The Most Expensive Supercomputers Ever Built
Supercomputers have always been at the forefront of technological advancement, pushing the boundaries of what's possible in computation.
These colossal machines not only represent significant scientific achievements but also substantial financial investments.
Let's delve into some of the most expensive supercomputers ever constructed.
Table of Contents
El Capitan
El Capitan, hosted at the Lawrence Livermore National Laboratory in California, is a testament to the United States' commitment to maintaining a leading edge in computational power.
With a staggering investment of approximately $600 million, this exascale supercomputer became operational in 2024.
It boasts a peak performance of 2.746 exaFLOPS, making it the world's fastest supercomputer as of November 2024.
El Capitan's primary mission is to support the U.S. National Nuclear Security Administration's stockpile stewardship program, ensuring the safety and reliability of the nation's nuclear arsenal.
Its architecture comprises 43,808 AMD 4th Gen EPYC CPUs and an equal number of AMD Instinct MI300A GPUs, resulting in a combined total of over 11 million cores.
This massive computational capability is housed within 87 compute racks, occupying a floor space comparable to two tennis courts, and consumes about 30 megawatts of power.
For more details, visit the official Lawrence Livermore National Laboratory page on El Capitan.
Tianhe-2
China's Tianhe-2, also known as "Milky Way 2," was developed by the National University of Defense Technology and is located at the National Supercomputer Center in Guangzhou.
With an investment of approximately 2.4 billion Yuan (around $390 million), Tianhe-2 held the title of the world's fastest supercomputer from June 2013 to June 2016.
It achieved a performance of 33.86 petaFLOPS, utilizing a combination of Intel Xeon processors and custom-built Matrix-2000 coprocessors.
Tianhe-2 serves various applications, including simulation, analysis, and government security tasks.
Its development marked a significant milestone in China's supercomputing capabilities, showcasing the nation's rapid advancements in high-performance computing.
Learn more about Tianhe-2 on its Wikipedia page.
NVIDIA DGX GH200
In May 2023, NVIDIA announced the DGX GH200, a groundbreaking supercomputer designed to handle terabyte-class models for massive recommender systems, generative AI, and graph analytics.
This system connects 32 NVIDIA Hopper Superchips into a singular superchip, comprising 256 H100 GPUs and 32 Grace CPUs, offering 19.5TB of shared memory with linear scalability for giant AI models.
The DGX GH200 represents NVIDIA's commitment to pushing the boundaries of AI and high-performance computing, providing researchers and enterprises with unprecedented computational resources to tackle complex AI challenges.
For more information, visit NVIDIA's official page on the DGX GH200.
Fugaku
Developed jointly by RIKEN and Fujitsu, Japan's Fugaku supercomputer is installed at the RIKEN Center for Computational Science in Kobe.
With an estimated development cost of around $1 billion, Fugaku became fully operational in 2021 and has been utilized for a wide range of applications, including drug discovery, weather forecasting, and advanced material simulations.
It achieved a peak performance of over 1 exaFLOPS, making it one of the world's most powerful supercomputers at the time.
Fugaku's architecture is based on ARM processors, marking a significant shift from the traditional x86 architecture commonly used in supercomputers.
This design choice emphasizes energy efficiency and performance, aligning with global trends toward more sustainable computing solutions.
For more details, visit RIKEN's official page on Fugaku.
Frontier
Frontier, located at the Oak Ridge National Laboratory in Tennessee, USA, is another monumental achievement in the realm of supercomputing.
With a construction cost of around $600 million, Frontier became the world's first exascale supercomputer upon its deployment in 2022, achieving a peak performance exceeding 1.5 exaFLOPS.
It utilizes AMD EPYC processors and AMD Instinct GPUs, reflecting a growing trend in the supercomputing community toward heterogeneous architectures that combine CPUs and GPUs for optimized performance.
Frontier supports a wide array of scientific endeavors, from climate modeling to nuclear physics, providing researchers with the computational power necessary to tackle some of the most complex challenges facing humanity today.
Learn more about Frontier on the Oak Ridge National Laboratory's official page.
These supercomputers represent the pinnacle of human engineering and investment in the pursuit of knowledge and technological advancement.
They not only serve as critical tools for scientific discovery but also as symbols of national pride and technological prowess.
As we continue to push the limits of computation, the investments in these monumental machines underscore the importance placed on solving the complex problems of our time.
Keywords: supercomputers, exascale computing, high-performance computing, El Capitan, Fugaku