What Is the Fastest Supercomputer in the World in 2025?

Why Trust Techopedia

Supercomputers are the peak of computing technology, capable of processing vast amounts of data and performing complex calculations rapidly. These high-performance machines are used for various applications, from scientific research to national security simulations, climate modeling, industrial design, and artificial intelligence (AI).

The escalating demand for massive volumes of cloud-based computing power is bringing supercomputing technologies into hyperscale data centers. And the rapid adoption of AI will likely see more supercomputers built worldwide.

In January 2025, the Lawrence Livermore National Laboratory (LLNL) in California, U.S., dedicated El Capitan, the fastest computer in the world. It is the first exascale computing system built for the National Nuclear Security Administration (NNSA). El Capitan’s powerful AI training and inference capabilities can support complex simulations for national security and unclassified research initiatives.

As more countries build large-scale supercomputers with the capacity to support AI development, it raises the question, what is the fastest supercomputer in the world in 2025?

Key Takeaways

  • Supercomputers are high-performance systems with exceptional processing power to perform complex scientific and engineering calculations and simulations.
  • The US is now home to the world’s four fastest supercomputers and five of the top 10, having achieved exascale speeds.
  • Fugaku in Japan was previously the world’s fastest, holding the top spot from 2020 to 2022.
  • Four of the 10 fastest supercomputers are in Europe, which is set to introduce its first exascale system with the Jupiter in Germany.
  • Advancements in AI, quantum computing, and exascale systems are expanding the possibilities of scientific research and data processing.

Defining Supercomputers

What is a supercomputer? Before delving into the numbers, it’s important to understand what sets a supercomputer apart from a high-performance computer.

Supercomputers are typically characterized by their exceptional processing power. Their speed is measured in floating-point operations per second (FLOPs), which refers to how many arithmetic calculations they can perform.

While there is no strict threshold that separates a supercomputer from a regular high-performance computing (HPC) system, supercomputers are generally capable of performing at speeds measured in teraFLOPs (trillions) and petaFLOPs (quadrillions). An exascale supercomputer like El Capitan can perform more than two quintillion (10¹⁸) FLOPs.

Governments, research institutions, and industrial corporations use supercomputers for various applications. They range in size from super-large scale to small-scale devices. There are so many supercomputers today that it is difficult to pinpoint the exact number. In addition, some systems are not known publicly for national security, geopolitical or competitive reasons.

However, the TOP500 project ranks the world’s 500 most powerful supercomputers biannually based on their performance on the Linpack benchmark, which measures the speed at which a computer can solve a system of linear equations. This provides a standardized way to compare supercomputer capabilities.

Top 10 Fastest Supercomputers in the World in 2025

The El Capitan supercomputer heads up the list as the third system to achieve exascale computing after the Frontier and Aurora systems, which are also installed at Department of Energy (DOE) laboratories in the U.S. New systems have made it to the latest Top 10.

Rank Country System Manufacturer Hardware Supplier Cores Max Performance (PFlop/s) Power (kW)
1 USA El Capitan HPE AMD 11,039,616 1,742.00 29,581
2 USA Frontier HPE AMD 9,066,176 1,353.00 24,607
3 USA Aurora Intel Intel 9,264,128 1,012.00 38,698
4 USA Eagle Microsoft NVIDIA 2,073,600 561.2
5 Italy HPC6 HPE AMD 3,143,520 477.9 8,461
6 Japan Fugaku Fujitsu Fujitsu 7,630,848 442.01 29,899
7 Switzerland Alps HPE NVIDIA 2,121,600 434.9 7,124
8 Finland LUMI HPE AMD 2,752,704 379.7 7,107
9 Italy Leonardo EVIDEN Intel/NVIDIA 1,824,768 241.2 7,494
10 USA Tuolumne HPE AMD 1,161,216 208.1 3,387

Source: The TOP500 list

1. El Capitan

El Capitan has displaced Frontier at the Oak Ridge National Laboratory (ORNL) as the world’s fastest supercomputer, achieving 1.742 exaFLOPs. It has 11,039,616 combined central processing unit (CPU) and graphics processing unit (GPU) cores. It is based on AMD processors and accelerators.

