Nvidia recently announced that it will build and launch 35 high-performance computing (HPC) supercomputers for artificial intelligence across Europe this year, which the company calls the largest single-year AI infrastructure expansion in Europe's history. These systems were officially disclosed at the ISC High Performance 2026 conference and will be deployed in a number of national supercomputing centers, AI factories and scientific research institutions, providing advanced computing resources to more than three million scientific researchers. NVIDIA founder and CEO Jensen Huang said, "AI is a new scientific tool, and Europe is building infrastructure to put this tool into the hands of millions of researchers."
According to reports, these new systems are based on NVIDIA's latest Blackwell and Hopper architectures and will focus on supporting model training and simulation work in key scientific research fields such as climate science, medical health, clean energy, and quantum computing. Among the key projects that have been announced, include the MareNostrum 5 AI upgrade plan of the Barcelona Supercomputing Center (BSC) in Spain, the Blue Swan platform of BavariaAI in Germany, the IT4LIA AI factory project in Italy, the HammerHAI project in Germany, and the Mimer AI Factory in Sweden.
In Spain, the Barcelona Supercomputing Center plans to bring a large-scale AI computing power upgrade to MareNostrum 5 by introducing NVIDIA GB300 NVL72 and GB200 NVL4 systems. The center estimates that after the upgrade is completed, the system can provide up to about 20 EFLOPS of AI training performance and 33 EFLOPS of AI reasoning performance, providing higher accuracy and higher efficiency simulation capabilities for large-scale scientific research projects in climate modeling, biotechnology, energy systems and other directions.
In Italy, the IT4LIA project will deploy more than 8,000 GPUs based on NVIDIA's GB200 NVL4 architecture, which is regarded as one of the largest AI factories announced so far. At the same time, Germany's Bavaria AI's Blue Swan platform will be equipped with about 1,000 GPUs to develop multi-modal AI models serving medical, robotics and various scientific research fields, focusing on intelligent analysis and assisted decision-making needs in practical application scenarios.
NVIDIA also emphasized in this release that the application of accelerated computing in energy and climate-related research is growing rapidly. For example, Siemens Energy is using an accelerated computing platform based on NVIDIA technology to significantly speed up the design and simulation process of hydrogen-compatible gas turbines. With the help of these computing power acceleration solutions, Siemens has reduced the time spent on related complex simulation tasks by up to 77%, helping to iterate faster on the design of new generation clean energy equipment.
Quantum computing is also one of the important directions of this layout. Italy's CINECA, EuroHPC and Pasqal are using the NVIDIA CUDA-Q platform to integrate quantum processing units into the CINECA supercomputing center, allowing traditional high-performance computing and quantum computing to closely collaborate. In Germany, researchers at the Jülich Supercomputing Center completed a simulation of a universal 50-qubit quantum computer on the JUPITER supercomputer, providing a verification basis for future research on higher-scale quantum systems.
NVIDIA pointed out that the construction of large-scale AI supercomputers in Europe reflects the region’s determination to continue to increase its AI and supercomputing infrastructure. As governments, scientific research institutions and technology companies of various countries compete to expand computing power and consolidate their competitiveness in cutting-edge scientific research fields, these 35 new systems are regarded as an important fulcrum to further improve Europe's AI scientific research capabilities.