Policy

China's LineShine Supercomputer Tops Global Rankings Without GPUs

The CPU-only system delivers 2,198 exaflops using entirely domestic hardware, signaling Beijing's progress despite US export restrictions.

Omega Editorial· June 28, 2026· 3 min read

China has reclaimed the title of world's fastest supercomputer with LineShine, a system that achieves unprecedented performance without relying on graphics processing units—the component backbone of most modern high-performance computing systems.

Installed at the National Supercomputing Center in Shenzhen, LineShine displaced the US-based El Capitan system to take first place in the TOP500 ranking, according to details first reported by WIRED. The system delivers 2,198 exaflops of computing power, exceeding El Capitan's capacity by more than 20 percent while consuming approximately 42.2 megawatts of power.

Why it matters

LineShine's achievement demonstrates that US export controls on advanced computing components have not halted China's supercomputing ambitions. Instead, restrictions appear to have accelerated domestic innovation in alternative architectures. The system's success using CPUs rather than GPUs—and entirely Chinese-designed hardware—suggests Beijing has developed viable pathways around Western technology dependencies, with implications for AI development, scientific research, and the broader technology competition between the two nations.

Built entirely with domestic technology

Unlike contemporary supercomputers that depend heavily on GPUs for parallel processing, LineShine uses only central processing units. The architecture is based on China's LingKun platform and incorporates roughly 45,000 LX2 processors. Each processor contains 304 cores operating at 1.55 GHz clock speeds.

The system's nodes connect through LingQi, a proprietary high-speed network designed to minimize latency during data exchange. LineShine runs on Kylin OS, a Linux-based operating system widely deployed across China's scientific and government computing infrastructure.

This marks China's return to the top of the TOP500 ranking after nearly a decade, with the country last holding first place in the mid-2010s. The TOP500 list, published twice yearly since 1993, evaluates supercomputers through standardized benchmarks measuring theoretical speed, real-world performance, and energy efficiency.

Response to escalating trade restrictions

The achievement arrives amid intensifying technology competition between Washington and Beijing. Both the Trump and Biden administrations imposed strict export controls on advanced computing components, software, and AI-related technologies aimed at limiting China's technological advancement. These restrictions have expanded during the current Trump administration, particularly targeting GPUs, advanced chips, and AI-enabling components.

China has responded with parallel restrictions on US technology imports and, as LineShine demonstrates, accelerated investment in indigenous computing architectures capable of competing with Western systems despite limited access to cutting-edge foreign components.

El Capitan, located in Livermore, California, had held the top ranking since 2024 before LineShine's benchmark results confirmed the new performance leader.

Details of LineShine's specifications and ranking were first reported by WIRED.

#supercomputing#china#semiconductors#high-performance computing#us-china tech competition#cpu architecture

This is an original analysis by the Omega editorial team. Source reporting: WIRED.

Want systems like this working for your business?

Book a Call

More in Policy

Policy· 3 min read

Meta AI Glasses Privacy Backlash Grows as Women Report Unwanted Recording

Social media users are speaking out after being filmed without consent by wearers of Meta's smart glasses, highlighting gaps in privacy protections.

Via AI Watch · Aug 12, 2026
Policy· 3 min read

CME Group to Launch AI Compute Power Futures Contracts

The exchange will partner with Silicon Data to create tradable benchmarks for GPU rental costs, starting with Nvidia's H100 and B200 chips.

Via AI Watch · Aug 11, 2026
Policy· 3 min read

Nvidia bets $500B on GPUs as infrastructure assets

Jensen Huang's financing model treats AI chips like real estate, but China and depreciation pose major risks to Wall Street backers.

Via AI Watch · Aug 11, 2026