The Leibniz Supercomputing Centre (LRZ) in Germany is set to introduce Blue Lion, a supercomputer poised to deliver approximately 30 times the computing power of its current flagship, SuperMUC-NG. For the first time, LRZ, a part of the Gauss Centre for Supercomputing, confirms it will deploy the NVIDIA Vera Rubin architecture, moving beyond generic references to ´next-generation´ NVIDIA hardware. This marks a significant step for both the center and the broader research community in adopting cutting-edge solutions for high-performance computing.
NVIDIA Vera Rubin is an integrated superchip platform, combining the forthcoming Rubin GPU (succeeding the Blackwell line) and the Vera CPU—NVIDIA´s first custom processor engineered to operate seamlessly with the GPU. Designed to consolidate simulation, data processing, and Artificial Intelligence, this architecture promises a unified, high-bandwidth, low-latency environment for scientific advancement. The platform features shared memory, coherent compute capabilities, and in-network acceleration, with Blue Lion set to support researchers in climate science, turbulence modeling, physics, and advanced machine learning. Hewlett Packard Enterprise (HPE) will build Blue Lion on Cray hardware, utilizing a direct liquid-cooling system that recycles heat to warm local infrastructure while minimizing energy waste.
Blue Lion’s significance is amplified by its parallel with the Doudna supercomputer, under construction at Lawrence Berkeley National Laboratory in California and due to utilize the same Vera Rubin architecture. Doudna, built by Dell Technologies and named for Nobel laureate Jennifer Doudna, will serve over 11,000 researchers and feature real-time workflows through high-speed NVIDIA Quantum-X800 InfiniBand networking. Both Blue Lion and Doudna are engineered to maximize scientific outputs with efficient power use, offering leaps in performance per watt over predecessors.
Together, these systems exemplify a new era where Artificial Intelligence and simulation are deeply intertwined, data flow is continuous, and computing is tightly integrated into the scientific process. They mark a pivotal transition from siloed research domains to unified, agile platforms capable of supporting real-time science at global scales. The adoption of NVIDIA Vera Rubin is at the forefront of this transformation, indicating a fundamental rethinking in the design and deployment of high-performance computing systems.