El Capitan uses 100% fanless direct liquid-cooling technology from Hewlett Packard Enterprise (HPE) to achieve an energy efficiency of 58.89 GigaFLOPs/watt, placing the supercomputer 18th on the GREEN500 list ranking supercomputers by their energy efficiency.

Application area:
El Capitan will be used from March to manage the US nuclear weapons stockpile and support complex high-fidelity modeling and simulation, as well as AI and machine learning for national security, materials discovery, and inertial confinement fusion.

2. Frontier

The world’s first exascale system, the Frontier at ORNL in Tennessee, U.S., had previously topped the list as the most powerful supercomputer since 2022. It clocked 1.1 exaFLOPs Rmax in May 2022, which has since increased to 1.353 exaFLOPs. Its CPU and GPU cores can perform double-precision operations at the same speed as single-precision.

When it was released, Frontier topped the Green500 list, measuring 62.68 gigaFLOPs/watt. However, more efficient systems have since surpassed it, and it currently ranks 11th.

Application area:
Frontier’s massive capacity is advancing scientific research in fields from medicine to material science, climate change and deep space. In November 2024, researchers at the DoE’s Argonne National Laboratory used the supercomputer to run the largest astrophysical simulation of the universe ever conducted.

3. Aurora

The Aurora system at Argonne Leadership Computing Facility in Illinois, U.S. is the third largest exascale system, achieving 1.012 exaFLOPs. The Aurora was built by Intel based on the HPE Cray EX – Intel Exascale Compute blade.

Researchers use Aurora to train large language models (LLMs) for science. The AuroraGPT project, for instance, is a science-focused foundation model trained on inputs across scientific domains, including biology and chemistry, to help researchers create new AI tools.

Application area:
Scientists are working to develop high-fidelity models of complex systems, such as the human circulatory system, nuclear reactors and supernovae, to gain new insights into their behavior. The supercomputer’s capacity to process massive datasets enables scientists to analyze the vast data streams from research facilities such as Argonne’s Advanced Photon Source (APS) and CERN’s Large Hadron Collider.

4. Eagle

The Eagle system is installed on the Microsoft Azure Cloud in the U.S. Its score of 561.2 petaFLOPs of computing power makes it the highest-ranked cloud-based system.

Launched in August 2023, Eagle contains 1.12 million computing cores, including 14,400 networked Nvidia H100 GPUs and Intel Xeon Platinum 8480C processors. It is the fastest system using the H100.

Application area:
The system is designed to provide access to LLMs like OpenAI’s GPT-4 hosted on the Azure cloud. It is readily accessible to users through the cloud infrastructure for various high-performance computing (HPC) tasks, allowing anyone with access to Azure to utilize its immense processing power.

5. HPC6

HPC6 is a supercomputing cluster at Italian energy company Eni’s Green Data Center Eni in Ferrera Erbognone, Italy.

HPC6 was launched in November 2024, increasing the center’s performance from 70 to 606 million billion mathematical operations per second, equivalent to a peak of 606 petaFLOPs and 477 petaFLOPs sustained. It has the same architecture as the Frontier and is the fastest supercomputer in Europe.

In addition to using a direct liquid cooling system to dissipate 96% of the heat generated, the Green Data Center is powered by a 1MW solar photovoltaic plant to reduce its carbon footprint.

Application area:
HPC6 is designed to help Eni implement its decarbonization strategy and address the challenges of the transition to clean energy. For instance, the supercomputer will be used to optimize the operations of industrial facilities, improve the accuracy of geological and fluid dynamic studies for carbon dioxide (CO2) storage, and develop new high-performance batteries.

