According to a plan announced by authorities on Monday, China intends to increase the country's aggregate computing capacity by more than 50% by 2025, as Beijing increases its focus on supercomputing and artificial intelligence developments.
The idea comes amid increased competition between China and the United States in numerous high-tech areas, including semiconductors and supercomputers, as well as U.S. export limitations on chipmaking equipment.
The strategy, announced by six Beijing departments including the Ministry of Industry and Information Technology (MIIT), has set a target of 300 EFLOPS for China's entire computing power by 2025. A computer's speed is measured in EFLOPS, which is equal to one quintillion floating-point operations per second.
According to the MIIT, China's computing power has reached 197 EFLOPS this year, up from 180 EFLOPS in 2022. The ministry stated that China is second only to the United States in terms of computing power, but did not clarify on the scale of computing power referred to.
Because AI training necessitates a considerable quantity of processing, Beijing is increasingly focusing on boosting the supply of computer power.
According to a blog post published by Google last month, top-tier generative AI models will "require tens of EFLOPs of AI supercomputing to maintain training times of several weeks or less."
China intends to establish more data centres across the country in order to improve enterprises' access to computing power resources, according to the plan.
Beijing also intends to improve computing infrastructure in western China in order to fulfil the demands of the fast developing AI industry.
China's vast but sparsely inhabited provinces, such as southwestern Guizhou, have long been charged with building large data centres to fuel the country's internet. Apple, for example, has established up data centres in Guizhou with a local partner to serve the country's users.
Another priority is to increase the compute network's speed and efficiency. According to the strategy, transmission speeds between vital computer facilities must not exceed 5 milliseconds.
(Source:www.theprint.in)
The idea comes amid increased competition between China and the United States in numerous high-tech areas, including semiconductors and supercomputers, as well as U.S. export limitations on chipmaking equipment.
The strategy, announced by six Beijing departments including the Ministry of Industry and Information Technology (MIIT), has set a target of 300 EFLOPS for China's entire computing power by 2025. A computer's speed is measured in EFLOPS, which is equal to one quintillion floating-point operations per second.
According to the MIIT, China's computing power has reached 197 EFLOPS this year, up from 180 EFLOPS in 2022. The ministry stated that China is second only to the United States in terms of computing power, but did not clarify on the scale of computing power referred to.
Because AI training necessitates a considerable quantity of processing, Beijing is increasingly focusing on boosting the supply of computer power.
According to a blog post published by Google last month, top-tier generative AI models will "require tens of EFLOPs of AI supercomputing to maintain training times of several weeks or less."
China intends to establish more data centres across the country in order to improve enterprises' access to computing power resources, according to the plan.
Beijing also intends to improve computing infrastructure in western China in order to fulfil the demands of the fast developing AI industry.
China's vast but sparsely inhabited provinces, such as southwestern Guizhou, have long been charged with building large data centres to fuel the country's internet. Apple, for example, has established up data centres in Guizhou with a local partner to serve the country's users.
Another priority is to increase the compute network's speed and efficiency. According to the strategy, transmission speeds between vital computer facilities must not exceed 5 milliseconds.
(Source:www.theprint.in)
