On October 9, the Chinese government unveiled its ambitious goal to amplify the nation’s computing power by over 50% by 2025, which would take it to a total of 300 EFLOPS (one quintillion floating-point operations per second). This initiative reflects China’s strong emphasis on supercomputing and cutting-edge AI technologies.
China’s growing computational capabilities are a response to the competitive landscape, particularly against the US, in crucial technology sectors including semiconductors, supercomputing and AI. US export controls on chip-making equipment have added to this competition. As of August, China’s computing power reached 197 EFLOPS, positioning it as the second global leader behind the US, which was estimated to have reached 200 EFLOPS in 2022 and has been steadily growing ever since.
The government’s drive to surge its computational capabilities appears to be influenced by the soaring computational demands of AI training. In fact, insights from a Google blog post highlighted that top-tier AI models in the future will necessitate tens of EFLOPS of supercomputing power to ensure efficient training durations.
Going forward, the initiative aims to establish an expansive network of data centres nationwide, particularly focusing on western China’s computational infrastructure. Provinces like Guizhou, known for their vast terrains but fewer inhabitants, are witnessing a surge in massive data centre establishments, such as the partnership between Apple and a local entity in Guizhou.
Overall, China’s push to enhance computational strength offers substantial opportunities for tech companies and businesses relying on high-end computing. The expansion of data centres provides access to significant computational power, crucial for scaling operations in AI, data analytics, and other compute-intensive areas.