China has revealed its ambitious plans to significantly expand its computing power by over 50% within the next two years. This initiative aligns with China’s aspirations to dominate the global artificial intelligence (AI) race. The need for extensive computation in AI training has made bolstering computing resources a priority for Beijing. These efforts are further fueled by intense competition with the United States in various high-tech sectors, including semiconductors, supercomputers, and AI.
The drive for more computing power
The escalating requirements of the AI industry have propelled China to focus on enhancing its computing capabilities. AI training involves substantial computation, and securing a robust supply of computing power is vital for further progress.
This endeavor also unfolds amid heightened rivalry between China and the United States, encompassing areas such as semiconductor production and AI. The U.S. has imposed export controls on chipmaking equipment, intensifying the competition.
China’s ambitious computing power target
To fortify its position in the AI race, China has set an ambitious goal to achieve an aggregate computing power of 300 EFLOPS by 2025. EFLOPS, equivalent to one quintillion floating-point operations per second, serves as a measure of a computer’s speed.
The Ministry of Industry and Information Technology (MIIT) disclosed in August that China’s computing power had reached 197 EFLOPS in the current year, up from 180 EFLOPS in 2022. While the MIIT positioned China as second to the United States in terms of computing power, specific details about the scale of the U.S.’s computing power were not provided.
Google noted in a recent blog post that top-tier generative AI models would necessitate tens of EFLOPs of AI supercomputing to maintain training times within a few weeks or less.
Infrastructure improvements in Western China
China’s rapid development of AI technology requires not only increased computing power but also enhanced computational infrastructure. To address these needs, Beijing is planning improvements in western China, aiming to facilitate the burgeoning AI industry.
Less densely populated provinces like southwestern Guizhou have been designated to host massive data centers to support China’s internet infrastructure. Tech giants like Apple have established data centers in Guizhou, in partnership with local companies, to serve their user base within the country.
Expanding data centers nationwide
The MIIT’s strategy includes the construction of additional data centers across China, allowing businesses better access to computing resources. These data centers play a pivotal role in supporting AI development and other data-intensive activities.
In addition to expanding computing infrastructure, China aims to enhance the speed and efficiency of its computation network. One key aspect emphasized in the plan is minimizing latency. Critical computing facilities are required to maintain transmission speeds that ensure a latency of no more than 5 milliseconds.
China’s comprehensive plan to augment its computing power by more than 50% within the next two years underscores the country’s determination to lead the global AI race. The expansion of computing resources is a response to the surging demand for AI technology, which relies heavily on extensive computation.
The competition between China and the United States in high-tech domains, coupled with the increasing role of AI in various sectors, has elevated the significance of computing power. By setting ambitious targets and improving computational infrastructure, China seeks to position itself as a major player in shaping the future of AI and high-performance computing.
