China's Lingshen Supercomputer to Achieve 2 Exaflops

Key Points
- 1China announces Lingshen supercomputer with over 2 exaflops capacity.
- 2Introduction of CPU-only architecture shifts conventional supercomputer design.
- 3Indigenous technology initiative enhances AI sovereignty and reduces foreign reliance.
China's National Supercomputing Center in Shenzhen is set to construct the Lingshen supercomputer, which aims for sustained performance exceeding 2 exaflops. Unlike typical supercomputers, Lingshen will utilize only general-purpose CPUs with no GPUs, incorporating 47,000 Chinese processors housed in Huawei Kunpeng servers. This moves away from current supercomputing norms, illustrated by the high-performance models like the US's El Capitán, which employs GPU technology to achieve its benchmarks.
The implications of this project are significant for China's AI and technology landscape. Should Lingshen meet its performance estimates, it could represent a major leap in national compute capabilities without foreign components. This endeavor supports China's aim for increased data sovereignty and independence from semiconductor technologies of other nations. However, uncertainties remain as Lingshen is still a proposed project, not yet realized, raising questions about its actual operational effectiveness and industry impact.
Free Daily Briefing
Top AI intelligence stories delivered each morning.
Related Articles

Alibaba Releases Qwen3.6-27B for Local AI Coding

Data Centers Embrace AI Chips for Enhanced Performance

Lenovo Launches Powerful AI Workstation ThinkPad P16 Gen 3

OCP Members Advocate for DC Power in Data Centers
