Supercomputers are the most powerful types of computers available at the moment. Unlike the regular computers that perform normal operations such as surfing the internet, using office applications and watching videos, supercomputers analyze vast amounts of information while carrying out billions of operations simultaneously. Supercomputers are used by governments, universities, research institutions and corporations in performing operations that would take regular computers years to perform.
The following list ranks the world’s top 10 supercomputers according to the November 2025 TOP500 ranking, which measures performance using the High Performance Linpack (HPL) benchmark.
1. El Capitan – United States
El Capitan currently holds the number-one position in the world. Located at Lawrence Livermore National Laboratory (LLNL) in California, the system achieved 1.809 exaflops on the HPL benchmark. It contains approximately 11.34 million cores and has a theoretical peak performance of 2.821 exaflops.
El Capitan uses an HPE Cray EX255a architecture, AMD fourth-generation EPYC processors and AMD Instinct MI300A accelerators. Its Slingshot-11 interconnect allows the enormous number of computing components to communicate efficiently.
One of its major purposes involves national security research and advanced scientific computing. Its capabilities also support simulations and modelling that require enormous computational resources.
El Capitan also performs exceptionally well on the HPCG benchmark, where it achieved 17.41 petaflops, making it the leading system on that benchmark as well.
2. Frontier – United States
Frontier, operated by Oak Ridge National Laboratory in Tennessee, ranks second. It achieved 1.353 exaflops on HPL and contains more than 9 million cores.
Frontier uses an HPE Cray EX235a architecture with AMD EPYC CPUs and AMD Instinct MI250X accelerators. It became the world’s first publicly recognised exascale supercomputer before El Capitan took the number-one position.
Researchers use Frontier for areas such as climate science, materials research, nuclear science, biology and artificial intelligence. Its position demonstrates how quickly supercomputing has progressed: a machine that once represented the absolute peak of computational performance now sits behind several newer systems.
3. Aurora – United States
Aurora occupies third place with an HPL performance of 1.012 exaflops. The system operates at Argonne National Laboratory in Illinois and represents another major US investment in exascale computing.
Aurora uses Intel Xeon CPU Max processors and Intel Data Center GPU Max accelerators within an HPE Cray EX architecture. It has more than 9.26 million cores and a theoretical peak performance of almost 1.98 exaflops.
Scientists can use Aurora for large-scale simulations, scientific discovery and AI-related research. Its architecture also demonstrates the growing importance of GPUs and specialised accelerators in modern supercomputing.
4. JUPITER Booster – Germany
JUPITER Booster has changed the geographical distribution of exascale computing. The system, located at the Jülich Supercomputing Centre in Germany, achieved exactly 1.000 exaflops, making it the fourth exascale system in the TOP500 ranking and the first exascale system outside the United States.
The system uses a BullSequana XH3000 platform with NVIDIA GH200 superchips and NVIDIA InfiniBand networking. It has approximately 4.8 million cores and a theoretical peak performance of 1.226 exaflops.
JUPITER represents a significant development for European high-performance computing. Its arrival demonstrates that exascale capability has expanded beyond the United States.
5. Eagle – United States
Eagle, operated by Microsoft Azure, ranks fifth with an HPL performance of 561.2 petaflops. Unlike several government and research-laboratory systems above it, Eagle operates within Microsoft’s cloud infrastructure.
The system uses NVIDIA H100 GPUs, Intel Xeon Platinum processors and NVIDIA InfiniBand networking. It has approximately 2.07 million cores.
Eagle highlights an important development in high-performance computing: supercomputing capabilities increasingly connect with cloud infrastructure. This approach can make very powerful computing resources available for workloads involving artificial intelligence, scientific research and large-scale data processing.
6. HPC6 – Italy
Italy’s HPC6 occupies sixth position with 477.9 petaflops of HPL performance. Eni operates the system at its computing centre in Ferrera Erbognone, Italy.
