Jupiter: Europe’s First Exascale Supercomputer Powers AI and Scientific Breakthroughs

Jupiter, the first supercomputer in Europe, has officially entered the exascale era, and is capable of performing more than one quintillion calculations per second. Constructed at the Forschungszentrum Juelich research facility in Germany, Jupiter represents a breakthrough in computer power and efficiency.
Intended to drive advances in AI, quantum simulation, and climate modeling, it is a symbolic move by Europe toward becoming digitally sovereign and scientifically independent, comparable in scale and performance to the best U.S. and Chinese systems.
Technical Specs & Architecture
The Jupiter is designed around a modular architecture (BullSequana XH3000) with almost 24,000 NVIDIA GH200 Grace Hopper superchips in the Booster module, which are linked by Quantum-2 InfiniBand interconnect. Every node is equipped with four GH200 chips and 288 Arm Neoverse cores and up to 2.3 TB of high-speed memory.
At the performance front, Jupiter has been able to achieve approximately 793 petaflops in FP64 (double-precision) benchmarks and is projected to achieve full exascale (1 exaflop) performance. It is expected to provide more than 70-90 exaflops on AI workloads (8-bit/FP8).
Its JEDI module tops the Green500 list in terms of energy efficiency, featuring a warm-water direct liquid-cooling system that recycles waste heat and provides the equivalent of approximately 60 GFLOPS/watt. Electrical power consumption of about 11 megawatts under normal operation, 18.2 MW under peak.
Funding, Governance, & Stakeholders
Jupiter was constructed by a French-German joint venture, headed by Eviden (a part of Atos) and ParTec. It is a joint ownership and funding: half the cost is funded by EuroHPC JU, and the other half by Germany – half of it by the Federal Ministry of Education and Research (BMBF) and the other half by the Ministry of Culture and Science of the State of North Rhine-Westphalia (MKW NRW) via the Gauss Centre for Supercomputing. It is housed and run by the Juelich Supercomputing Centre (JSC), a division of Forschungszentrum Juelich, in Germany.

Purposes & Use Cases
Jupiter is designed to drive advanced scientific studies in a wide range of disciplines. It contributes to climate modelling (such as km-scale global weather forecasts through the Destination Earth program at ECMWF), quantum and molecular simulations, astrophysics, structural biology and other core science areas.
Jupiter accelerates the training of large models (such as foundation models, multilingual LLMs) on the AI / ML side, and inference workloads. It belongs to the JUPITER AI Factory (JAIF) that reduces the barriers to start up, SMEs, industry and academia to access training, testing and deployment of AI models, particularly in EU data protection and security standards.
In the case of industry, small business, and startups, open calls and the AI Factory allow them to access the scale of Jupiter without constructing their own supercomputers. That contributes to the faster innovation of areas such as healthcare, energy, manufacturing, and environment.
Impact & Significance
Jupiter signifies a pivot to digital sovereignty in Europe, contributing to narrowing the divide between the U.S. and China in high-performance computing and AI. As the first exascale supercomputer in Europe, it enhances the autonomy of science and minimizes the use of foreign infrastructure.
To science and the environment, Jupiter can be much more precisely modelled, such as weather, climate, extreme events (floods, heatwaves) at kilometre-scale resolution, and sustainably harness computing resources, including energy efficiency, renewable energy sources, and waste heat reuse.
It might lead to breakthroughs in health, energy, materials and AI available to startups and to public research institutions; in the long term, it provides long-term economic, educational, and environmental returns.

Challenges & Limitations
A machine as powerful as Jupiter has its trade-offs and limits to consider. A large issue is power consumption and infrastructure: to run at average loads requires about 11-15 megawatts, and requires a strong power supply system, cooling, and facilities to handle heat generation.
Next are the costs and maintenance – the price tag of €500 million is only a portion of it. Maintenance, updates, staffing and software and hardware maintenance to meet changing demands all have a cumulative effect.
Finally, regulatory bottlenecks and accessibility may restrict the actual users of Jupiter. Users have to apply through competitive calls; the sovereignty of data, compliance with regulations, and the fair access of EU members or academic and industrial users poses challenges.
Read Also: 5 Practical Ways to Enhance Apartment Door Security When Renting
Future Prospects
JUPITER is already entering its second stage. A key advance in development is the JUPITER inference module JARVIS, which will enable fast model inference and refinements learning to supplement the already existing Booster and Cluster modules.
More than that, Europe is spreading its network of AI Factories, with other locations in Austria, Bulgaria, France, Germany, Poland and Slovenia. These will collaborate with JUPITER to expand access and AI innovation in research, startups, industries.
Some of the new areas that will be useful include quantum simulation, climate digital twins, high-resolution probabilistic weather forecasting, and multilingual AI. JUPITER is likely to have an increased scientific, environmental, and social impact as it opens up more hours to outside projects (such as ECMWF and the Destination Earth project).
Final Word
Jupiter is the next step in the scientific and technological history of Europe – the first exascale supercomputer, and one of the most energy-efficient on Earth. It is a radical step towards digital sovereignty and dominance in AI and advanced research. It is an opportunity for global science: quicker discoveries, more comprehensive simulations and cleverer innovation. With the ever-growing demand of computers, the ability to keep up with them, such as Jupiter, will determine the next generation of knowledge and advancement.
Read More: Remote Work in Tech – How to Adapt Your Recruitment Strategy?
No comment