6. Fugaku

The Fugaku has been jointly developed by Fujitsu and the RIKEN Center for Computational Science (R-CCS) since 2014, starting trial operation in 2020 before its completion in March 2021.

With a capacity of 442.01 petaFLOPs, it was the fastest supercomputer in the world from the June 2020 TOP500 list until it was surpassed by the Frontier supercomputer in 2022. It is the first ARM-based supercomputer to become the world’s fastest. Fugaku uses more than 150,000 Fujitsu A64FX ARM-based processors.

Application area:
Fugaku is used in disaster prevention and mitigation to predict extreme weather events and simulate the impact of earthquakes and tsunamis. Fugaku is also used in research for life sciences, energy, manufacturing, basic science, and socioeconomic applications.

Virtual Fugaku on AWS

Fujitsu and RIKEN have collaborated with Amazon Web Services (AWS) to replicate the supercomputer on the cloud. Virtual Fugaku provides researchers with an alternative way to run software developed for the supercomputer on the cloud, keeping their work confidential.

Proofs of concept (POCs) in 2023 found the Virtual Fugaku on AWS to be twice as fast as its on-premises counterpart in complex calculations and simulations.

Based on the technology Fugaku has made possible, Fujitsu is developing its FUJITSU-MONAKA ARM-based CPU for launch in 2027.

The company is working with Super Micro Computer to develop a high-performance, energy-saving AI computing platform and with AMD to improve the processing capacity of large-scale AI workloads.

7. Alps

The Alps supercomputer is part of the Swiss National Supercomputing Centre (CSCS), which provides computing services for the scientific community.

The supercomputer’s main location is Lugano, Switzerland, but it is distributed over several different sites to provide geographical redundancy and access to large amounts of data stored in different locations.

Aside from the CSCS, Alps is located at the Swiss Federal Technology Institute (EPFL) in Lausanne, Switzerland, the Paul Scherrer Institute (PSI) in Villingen-Schwenningen, Germany, for data archiving, and the ECMWF data center in Bologna, Italy, for access to meteorological data.

The HPE Cray supercomputer achieves a sustained peak performance of 434.90 petaFLOPs. Construction started in 2020, and the supercomputer was officially inaugurated in September 2024.

Application area:
Alps enables the creation of versatile clusters (vClusters) that can be customized to the needs of specific users, such as MeteoSwiss for its numerical weather forecasts, as well as machine learning and AI applications.

8. LUMI

The Large Unified Modern Infrastructure (LUMI) supercomputer is a petascale system located at the IT Centre for Science (CSC) data center in Kajaani, Finland.

LUMI is one of three pre-exascale systems financed by the European High-Performance Computing Joint Undertaking (EuroHPC JU), which pools European resources to develop exascale supercomputers. The consortium countries are Finland, Belgium, the Czech Republic, Denmark, Estonia, Iceland, the Netherlands, Norway, Poland, Sweden, and Switzerland.

LUMI is an HPE Cray EX-based system using AMD CPUs and GPUs to achieve 379.70 petaFLOPs. Most of the computing power comes from the GPU cores to optimize cost and performance.

LUMI became the fastest supercomputer in Europe in January 2023 until the launch of Alps. It is completely powered by hydroelectricity and the heat it generates is captured to heat buildings locally. The system ranks 25th on the Green500 list.

Application area:
LUMI can be used for climate change modeling, drug and vaccine research, analyzing data to understand diseases and hereditary illnesses, artificial intelligence (AI) purposes (training deep neural networks and LLMs), cybersecurity solutions, genome sequencing, social sciences, and more.

9. Leonardo

Leonardo is hosted and managed by Cineca in a data center in Bologna, Italy, for the EuroHPC JU. It was inaugurated in November 2022 with a performance of 241.20 petaFLOPs, which at the time made it the fourth fastest supercomputer in the world.

