China aims to increase the country’s total computing power by more than 50% by 2025, according to a plan released by officials on Monday, as Beijing increases its focus on supercomputing and artificial intelligence innovations.

The plan comes amid growing competition between China and the US in a number of high-tech sectors from semiconductors and supercomputers to AI, including US export controls on chip-making equipment.

The plan issued by six departments in Beijing, including the Ministry of Industry and Information Technology (MIIT), targets China’s total computing power to reach 300 eFLOPS by 2025. EFLOPS is equivalent to one quintillion floating-point operations per second. Measures computer speed.

MIIT revealed in August that China’s computing power has reached 197 EFLOPS this year, exceeding 180 EFLOPS in 2022. The ministry said it ranked China second only to the United States, but did not elaborate on the scale of American computing power. ,

As AI training requires large amounts of computation, efforts to expand the supply of computing power are increasingly becoming a focus for Beijing.

According to a Google blog post last month, the world’s top-tier generative AI models would “require tens of thousands of eFLOPs of AI supercomputing to maintain training times of several weeks or less”.

According to the plan, China aims to build more data centers across the country to facilitate businesses’ access to computing power.

To meet the demands of the rapidly growing AI industry, Beijing also plans to improve computational infrastructure in western China.

Vast but sparsely populated provinces in China, such as southwestern Guizhou, have long been tasked with setting up massive data centers to power the country’s internet. For example, Apple has set up data centers in Guizhou with a local partner to serve its users in the country.

The second focus is to improve the speed and efficiency of computation networks. The plan states that transmission speeds between critical computing facilities should not have a latency of more than 5 milliseconds.