HPC6 uses the HPE Cray EX235a architecture, AMD EPYC processors and AMD Instinct MI250X accelerators. It has approximately 3.14 million cores and consumes around 8.46 MW according to the TOP500 data.
The system supports computationally demanding energy-sector research. Its presence in the top 10 also shows that major industrial organisations increasingly invest in supercomputing rather than leaving the technology exclusively to governments and universities.
7. Fugaku – Japan
Japan’s Fugaku ranks seventh with 442.01 petaflops. Located at the RIKEN Center for Computational Science in Kobe, Fugaku has remained one of the world’s most influential supercomputers since it entered the TOP500 ranking in 2020.
Fugaku uses Fujitsu’s A64FX processors and Tofu interconnect technology. It contains approximately 7.63 million cores.
Although newer systems have surpassed Fugaku in HPL performance, its capabilities remain impressive. It also performs strongly on HPCG, where it ranks second with approximately 16 petaflops.
This distinction matters because HPL does not represent every type of computational workload. A supercomputer can rank lower on one benchmark while delivering exceptional performance for particular scientific applications.
8. Alps – Switzerland
Alps, operated by the Swiss National Supercomputing Centre (CSCS), ranks eighth with 434.9 petaflops.
The system uses an HPE Cray EX254n architecture, NVIDIA Grace processors and NVIDIA GH200 superchips. It contains approximately 2.12 million cores.
Alps provides computing resources for scientific research in Switzerland and Europe. Its architecture reflects the growing use of tightly integrated CPU-GPU systems, which can handle traditional simulations as well as increasingly demanding AI workloads.
9. LUMI – Finland
Finland’s LUMI ranks ninth, achieving 379.7 petaflops on HPL. The system operates at the CSC data centre in Kajaani and forms part of Europe’s EuroHPC infrastructure.
LUMI uses an HPE Cray EX235a platform with AMD EPYC CPUs and AMD Instinct MI250X accelerators. It has approximately 2.75 million cores.
LUMI serves researchers across Europe and supports work involving climate modelling, artificial intelligence, materials science, health research and other computational disciplines. Its role demonstrates how countries can pool resources to create large-scale scientific computing infrastructure.
10. Leonardo – Italy
Leonardo, located at the CINECA supercomputing centre in Italy, takes tenth place with 241.2 petaflops.
The system uses a BullSequana XH2000 architecture, Intel Xeon Platinum processors and NVIDIA A100 GPUs. It contains more than 1.82 million cores.
Leonardo forms part of Europe’s broader EuroHPC strategy and provides computing resources for scientific and technological research. Its capabilities support areas such as weather and climate modelling, materials research, engineering and artificial intelligence.
The Future of Supercomputing
The next stage of supercomputing will not simply involve building machines with more processors. Designers must also address energy consumption, cooling, data movement, memory capacity and software efficiency.
The appearance of four exascale systems in the November 2025 TOP500 list shows how quickly the field has progressed. Europe has also entered the exascale era through JUPITER Booster, while Japan and Switzerland continue to operate highly capable systems with distinctive architectures.
Future systems will likely place even greater emphasis on AI acceleration, heterogeneous computing and energy efficiency. The challenge will be to increase computational performance without allowing power consumption and operating costs to rise at the same rate.
Conclusion
The world’s best supercomputers show the advancements made in computing technology. El Capitan tops the list of November 2025 TOP500 at 1.809 exaflops, followed by Frontier and Aurora. Also, Germany has produced its first exascale computer called JUPITER Booster.
These machines have applications that go well beyond racing for speed. They are useful in studying climate change, material science, engineering design, biological systems, and artificial intelligence. These developments also illustrate an interesting trend: in the realm of supercomputing today, teamwork among CPUs, GPUs, fast networks, and advanced software is increasingly necessary.
As computational demands continue to increase, the question is no longer simply how fast a computer can calculate. The more significant question is what scientific and technological problems humanity can solve when that computational power becomes available.
