The global race for supercomputing supremacy has taken a new turn. China has reclaimed the top spot in the TOP500 ranking with its new LineShine system, which dethroned the US-based El Capitan with a power of 2.198 exaflops on the High Performance Linpack (HPL) benchmark. This milestone marks the first time a Chinese system has led the list since Sunway TaihuLight in 2017, demonstrating that China's domestic development strategy is bearing fruit.

A Giant Leap: Over 20% Advantage

LineShine, installed at the National Supercomputing Centre in Shenzhen and built by the Shenzhen Cloud Computing Center, has achieved 2.198 exaflops, surpassing the second-place El Capitan at the Lawrence Livermore National Laboratory by more than 20%, which stands at 1.809 exaflops. This performance makes LineShine the first system to officially break the 2 exaflops barrier, setting a new standard in high-performance computing.

The new ranking also reveals that there are now five exascale systems operational globally: LineShine, El Capitan, Frontier, Aurora, and Europe's JUPITER Booster. While US systems rely heavily on AMD and Intel accelerators, LineShine achieved the top spot with a CPU-only design, built around China's LingKun platform, the proprietary LingQi interconnect, and the Kylin operating system. This demonstrates that China's bet on technological sovereignty is paying off, as we have already seen in other areas such as post-quantum cryptography.

Top 10: Europe Gains Ground, Cloud Consolidates

The top 10 of the June 2026 TOP500 list has undergone significant changes. Italy has burst onto the scene: HPC7, owned by energy company Eni, debuts in sixth place with 571.5 petaflops, while HPC8 sits in eighth place with 477.9 petaflops. This makes Eni the only company with two systems in the top 10, reflecting the growing demand for computing capacity in sectors such as energy exploration.

Meanwhile, Finland's LUMI and Italy's Leonardo have exited the top 10, while Microsoft's Eagle remains in seventh place with 561.2 petaflops. Eagle is a clear example of how hyperscale cloud providers are gaining influence in high-performance computing, a phenomenon we have already analyzed in articles such as Qualcomm buys Modular: The end of Nvidia's monopoly in AI data centers?.

The overall performance of the TOP500 continues to grow at a breakneck pace. Combined computing power has reached 18.74 exaflops, up from 14.99 exaflops six months ago, an increase of 25%. Additionally, the adoption of accelerators has increased to 277 systems, up from 255 in the previous edition, indicating that the trend toward computational heterogeneity is consolidating.

Details of the 10 Most Powerful Systems

1. LineShine (China)
Location: National Supercomputing Centre in Shenzhen (NSCS).
HPL Score: 2.198 exaflops.
Model: LingKun Platform.
Architecture: LX2 processors with 304 cores at 1.55 GHz.
Total Cores: 13,789,440.
Interconnect: LingQi.

2. El Capitan (USA)
Location: Lawrence Livermore National Laboratory, California.
HPL Score: 1.809 exaflops.
Model: HPE Cray EX255a.
Architecture: AMD EPYC 4th Gen (24 cores, 1.8 GHz) + AMD Instinct MI300A.
Total Cores: 11,340,000.
Interconnect: Cray Slingshot-11.

3. Frontier (USA)
Location: Oak Ridge National Laboratory, Tennessee.
HPL Score: 1.353 exaflops.
Model: HPE Cray EX235a.
Architecture: AMD EPYC 3rd Gen (64 cores, 2 GHz) + AMD Instinct MI250X.
Total Cores: 9,066,176.
Interconnect: Cray Slingshot-11.

4. Aurora (USA)
Location: Argonne Leadership Computing Facility, Illinois.
HPL Score: 1.012 exaflops.
Model: HPE Cray EX – Intel Exascale Compute Blade.
Architecture: Intel Xeon CPU Max Series + Intel Data Center GPU Max Series.
Total Cores: 9,264,128.
Interconnect: Cray Slingshot-11.

5. JUPITER Booster (Germany)
Location: Jülich Supercomputing Centre.
HPL Score: 1.000 exaflops.
Model: BullSequana XH3000.
Architecture: GH Superchip 72C at 3 GHz.
Total Cores: 4,801,344.
Interconnect: NVIDIA InfiniBand NDR200.

6. HPC7 (Italy)
Location: Eni S.p.A.
HPL Score: 571.5 petaflops.
Model: HPE Cray EX255a.
Architecture: AMD EPYC 4th Gen (24 cores, 1.8 GHz) + AMD Instinct MI300A.
Total Cores: 3,461,472.
Interconnect: Cray Slingshot-11.

7. Eagle (USA)
Location: Microsoft Azure.
HPL Score: 561.2 petaflops.
Model: Microsoft NDv5.
Architecture: Intel Xeon Platinum 8480C (48 cores, 2 GHz) + NVIDIA H100.
Total Cores: 2,073,600.
Interconnect: NVIDIA InfiniBand NDR.

8. HPC6 (Italy)
Location: Eni S.p.A., Ferrera Erbognone.
HPL Score: 477.9 petaflops.
Model: HPE Cray EX235a.
Architecture: AMD EPYC 3rd Gen (64 cores, 2 GHz) + AMD Instinct MI250X.
Total Cores: 3,143,520.
Interconnect: Cray Slingshot-11.

9. Fugaku (Japan)
Location: RIKEN Center for Computational Science, Kobe.
HPL Score: 442.0 petaflops.
Model: Fugaku.
Architecture: Fujitsu A64FX (48 cores, 2.2 GHz).
Total Cores: 7,630,848.
Interconnect: Tofu Interconnect D.

10. Alps (Switzerland)
Location: Swiss National Supercomputing Centre (CSCS).
HPL Score: 434.9 petaflops.
Model: HPE Cray EX254n.
Architecture: NVIDIA Grace 72C (3.1 GHz) + NVIDIA GH200 Superchip.
Total Cores: 2,121,600.
Interconnect: Cray Slingshot-11.

Implications for Industry and Technology

LineShine's dominance is not only a technical achievement but also has profound geopolitical and commercial implications. China demonstrates that it can compete at the forefront of supercomputing without relying on Western technologies, reinforcing its self-sufficiency in a context of trade restrictions. For businesses and research centers, this means that the supply of computing capacity is diversifying, with options based on both accelerators and pure CPUs.

Furthermore, the presence of cloud systems like Microsoft's Eagle indicates that high-performance computing is no longer exclusive to government or academic data centers. Cloud providers are democratizing access to supercomputing, allowing smaller companies to perform complex simulations without multi-million dollar investments. This shift is also driving the need for runtime verification to ensure the reliability of codes running in these environments.

On the other hand, the growing computing power opens new possibilities in fields such as artificial intelligence, climate simulation, and drug discovery. However, it also poses challenges in terms of energy consumption and cooling. Systems like JUPITER Booster, which uses liquid cooling, point the way toward more sustainable supercomputing.

In summary, the arrival of LineShine not only changes the ranking but redefines the rules of the game. Supercomputing is moving toward a more diverse future, with multiple architectures and players, where innovation and technological sovereignty will be key. To stay up to date with these trends, don't miss our analysis on how code must be regenerated, not maintained in a world of extreme computing.

Original source: ComputerWorld. Analysis and adaptation by ForgeNEX.