Fastest Supercomputers by Performance
The fastest supercomputer in public rankings is Frontier, operated by Oak Ridge National Laboratory in the United States, which sustained over 1.1 exaflops on the LINPACK benchmark. The system uses AMD EPYC processors and Instinct accelerators and is designed for open science, climate modeling, and national security workloads. Frontier is maintained by Hewlett Packard Enterprise and the U.S. Department of Energy, and its performance figures are reported through the TOP500 project and Oak Ridge leadership pages TOP500 June 2024 list.
El Capitan at Lawrence Livermore National Laboratory is another leading system, delivering over 1.7 exaflops on the HPL benchmark and optimized for nuclear stockpile stewardship and weapons simulation. It is built by HPE Cray and uses AMD processors and accelerators, with the U.S. Department of Energy and National Nuclear Security Administration funding the deployment. El Capitan is part of a broader trend of exascale computing in the United States, where multiple facilities now operate at or above one quintillion floating-point operations per second Lawrence Livermore National Laboratory El Capitan page.
Largest and Most Expensive Computer Systems
The largest supercomputing systems by physical scale and cost include Frontier, Aurora at Argonne National Laboratory, and El Capitan, each costing hundreds of millions of dollars in hardware, integration, and facility upgrades. Aurora, delivered by Intel and HPE Cray, uses Intel Xeon Max Series processors and Intel Data Center GPU Max Series accelerators, and it targets materials science, cosmology, and medicine workloads at Argonne National Laboratory Argonne Aurora system page.
Cost Drivers and Funding Models
Cost drivers for these systems include custom cooling infrastructure, high-bandwidth interconnects, specialized power delivery, and long-term software and operations staffing. Most large-scale systems are funded by national governments, with the U.S. Department of Energy, the European Commission, and national research agencies allocating capital through multi-year contracts. The price tags often exceed one billion dollars when including facility construction, networking, and software ecosystem development, and procurement is typically tied to national competitiveness in science and industry Forbes TOP500 coverage.
Companies, Dates, and Technical Specifications
Key vendors building the fastest and biggest systems include Hewlett Packard Enterprise, Dell Technologies, Lenovo, and Cray, with AMD, Intel, and Nvidia supplying processors and accelerators. Frontier uses HPE Cray EX cabinets and Slingshot interconnect, while El Capitan relies on HPE Cray Shasta architecture and Slingshot-11 networking, and Aurora uses HPE Cray Ex supercomputing hardware with Intel software ecosystems HPE Supercomputing solutions.
Performance Metrics and Use Cases
Performance is measured in FLOPS, with the fastest systems reaching exascale levels above one quintillion calculations per second, and rankings are updated semi-annually through the TOP500 and Green500 lists. These computers support tasks such as fusion energy research, astrophysics simulations, drug discovery, and artificial intelligence training at scales that require massive parallelism and high-speed data movement across thousands of nodes Nvidia Supercomput