NVIDIA has just made history in Europe. The newly launched JUPITER supercomputer, built with nearly 24,000 NVIDIA Grace Hopper GH200 Superchips, is now officially Europe’s fastest and most energy-efficient supercomputer. Hosted in Germany and backed by the EuroHPC Joint Undertaking, JUPITER is built to lead the future of science, research, and AI.

Double the Speed, Exascale Power

JUPITER delivers more than 2x the performance of the next-fastest system in Europe. It’s also on track to become Europe’s first exascale supercomputer, meaning it will soon be capable of running 1 quintillion (1,000,000,000,000,000,000) FP64 operations per second.

This massive computing power is driving breakthroughs in AI, climate science, quantum research, structural biology, and more.

Designed for the Future of AI and HPC

JUPITER isn’t just fast — it’s smart. It supports training and inference of the largest AI models, helping researchers simulate the most complex scientific phenomena. From designing new drugs to building climate digital twins, this system is built to solve the world’s hardest problems.

It’s powered by,

• NVIDIA Quantum-2 InfiniBand networking
• Eviden’s BullSequana XH3000 liquid-cooled architecture
• A full NVIDIA AI software stack for optimized performance

With over 90 exaflops of AI performance, JUPITER ranks among the top five supercomputers in the world and leads the pack in energy efficiency, with 60 gigaflops per watt.

Built to Drive Scientific Discovery

Hosted at the Jülich Supercomputing Centre in Germany, JUPITER was created in partnership with NVIDIA, Eviden (Atos Group), and EuroHPC JU. It took just nine months to design, build, and launch — a remarkable achievement in high-performance computing.

Real-World Impact Areas

JUPITER will help advance innovation across key scientific domains.

Climate Modeling: High-resolution simulations to predict and address climate change (via NVIDIA Earth-2 and Earth Virtualization Engines).

Drug Discovery: Faster AI-based research using NVIDIA BioNeMo.

Computational Engineering: AI-driven design through NVIDIA PhysicsNeMo, CUDA-X, and Omniverse.

Quantum Research: Hybrid quantum-classical computing with CUDA-Q and cuQuantum.

Astrophysics & Structural Biology: Training of large foundation models for scientific innovation.

Voices from the Launch

AI will supercharge scientific discovery and industrial innovation.

Jensen Huang, CEO of NVIDIA

With JUPITER’s extreme performance, Europe has taken a giant leap into the future of science and sovereignty,

Anders Jensen, Executive Director, EuroHPC JU

JUPITER is advancing the frontier of AI and simulation,

Thomas Lippert, Co-director, Jülich Supercomputing Centre

Apply for Access

Researchers across Germany and Europe can now apply for access to JUPITER to accelerate their scientific projects and innovations.