Nvidia will provide the processing power for the world’s most powerful AI supercomputer, a cutting-edge system named “Leonardo” currently under development by CINECA, an Italian multi-university consortium and a prominent international leader in supercomputing. Upon completion, the Leonardo system is projected to deliver up to 10 exaflops of FP16 AI performance, utilizing over 14,000 Nvidia GPUs based on the Ampere architecture.

Leonardo represents one of four new supercomputers being developed through a collaborative, pan-European initiative aimed at enhancing high-performance computing resources throughout the region. These advanced systems will ultimately provide sophisticated AI capabilities to support a wide range of applications in both scientific research and various industries. To facilitate high performance and efficient data transfer across the clusters, Nvidia will also provide its Mellanox HDR InfiniBand networking technology, enabling low-latency, high-bandwidth connections.

Alongside Leonardo, the initiative includes the MeluXina supercomputer in Luxembourg and Vega in Slovenia, as well as a new supercooling facility being established in the Czech Republic. The European consortium is also planning to deploy four additional supercomputers in Bulgaria, Finland, Portugal, and Spain, although details regarding their specifications and precise locations are not currently disclosed.

CINECA and the other supercomputers will be utilized for diverse applications, such as genomic analysis and the identification of novel therapeutic strategies; integrating data from numerous sources to support space exploration and the study of planets beyond Earth; and creating models of weather systems, including the prediction of severe weather occurrences.