The system comprises an Atos BullSequana XH2000 computer and around 14,000 Nvidia Ampere GPUs. Leonardo was constructed as three separate modules: a booster module housing the GPUs, a “data-centric module” comprising CPUs, and a “front-end & service module.”

EuroHPC JU and the Italian government are financing an upgrade, known as the Leonardo Improved Supercomputing Architecture (LISA) to be installed by July 2025.

Application area:
Leonardo is used to mitigate and manage the risks related to extreme weather events and natural disasters, as well as pandemics and epidemics.

10. Tuolumne

Commissioned in 2024, Tuolumne runs at the LLNL as a sibling to El Capitan. It has the same HPE Cray-based architecture but a fraction of its speed at 208.10 petaFLOPs. That is still enough to make it the world’s 10-largest supercomputer.

It more than doubles the speed of the Sierra supercomputer at LLNL, which preceded El Capitan and Tuolumne. Sierra can run at 125 petaFLOPs and peaked as the world’s second-fastest supercomputer in 2018.

Application area:
Tuolumne is an unclassified system, which allows it to be used for scientific research, from engineering simulations to climate and earthquake predictions, energy sustainability, and computational biology.

The Future of Supercomputing

Europe will soon join the U.S. as home to an exascale supercomputer with the Jupiter system currently under construction in Germany. Work started in 2024 on the system, which uses the hybrid cluster approach seen in the Leonardo supercomputer.

The Jupiter Exascale Development Instrument (JEDI) precursor system has the same equipment as the subsequent Jupiter booster module. The first module, which was installed in April 2024, is capable of 72 billion FLOPs per watt.

JEDI is one of the first supercomputing systems in the world to use NVIDIA’s GH200 Grace Hopper Superchip, which combines the Hopper GPU and Grace CPU on a single module.

As more exascale systems emerge, supercomputing will come up against technical limits, as processor speeds cannot rise much further without producing more heat and semiconductors cannot be manufactured much smaller.

ORNL is running a $23 million research program, New Frontiers, to accelerate the development of technologies that will be needed for post-exascale computing in 2029 and beyond.

Ceren Susut, associate director of DOE’s Office of Science for Advanced Scientific Computing Research, said:

“There is a growing consensus that urgent action is needed to address an array of bottlenecks in advanced computing, including energy efficiency, advanced memory, interconnects, and programmability to maintain economic leadership and national security.”

ORNL’s Christopher Zimmer, project director for New Frontiers, added:

“Current technology trends threaten to have a disruptive and costly impact on the development of DOE applications and potentially a negative impact on the productivity of DOE scientists.”

Advancing high-performance computing (HPC) architectures will require “numerous significant modifications” to the tools and techniques used in today’s systems, according to the lab.

The Bottom Line

Supercomputers bring more and more computational power, enabling groundbreaking research in fields like climate modeling, life sciences, engineering simulations and AI. The world’s fastest supercomputers operate at exascale speeds, performing more than a quintillion calculations per second.

As HPC technology advances, quantum computing and AI-driven optimizations may redefine performance benchmarks. Developers will also need to address the bottlenecks that could constrain the next generation of supercomputers by the end of the decade.

FAQs

Which is the fastest supercomputer in the world?

Is Google’s quantum computer 158 million times faster than the world’s fastest supercomputer?

Is Fugaku the world’s fastest supercomputer?

How fast are NASA supercomputers?

Related Reading

Related Terms

Advertisements
Nicole Willing
Technology Journalist
Nicole Willing
Technology Journalist

Nicole is a professional journalist with 20 years of experience in writing and editing. Her expertise spans both the tech and financial industries. She has developed expertise in covering commodity, equity, and cryptocurrency markets, as well as the latest trends across the technology sector, from semiconductors to electric vehicles. She holds a degree in Journalism from City University, London. Having embraced the digital nomad lifestyle, she can usually be found on the beach brushing sand out of her keyboard in between snorkeling trips.

Advertisements