HPE takes six of top 10 spots in supercomputer ranking

HPE has taken six of the top 10 positions in the latest TOP500 ranking of the world's fastest supercomputers. Three of those systems are among the five verified exascale machines on the list.

The result gives HPE the largest share of the top tier of the global supercomputing table, with systems it built accounting for more than 11.4 exaflops of combined performance.

In the TOP500 ranking, El Capitan at Lawrence Livermore National Labouratory is second with 1.809 exaflops, Frontier at Oak Ridge National Labouratory is third with 1.353 exaflops, and Aurora at Argonne National Labouratory is fourth with 1.012 exaflops. All three are verified exascale systems, a small group still concentrated in the US national laboratory network.

Beyond those three HPE-built machines, the company also has systems ranked sixth, eighth and tenth: HPC7 and HPC6, both used by Italian energy group Eni, and Alps at the Swiss National Supercomputing Centre.

HPC7 entered the list at number six with 571.5 petaflops, making it the highest-ranked enterprise system in the current table. The machine is used for industrial applications including facility operations, geological and fluid dynamics studies for carbon dioxide storage, battery development, and work on sector-specific AI models.

Its older counterpart, HPC6, ranked eighth with 477.9 petaflops. Alps, which rounded out HPE's top-10 presence, reached 434.9 petaflops and supports research in health, climate, materials science, and engineering, while also serving the Swiss AI Initiative.

Benchmark standings

HPE also led a separate benchmark focused on mixed-precision calculations used in both traditional high-performance computing and AI workloads. El Capitan, Frontier and Aurora took the top three places on the HPL-MxP benchmark, reinforcing the role of large scientific machines in AI-related research as well as simulation work.

El Capitan led that benchmark while remaining the second-fastest machine overall. Frontier ranked second on HPL-MxP and Aurora third.

HPE's showing also extended to energy-efficiency rankings. Four HPE-built systems appeared in the top 10 of the Green500 list, which measures supercomputing efficiency.

Those systems were Isambard-AI at the University of Bristol's Bristol Centre for Supercomputing, Samsung Electronics' SSC-24 Energy Module, Helios GPU at the Academic Computer Centre Cyfronet AGH in Poland, and Portage, an HPE benchmarking system. Isambard-AI ranked fourth on Green500, while SSC-24 Energy Module was seventh, Helios GPU eighth and Portage tenth.

Several of HPE's highest-ranked machines on the TOP500 also placed on the efficiency list, though outside the top 10. El Capitan ranked 28th on Green500, Frontier 20th, and DAEDALUS 23rd.

Greek debut

A new entrant from Greece also featured in the rankings. DAEDALUS, built for the National Infrastructures for Research and Technology with the Ministry of Digital Governance and EuroHPC JU, debuted at number 31 on the TOP500 with 85.69 petaflops.

The system is now the fastest supercomputer in Greece and serves as the computing core of the AI Factory Pharos. It is intended to support research and innovation in AI, medicine, meteorology, big data analysis, and smart transport systems.

The concentration of HPE-built systems near the top of the rankings underlines the limited number of vendors able to design and integrate machines at this scale. The exascale tier in particular remains dominated by projects commissioned by governments and national laboratories, with only a handful of installations worldwide verified at that level.

At the same time, the appearance of Eni's systems in the top 10 points to continued industrial demand for supercomputing outside public research institutions. Enterprise users have increasingly turned to large-scale computing for modelling, energy analysis, and AI development, even as the highest end of the market remains shaped by public funding and sovereign research priorities.

HPE has also widened the software and infrastructure around its supercomputing portfolio. That includes new programming software now available on HPE ProLiant Compute servers, as well as multi-tenant features in networking and storage aimed at high-performance computing environments.

The company has also introduced financial services for customers retiring older air-cooled computing infrastructure. The move reflects a broader industry shift toward liquid-cooled systems as operators seek to manage power use and heat density in larger AI and scientific workloads.

Among the systems highlighted by HPE, Isambard-AI stands out as a UK public-sector machine tied to the government's AI Research Resource. In Switzerland, Alps supports a national AI programme involving more than 1,200 researchers, while DAEDALUS is positioned as a national asset for Greece's research base.

The rankings also show how supercomputers are now judged across multiple measures rather than raw speed alone. Performance on mixed-precision benchmarks and energy-efficiency tables has become increasingly important as AI workloads reshape the economics and design priorities of large computing systems.

HPE said its Green500 results were supported by its fanless direct liquid-cooling design, with Isambard-AI among the strongest examples of that approach in operation.