A new supercomputer in Germany has achieved a processing speed of one quintillion calculations per second, marking a significant milestone for European computing infrastructure. The Jupiter supercomputer, located at the Jülich Supercomputing Centre in North Rhine-Westphalia, operates using 24,000 Nvidia chips and represents Europe’s first entry into the highest tier of supercomputing dedicated to artificial intelligence training.
For years, this level of computing power was concentrated in the United States and China, with Europe relying on foreign infrastructure for advanced AI work. Jupiter changes this dynamic by providing European researchers with domestic computing capacity for training large AI models, including language systems and computer vision tools. The machine is the flagship project of the European High-Performance Computing Joint Undertaking (EuroHPC), a multinational initiative that has coordinated funding and development efforts across Europe. The Jülich facility has housed some of Europe’s fastest computers for decades, but none have reached this level of performance.
During preliminary testing, Jupiter recorded 793.4 Petaflop/s, placing it fourth on the global Top 500 list of supercomputers. Only three machines currently rank higher: the US Department of Energy’s El Capitan, Frontier, and Aurora.
All three are American systems. Jupiter’s architecture is specifically optimized for AI training, representing a shift from older supercomputers that focused primarily on simulations such as weather modeling, physics calculations, and drug discovery. While Jupiter can perform those tasks as well, its design centers on the parallel processing required by modern AI applications.
Data sovereignty considerations played a key role in the project’s development. The European Union’s General Data Protection Regulation (GDPR) restricts the transfer of personal data outside the EU, creating legal complications for AI companies that previously relied on American or Chinese supercomputers.
Jupiter allows data to remain within European borders throughout the training process. The timing of Jupiter’s deployment coincides with China’s rapid expansion of state-backed supercomputing projects, which have raised concerns about data privacy, intellectual property protection, and military applications. Europe has been caught between these competing powers, dependent on foreign computing resources for cutting-edge AI work.
While Jupiter represents a single machine and does not close the overall computing capacity gap between Europe and the leading nations, it signals Europe’s intent to compete in AI infrastructure. The supercomputer is now operational, with initial tests confirming its performance levels.
Researchers have begun using the system for their work.
