Jupiter Booster
 | This article may incorporate text from a large language model. (September 2025) |
 | |
| Active |
|
|---|---|
| Sponsors | EuroHPC JU |
| Operators | Forschungszentrum Jülich, Jülich Supercomputing Centre |
| Location | Forschungszentrum Jülich, North Rhine-Westphalia, Germany |
| Architecture | Modular architecture with NVIDIA GH200 Grace Hopper Superchips |
| Power | 18.2 megawatts |
| Operating system | Linux-based HPC OS |
| Space | TBA |
| Memory | 288 Arm Neoverse cores capable of achieving 16 petaflops of AI performance using up to 2.3 terabytes of high-speed memory |
| Storage | High-capacity, high-speed storage solutions |
| Speed | 1.000 exaFLOPS (Rmax) / 1.226 exaFLOPS (Rpeak) (as of November 2025) |
| Cost | €500 million |
| Purpose | Scientific research and development, AI model training, climate modeling |
Jupiter is an exascale supercomputer hosted at Forschungszentrum Jülich in North Rhine-Westphalia, Germany. Developed by the Jülich Supercomputing Centre (JSC) and owned by the European High-Performance Computing Joint Undertaking (EuroHPC JU), Jupiter became operational in June 2025. It is based on a modular architecture featuring NVIDIA GH200 Grace Hopper Superchips and is recognized as Europe's fastest supercomputer, ranking 4th on the November 2025 TOP500 list of the world’s fastest supercomputers. Jupiter is also the most energy-efficient system among the top five.[1][2]
Design
[edit]Jupiter uses a modular architecture with around 24,000 NVIDIA GH200 Grace Hopper Superchips, optimized for highly parallel applications such as training AI models or numerically demanding simulations. This enables the training of the largest AI models, known as large language models (LLMs). At full capacity, Jupiter can complete such tasks in less than one week.[3]
The supercomputer is designed with advanced cooling techniques, including warm water cooling, to manage heat dissipation and improve energy efficiency. The waste heat generated is used to heat buildings and is integrated into the Jülich campus heating network.[4]
Jupiter's storage system includes high-capacity, high-speed solutions to handle vast amounts of data, providing rapid data access and retrieval to support intensive computational tasks. The network infrastructure employs high-bandwidth, low-latency interconnects to ensure high-speed data transfer and communication between nodes.[5]
History
[edit]Jupiter was developed as part of a broader initiative to enhance Europe's computational infrastructure, crucial for maintaining competitiveness in scientific research, technological innovation, and industrial applications. The project involves collaboration among several European countries, research institutions, and technology companies.[6]
In the future, the Booster partition of Jupiter will be supported by a Cluster partition, supplied by ParTec, featuring conventional central processing units (CPUs) with high memory bandwidth. The Cluster is specially designed for data-intensive tasks.[7]
Jupiter was presented at the International Supercomputing Conference (ISC) in Hamburg in June 2025, where it was recognized as the most energy-efficient system among the top five supercomputers globally.[8]
In September 2025, German Chancellor Friedrich Merz, his Minister of Research, Technology and Space Dorothee Bär and European Commissioner for Startups, Research and Innovation Ekaterina Zaharieva inaugurated Jupiter during a visit to Forschungszentrum Jülich.[9]
Funding
[edit]Jupiter is funded by the European High-Performance Computing Joint Undertaking (EuroHPC JU), the Federal Ministry of Research, Technology and Space (BMFTR), and the Ministry of Culture and Science of the State of North Rhine-Westphalia (MKW NRW) via the Gauss Centre for Supercomputing (GCS). The total budget for the acquisition, delivery, installation, and maintenance of Jupiter is €500 million, with €273 million allocated for hardware, software, and services, and the remaining €227 million for power, cooling, and operations.[10]
Early access and research
[edit]Jupiter and its storage systems have been fully installed, with scientific users gaining access in the coming weeks.[when?] Over 100 national and international applications have been selected via the Jupiter Research and Early Access Program (JUREAP), the GCS Exascale Pioneer Call, and the Gauss AI Compute Competition for AI projects.[11]
Applications
[edit]Jupiter supports a variety of applications, including but not limited to:
- Climate modeling and weather forecasting: Enhancing climate and weather simulations to significantly improve predictions of local extreme weather events, such as heavy rain and severe thunderstorms.
- Molecular dynamics and materials science
- Astrophysics and cosmology
- Industrial simulations and engineering
- Training of large language models (LLMs) and other AI models[12]
Collaborations and partnerships
[edit]The project involves collaboration among several European countries, research institutions, and technology companies. Key partners include:
- Forschungszentrum Jülich
- European High-Performance Computing Joint Undertaking (EuroHPC JU)
- Various academic and industrial research organizations[13]
Impact
[edit]Jupiter is anticipated to have a profound impact on scientific research and industrial innovation in Europe. By providing unprecedented computational power, it will[timeframe?] enable researchers to tackle complex problems that were previously infeasible to solve. The supercomputer is a key pillar[peacock prose] of Europe’s digital sovereignty and technological advancement.[14]
Recognition and achievements
[edit]Jupiter was presented at the International Supercomputing Conference (ISC) in Hamburg in June 2025, where it was recognized as the most energy-efficient system among the top five supercomputers globally.[15]
References
[edit]- ^ "JUPITER - Exascale for Europe". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Peeling The Covers Off Germany's Exascale "Jupiter" Supercomputer". NextPlatform. June 12, 2025. Retrieved July 17, 2025.
- ^ "NVIDIA GH200 Grace Hopper Superchip". NVIDIA. Retrieved July 17, 2025.
- ^ "Innovative Cooling Techniques in Jupiter Supercomputer". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Storage and Network Infrastructure of Jupiter". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "EuroHPC Initiative". EuroHPC JU. Retrieved July 17, 2025.
- ^ "ParTec Cluster Partition for Jupiter". ParTec. October 5, 2024. Retrieved July 17, 2025.
- ^ "International Supercomputing Conference 2025". ISC. Retrieved July 17, 2025.
- ^ Andreas Rinke (5 September 2025), Germany's Merz inaugurates Nvidia supercomputer for research Reuters.
- ^ "Funding for Jupiter Supercomputer". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Early Access and Research Programs for Jupiter". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Applications of Jupiter Supercomputer". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Collaborations and Partnerships for Jupiter". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Impact of Jupiter Supercomputer". Forschungszentrum Jülich. Retrieved July 17, 2025.
- ^ "Recognition and Achievements of Jupiter". Forschungszentrum Jülich. Retrieved July 17, 2025.
External links
[edit]- [Official Jupiter Project Page](https://www.fz-juelich.de/en/ias/jsc/jupiter)
- [NextPlatform Article on Jupiter](https://www.nextplatform.com/2025/06/11/peeling-the-covers-off-germanys-exascale-jupiter-supercomputer/)
- [EuroHPC JU Official Website](https://eurohpc-ju.europa.eu/index_en)
- [TOP500 List](https://www.top500.org/lists/top500/2025/06/)
- [NVIDIA Newsroom Article on Jupiter](https://nvidianews.nvidia.com/news/nvidia-grace-hopper-superchip-powers-jupiter-defining-a-new-class-of-supercomputers-to-propel-ai-for-scientific-discovery)
- [Digital Strategy Page on Jupiter](https://digital-strategy.ec.europa.eu/en/factpages/digital-success-stories-jupiter-first-european-exascale-supercomputer)