The Vera Rubin Platform: 7+ Exaflops of AI in a Single Rack

The centerpiece of NVIDIA's hardware announcements was the Vera Rubin accelerated computing platform. This rack-scale system is designed for scientific supercomputing workloads, including climate modeling, computational fluid dynamics, and energy exploration .

• Performance: A single rack of Vera Rubin delivers more than 7 exaflops of AI for science and 5 petaflops of native FP64 (double-precision) performance. According to NVIDIA, this packs TOP500-class supercomputing into a single rack .
• Architecture: The platform combines NVIDIA Rubin GPUs with Vera CPUs, HBM4 memory (offering 22 Tbps bandwidth), and 3rd-gen NVLink. It was first detailed at CES 2026 and is expected to be available through partners in the second half of 2026 .
• Target Users: Scientific research organizations, national laboratories, and entities involved in energy exploration .

Europe's Record Expansion: 35 New AI Supercomputers

NVIDIA announced a record deployment of 35 NVIDIA AI HPC supercomputers in development across 23 European countries. These systems are expected to provide next-generation infrastructure to over 3 million researchers .

• Aggregate Compute: The combined systems deliver approximately 800 AI exaflops, utilizing both Blackwell and Hopper platforms .
• Notable Systems: Key projects include the Barcelona Supercomputing Center's EuroHPC AI Factory, BavariaAI's Blue Swan, IT4LIA (Czech Republic), HLRS's HammerHAI (Germany), and NAISS's Mimer EuroHPC AI Factory (Sweden) .
• Applications: The systems will support work in climate science, healthcare, clean-energy decarbonization, and industrial digital twins .

Partner Announcements and Ecosystem

Several partner announcements were highlighted at or around ISC 2026, including those from Siemens Energy, HPE, and Dell.

• Siemens Energy + HPE: Siemens Energy is deploying a dedicated global HPC platform through HPE GreenLake for engineering-driven AI. This includes work on hydrogen-capable gas turbine burners, cited as a key clean-energy initiative in NVIDIA's press materials .
• HPE: HPE announced one of the first NVIDIA Vera CPU compute blades (the HPE Cray Supercomputing GX240, available in 2027) and the HPE Compute XD700 server built on the NVIDIA HGX Rubin NVL8 platform. The Vera Rubin NVL72 by HPE is expected to be available in December 2026 .
• Dell: At GTC 2026 (March), Dell introduced a Data Orchestration Engine and new high-performance storage solutions for AI data centers. No new Dell-specific announcement was made at ISC 2026, but Dell remains a key system integrator for the Rubin platform and European supercomputing deployments .

Summary of Key Announcements

The Vera Rubin Platform: 7+ Exaflops of AI in a Single Rack

The centerpiece of NVIDIA's hardware announcements was the Vera Rubin accelerated computing platform. This rack-scale system is designed for scientific supercomputing workloads, including climate modeling, computational fluid dynamics, and energy exploration .

• Performance: A single rack of Vera Rubin delivers more than 7 exaflops of AI for science and 5 petaflops of native FP64 (double-precision) performance. According to NVIDIA, this packs TOP500-class supercomputing into a single rack .
• Architecture: The platform combines NVIDIA Rubin GPUs with Vera CPUs, HBM4 memory (offering 22 Tbps bandwidth), and 3rd-gen NVLink. It was first detailed at CES 2026 and is expected to be available through partners in the second half of 2026 .
• Target Users: Scientific research organizations, national laboratories, and entities involved in energy exploration .

Europe's Record Expansion: 35 New AI Supercomputers

NVIDIA announced a record deployment of 35 NVIDIA AI HPC supercomputers in development across 23 European countries. These systems are expected to provide next-generation infrastructure to over 3 million researchers .

• Aggregate Compute: The combined systems deliver approximately 800 AI exaflops, utilizing both Blackwell and Hopper platforms .
• Notable Systems: Key projects include the Barcelona Supercomputing Center's EuroHPC AI Factory, BavariaAI's Blue Swan, IT4LIA (Czech Republic), HLRS's HammerHAI (Germany), and NAISS's Mimer EuroHPC AI Factory (Sweden) .
• Applications: The systems will support work in climate science, healthcare, clean-energy decarbonization, and industrial digital twins .

Partner Announcements and Ecosystem

Several partner announcements were highlighted at or around ISC 2026, including those from Siemens Energy, HPE, and Dell.

• Siemens Energy + HPE: Siemens Energy is deploying a dedicated global HPC platform through HPE GreenLake for engineering-driven AI. This includes work on hydrogen-capable gas turbine burners, cited as a key clean-energy initiative in NVIDIA's press materials .
• HPE: HPE announced one of the first NVIDIA Vera CPU compute blades (the HPE Cray Supercomputing GX240, available in 2027) and the HPE Compute XD700 server built on the NVIDIA HGX Rubin NVL8 platform. The Vera Rubin NVL72 by HPE is expected to be available in December 2026 .
• Dell: At GTC 2026 (March), Dell introduced a Data Orchestration Engine and new high-performance storage solutions for AI data centers. No new Dell-specific announcement was made at ISC 2026, but Dell remains a key system integrator for the Rubin platform and European supercomputing deployments .

Summary of Key Announcements