en.Wedoany.com Reported - According to the latest ranking released this week at the ISC High Performance Computing conference in Hamburg, Germany, NVIDIA technology powers over 400 of the world's top 500 supercomputers, accounting for 81% of the TOP500 list.
Compared to the previous list, the number of NVIDIA-powered systems has increased by 17, with 90% of new entrants built on its technology. This proportion reflects a market preference for machines designed for artificial intelligence (AI), simulation, and scientific computing. Among the TOP500, NVIDIA systems deliver more than twice the AI training throughput of all other platforms combined, and nearly three times the AI inference throughput.
GPU and network adoption rates have reached historic highs. NVIDIA GPUs accelerate a record 238 systems, and NVIDIA networking connects a record 376 systems, the vast majority based on NVIDIA Quantum InfiniBand, with the remainder using Ethernet. Systems featuring the NVIDIA Grace CPU total 26, an increase of 8 from the previous list, with nearly 2.5 million Grace CPUs shipped. The NVIDIA Vera CPU, released earlier this year, builds on Grace's success, elevating CPU performance and energy efficiency to new levels to handle the most demanding AI workloads in modern data centers, where agents transition from answering basic questions to executing actions, running code, using tools, and evaluating results.
In terms of energy efficiency, NVIDIA dominates the Green500 list. The top eight systems all feature NVIDIA GPUs, and nine of the top ten use its technology. Topping the list is the KAIROS system at the University of Toulouse in France, equipped with a single NVIDIA Grace Hopper superchip, achieving an energy efficiency of 73.3 gigaflops/watt. NVIDIA Grace-based machines rank high on both lists: JUPITER is fifth on the TOP500, and Alps is tenth. Each system pairs NVIDIA GPUs with Grace CPUs integrated into a single superchip, allowing both to share memory with minimal overhead.
A record 35 NVIDIA AI HPC supercomputers are under development in Europe, providing infrastructure for over 3 million researchers. This includes Europe's fastest supercomputer, JUPITER, also its first to achieve exascale performance, located at the Jülich Supercomputing Centre in Germany. The latest systems on the list are based on the NVIDIA Blackwell architecture, with B200 and GB200 systems entering rankings in Asia, Europe, and the United States, and the first GB200 systems already debuting in Japan. AI infrastructure construction is unfolding globally, from new AI factories in South Africa to national AI systems in Saudi Arabia, Singapore, and Vietnam. The TOP500 list, updated twice a year, measures the world's fastest supercomputers, while the Green500 list measures computing performance per watt of power consumption.
This article is compiled by Wedoany. All AI citations must indicate the source as "Wedoany". If there is any infringement or other issues, please notify us promptly, and we will modify or delete it accordingly. Email: news@wedoany.com
