NVIDIA hardware currently powers 81% of the world’s 500 fastest supercomputers. According to the latest TOP500 list released at the ISC High Performance conference in Hamburg, Germany, NVIDIA technologies are utilized in more than 400 of the 500 ranked systems. This represents an increase of 17 systems since the previous biannual ranking, with nearly nine of every 10 systems new to the ranking built on NVIDIA technologies.
How NVIDIA Dominates the TOP500 Rankings
The reach of NVIDIA’s hardware extends across the entire architecture of modern supercomputing, encompassing GPUs, networking, and CPUs. Data from the TOP500 indicates that NVIDIA GPUs accelerate 238 systems, while 376 systems utilize NVIDIA networking to manage data traffic.
The adoption of the NVIDIA Grace CPU has also seen significant growth. Currently, 26 systems on the list feature the Grace CPU, an increase of eight units compared to the previous reporting period. This growth is largely driven by the NVIDIA Grace Hopper Superchip, which integrates the GPU and CPU to share memory with reduced overhead, a design optimized for the memory-intensive requirements of modern AI.
Why Energy Efficiency Matters in Supercomputing
The Green500 list, which ranks supercomputers based on performance per watt, shows that NVIDIA-powered systems hold the top eight positions. The KAIROS supercomputer, located at the University of Toulouse in France, currently leads the Green500, delivering 73.3 gigaflops per watt using the Grace Hopper architecture.
This emphasis on efficiency is critical as data centers face increasing power constraints. By pairing high-throughput GPUs with energy-efficient CPUs, operators can manage larger AI workloads. Other prominent systems, such as the JUPITER supercomputer in Germany—Europe’s fastest supercomputer and its first to reach exascale—demonstrate how these technologies are being applied to complex scientific challenges, including climate modeling and 6G network development.
What Happens as AI Infrastructure Expands Globally
The deployment of accelerated computing is no longer limited to a few major research hubs. New systems based on the NVIDIA Blackwell architecture, including B200 and GB200 configurations, are now appearing in rankings across the United States, Europe, and Asia.
This global expansion includes national AI initiatives in countries such as Singapore, Saudi Arabia, and Vietnam, alongside private-sector investments. As of the latest reporting, there are 35 NVIDIA AI HPC supercomputers in development across Europe. These projects are designed to provide more than 3 million researchers with the necessary infrastructure for accelerated science and industrial innovation.
Key Takeaways
- Market Share: NVIDIA technology powers more than 400 of the 500 systems on the latest TOP500 list.
- Networking: 376 systems use NVIDIA networking, the vast majority on NVIDIA Quantum InfiniBand, to facilitate large-scale AI operations.
- Efficiency: Nine of the top 10 most energy-efficient supercomputers on the Green500 list utilize NVIDIA hardware.
- CPU Growth: The NVIDIA Grace CPU is now featured in 26 systems, with nearly 2.5 million Grace CPUs shipped.
- AI Throughput: NVIDIA-based systems currently deliver more than 2x the AI training throughput and nearly 3x the AI inference throughput of all other platforms combined.