NVIDIA at ISC 2026: 35 New AI Supercomputers, Vera Rubin Platform, and Europe's Sovereign AI Infrastructure

Italy's IT4LIA (CINECA): A new supercomputer joining the existing infrastructure at the Italian supercomputing center. IT4LIA already operates over 20,000 GPUs across its LISA, GAIA, Leonardo, and MEGARIDE systems, and the new IT4LIA system will expand capacity further .

Germany's HammerHAI (HLRS Stuttgart): The first standalone supercomputer deployed under the EuroHPC AI Factories initiative. A contract with HPE was signed in March 2026 for installation at the High-Performance Computing Center Stuttgart (HLRS) .

Germany's BavariaAI — Blue Swan: The BavariaAI consortium's "Blue Swan" system was named among the key installations building on next-generation NVIDIA AI infrastructure .

Vera Rubin Platform — Over 7 Exaflops per Rack

NVIDIA unveiled the Vera Rubin NVL72 rack-scale system, the flagship of the Vera Rubin platform. Key specifications confirmed at ISC 2026:

• 7 exaflops of AI for science performance per rack-scale system
• 5 petaflops of native FP64 performance for traditional HPC simulation
• The NVL72 integrates 72 Rubin GPUs and 36 Vera CPUs per rack, unified via NVLink 6 and Quantum-X800 InfiniBand
• Global system manufacturers — Bull, Dell Technologies, GIGABYTE, HPE, and Supermicro — announced custom high-density Vera Rubin systems with up to 144 GPUs
• A Supermicro DCBBS blueprint was also revealed, scaling up to 1,152 Rubin GPUs and 576 Vera CPUs in liquid-cooled racks
• Vera Rubin is described as packing "TOP500 supercomputing in a single rack"

Clean Energy: Siemens Energy Hydrogen-Capable Turbines

NVIDIA highlighted its work with Siemens Energy on accelerating hydrogen-capable gas turbine burner design using AI-driven simulation — part of a broader push to apply the new European supercomputing capacity toward clean-energy decarbonization .

Quantum Computing: CUDA-Q and QPU Integration

Multiple quantum computing initiatives were announced:

• CINECA (IT4LIA) is working with EuroHPC and French quantum company Pasqal to integrate a neutral-atom quantum processing unit (QPU) into its computing infrastructure, using NVIDIA's CUDA-Q platform as the bridge between classical GPU compute and quantum processors .
• Barcelona Supercomputing Center, CINECA, Fraunhofer, and Jülich Supercomputing Centre are the latest institutes adopting CUDA-Q for hybrid quantum-classical workflows .
• FirstQFM demonstrated a quantum-powered forecasting system at ISC 2026 built on NVIDIA accelerated computing and CUDA-Q, outperforming leading classical forecasting models .
• CUDA-Q is NVIDIA's open-source platform for programming hybrid CPU-GPU-QPU architectures, supporting Python and C++ .

Connection to JUPITER

The 2026 announcements build directly on last year's JUPITER supercomputer — Europe's first exascale system, hosted at Jülich Supercomputing Centre in Germany. JUPITER, also NVIDIA-powered, set the template for the EuroHPC AI Factories model. The 2026 wave of 35 new systems scales that concept across the continent, with HammerHAI as the first new standalone AI Factory following JUPITER's architecture .

Broader Significance for Europe's Sovereign AI Infrastructure

• Strategic scale: With 35 systems across 23 countries, Europe now has the most geographically distributed sovereign AI infrastructure of any region, reducing dependence on non-European cloud providers .
• Science and industrial use: The infrastructure targets climate modeling, healthcare, computational fluid dynamics, clean energy (Siemens Energy), and quantum computing — not just large language models .
• Vera Rubin as a multiplier: The new platform lets European centers deploy TOP500-class capability in a single rack, dramatically lowering the facility cost barrier for smaller national centers .
• Open ecosystem: CUDA-Q enables European quantum startups (Pasqal, IQM) to integrate QPUs with GPU supercomputers, establishing a hybrid classical-quantum computing stack within Europe .