Europe is preparing for a major increase in AI computing capacity through 35 new NVIDIA based supercomputers spread across 23 countries. The planned systems are expected to support scientific research, public sector projects, healthcare, climate studies, advanced manufacturing, and AI model development.
Together, the new deployments are expected to provide as much as 800 exaflops of AI compute. That level of capacity reflects how quickly European governments, research institutions, and technology organizations are investing in large scale AI infrastructure.
The systems will use a mix of NVIDIA Hopper, Blackwell, and upcoming Vera Rubin hardware. They will also rely on NVIDIA networking, software libraries, enterprise tools, and AI services to help research teams use the systems for complex workloads.
Europe’s AI Infrastructure Push Will Support Millions of Researchers
The new supercomputers are expected to support more than three million researchers across Europe. The projects will be hosted at national supercomputing centers, universities, AI factories, and research institutions.
The goal is to give scientists and public organizations access to advanced computing resources without relying entirely on infrastructure outside Europe. These systems could be used for training AI models, running scientific simulations, analyzing medical data, improving weather forecasting, studying climate change, and developing new materials.
Several major projects are already planned across Spain, Germany, Italy, Sweden, and other countries. Some systems will focus on general AI research, while others will support government services, energy research, robotics, biotechnology, and industrial simulation.
| Supercomputer Project | Location | Planned Hardware | Expected AI Performance |
|---|---|---|---|
| MareNostrum 5 AI Factory expansion | Spain | GB300 NVL72 and GB200 NVL4 | Up to 20 exaflops training |
| Blue Swan | Germany | GB200 NVL4 systems | Up to 11 exaflops training |
| IT4LIA AI Factory | Italy | More than 8,000 GB200 GPUs | Up to 82 exaflops training |
| HammerHAI | Germany | More than 850 GB200 GPUs | Up to 8 exaflops training |
| Mimer AI Factory | Sweden | 100 GB200 NVL4 systems | Up to 4 exaflops training |
| JUPITER | Germany | Advanced accelerated supercomputer platform | Europe’s first exascale system |
JUPITER Will Support Climate, Brain and Wireless Research
One of the most important systems in the wider rollout is JUPITER, Europe’s first exascale supercomputer. It is expected to help researchers model the human brain at cellular scale, simulate global climate conditions at very high detail, and support research into future wireless technologies including 5G and 6G.
These workloads require huge amounts of computing power because they involve massive datasets and highly detailed simulations. Climate modeling alone can require years of data, complex physics calculations, and extremely large memory systems.
JUPITER is also expected to support quantum computing research by simulating advanced quantum systems. This could help scientists test algorithms and develop new methods before large scale quantum computers become widely available.
Vera Rubin Systems Will Arrive in Future European Deployments
Europe is also preparing for systems based on NVIDIA’s upcoming Vera Rubin platform. Vera Rubin is expected to deliver high AI performance in a rack scale design, giving research centers another option for deploying large AI systems.
Some future installations are expected to use high density racks with up to 144 GPUs. These configurations could provide significantly more computing capacity than existing systems while reducing the amount of physical space needed in data centers.
The planned expansion shows that Europe is treating AI computing as a long term strategic investment. The challenge will be making sure these systems are used effectively, shared across institutions, and supported with enough energy, networking, and skilled researchers to match their scale.
Discussion (0)
Be the first to comment.