing power reached 1.88 million PFLOPS (FP16). To put it roughly, this is equivalent to the computing power of hundreds of millions of ordinary home computers running at their limit simultaneously.
In terms of rankings, according to the "Global Computing Index Evaluation Report", the United States has a computing power score of 77 and China 70, both belonging to the "frontrunner" tier; Japan, Germany, and the United Kingdom are in the "chaser" range, with scores between 40 and 55.
The rapid growth of computing power is inseparable from the country's strategic deployment. The "Eastern Data, Western Computing" project is a key move among them.
Simply put, "Eastern Data, Western Computing" is about channelling the massive amounts of data generated in the eastern regions to the western regions, where land and electricity resources are more abundant and cheaper, for computation.
Just like the "South-to-North Water Diversion" and "West-to-East Power Transmission" projects, it optimises resource allocation and improves efficiency at a national level.
Under the deployment of "Eastern Data, Western Computing", China has built eight national computing power hub nodes in eastern hubs such as Beijing-Tianjin-Hebei, the Yangtze River Delta, the Greater Bay Area, and Chengdu-Chongqing, as well as in western regions like Inner Mongolia, Guizhou, Gansu, and Ningxia.
Within these hubs, 10 data centre clusters have been planned and developed, serving as the "backbone" for token generation.
After computing power is produced, how to distribute it efficiently is equally important.
China is actively promoting the construction of a "computing power network", which is like a computing power version of the national power grid that connects data centres or token factories across the country to realise unified monitoring and intelligent scheduling.
In the future, whether they are large enterprises or small and medium-sized companies, they can directly purchase tokens from the computing power network on demand. This will make computing power universally accessible and as convenient to use as water and electricity.
In short, by ensuring the supply of computing power through "Eastern Data, Western Computing" and then optimising resource allocation with a computing power network, China is building an efficient, nationwide computing power system.
If China is to become a "world token factory", it must leverage its inherent advantages in electricity while seizing opportunities in the development of computing power. This is the foundation for becoming a powerful nation in AI.
However, development alone is not enough. The two need to be seamlessly synergised, and the next article will focus on "electricity-computing synergy".
Read more: What is "Eastern Data, Western Computing"?
Read more: What is "space computing" ?