Nvidia and Hewlett Packard Enterprise unveiled plans at a major supercomputing event in Hamburg to develop a new high-performance computing system named Blue Lion in partnership with Germany´s Leibniz Supercomputing Centre. Blue Lion is set to deliver around 30 times the computational capability of the current SuperMUC-NG system. At the heart of Blue Lion will be Nvidia´s forthcoming Vera Rubin architecture, which melds the Rubin GPU with Nvidia’s first custom CPU, Vera. This integrated platform aims to bring together simulation, data processing, and Artificial Intelligence under a single high-bandwidth, low-latency umbrella, specifically tailored for ambitious scientific tasks.
The technical build will feature HPE’s latest Cray supercomputing technologies, leveraging Nvidia GPUs alongside state-of-the-art storage and connectivity solutions. An innovative aspect of the Blue Lion project is its use of HPE’s entirely fanless direct liquid-cooling system: warm water circulates through pipes to draw heat away from the supercomputer’s critical components. These environmental considerations extend further, as the system’s waste heat will actively provide warmth to surrounding buildings, demonstrating a sustainable approach to energy use in large-scale computing environments.
Scheduled for researcher access by early 2027 at the Leibniz Supercomputing Centre, Blue Lion will serve a broad range of scientific fields, notably climate research, physics, and machine learning. The announcement follows Nvidia´s disclosure of a parallel Vera Rubin-based supercomputer project at the Lawrence Berkeley National Lab in the United States, named Doudna, which will process data sources including telescopes, genome sequencers, and fusion experiments. As these new systems come online, they promise to set new benchmarks for scientific discovery, sustainability, and integration of Artificial Intelligence with traditional numerical simulation.
