NVIDIA Powers Global AI Supercomputing Surge

At the SC25 conference in St. Louis, NVIDIA revealed a massive expansion of AI-accelerated scientific computing infrastructure globally. More than 80 new systems powered by NVIDIA technology have been deployed worldwide in the past year, collectively delivering 4,500 exaflops of AI performance. This represents a significant scaling of computational resources available to researchers across diverse scientific fields.

The centerpiece of this expansion is Horizon, America's largest academic supercomputer being built at the Texas Advanced Computing Center. Scheduled for 2026 deployment, the 300-petaflop system will feature NVIDIA GB200 NVL4 and Vera CPU servers interconnected with Quantum-X800 InfiniBand networking. With 4,000 Blackwell GPUs, Horizon is designed to deliver up to 80 exaflops of AI compute at FP4 precision.

John Cazes, director of high-performance computing at TACC, emphasized the system's transformative potential. "Horizon will enable our scientists to pursue ambitious scientific research at unprecedented scale," he said. The supercomputer will support applications ranging from molecular dynamics of viral infections to seismic activity simulation and astronomical data analysis.

The U.S. Department of Energy is significantly expanding its NVIDIA-powered infrastructure with seven new AI supercomputers at Argonne and Los Alamos National Laboratories. The largest system, Solstice at Argonne, will feature 100,000 NVIDIA Blackwell GPUs capable of reaching 1,000 exaflops of AI training compute—reportedly over 50 times the combined AI training capacity of the entire TOP500 list from June 2025.

European research institutions are also scaling up their computational capabilities. Germany's Jülich Supercomputing Centre has inaugurated JUPITER, Europe's first exascale computer featuring 24,000 NVIDIA GH200 Grace Hopper Superchips. Thomas Lippert, director of the center, noted that "with over 1 exaflop of computing power on JUPITER, our researchers can now run global simulations at kilometer-scale resolution."

Asian nations are making substantial investments in NVIDIA-accelerated infrastructure through sovereign AI initiatives. Japan's RIKEN research institute is integrating NVIDIA GB200 NVL4 systems in two new supercomputers for AI science and quantum computing research. The country's National Institute of Advanced Industrial Science and Technology has launched ABCI-Q, described as the world's largest research supercomputer dedicated to quantum computing.

South Korea plans to deploy over 50,000 NVIDIA GPUs across sovereign clouds and AI factories, with industry leaders Samsung, SK Group, and Hyundai Motor Group building additional AI infrastructure. In Taiwan, NVIDIA is collaborating with Foxconn to construct an AI factory supercomputer with 10,000 Blackwell GPUs to support research and industrial innovation.

This global expansion of accelerated computing infrastructure comes as research institutions increasingly rely on AI-driven approaches for complex scientific s. The systems are being deployed across quantum physics, digital biology, climate research, materials science, and healthcare applications, potentially accelerating timelines across multiple scientific domains.