The latest TOP500 and Green500 supercomputer rankings, released this week at the ISC High Performance conference in Hamburg, Germany, underscore NVIDIA’s accelerating dominance in high-performance computing. The company now powers more than 400 of the world’s 500 fastest systems, capturing 81 percent of the total. This represents a gain of 17 installations since the previous assessment, with deployment momentum heavily skewed toward accelerated architectures. Nearly nine of every ten newly ranked systems rely on NVIDIA infrastructure, reflecting a strategic industry shift toward unified platforms optimized for artificial intelligence, scientific simulation, and enterprise workloads. Performance metrics highlight a widening competitive gap. NVIDIA-accelerated systems currently deliver over twice the artificial intelligence training throughput and nearly three times the inference capacity of all competing platforms combined. Adoption rates for core components also reached unprecedented levels. GPU acceleration was deployed across a record 238 systems, while NVIDIA networking solutions connected 376 installations. The vast majority utilize NVIDIA Quantum InfiniBand as the foundational fabric for large-scale AI and high-performance computing, with the remainder transitioning to high-speed Ethernet. NVIDIA’s market leadership is reinforced by a comprehensive full-stack approach encompassing GPUs, networking, and increasingly, CPU architectures. Grace CPU deployment has expanded to 26 systems, with cumulative shipments surpassing 2.5 million units. Machines built on the NVIDIA Grace Hopper Superchip, which unifies GPU and CPU memory access to minimize latency, currently occupy top positions across both major rankings. JUPITER, Europe’s first exascale system, ranks fifth on the TOP500 and leads scientific initiatives including cellular-scale human brain mapping, climate simulation, and 6G network development. Meanwhile, KAIROS, located at the University of Toulouse, claims the number one spot on the Green500 efficiency list, achieving 73.3 gigaflops per watt. Grace Hopper platforms occupy the top four efficiency rankings, demonstrating the architecture’s advantages for memory-intensive AI workloads. The industry’s transition toward next-generation hardware is already visible in the latest rankings. Systems built on the NVIDIA Blackwell architecture, including B200 and GB200 configurations, have entered the rankings across Asia, Europe, and the United States, with initial deployments debuting in Japan. This hardware refresh coincides with a massive infrastructure expansion in Europe, where 35 AI-enabled high-performance computing supercomputers are currently under development to support over three million researchers and drive industrial innovation. Beyond established markets, NVIDIA’s infrastructure is rapidly scaling globally. New deployments span continents, ranging from emerging AI factories in South Africa to national-scale systems in Saudi Arabia, Singapore, and Vietnam. The sustained expansion of Grace CPU lines, alongside the upcoming Vera CPU launch designed to further enhance data center efficiency and agent-driven workloads, signals continued architectural innovation. As accelerated computing solidifies its role as the foundation for the world’s most demanding computational tasks, NVIDIA’s comprehensive stack continues to define the standard for global AI and scientific infrastructure.