# Nvidia: 35 AI Supercomputers Under Way in Europe > Source: [https://botensten.com/articles/nvidia-35-ai-supercomputers-europe](https://botensten.com/articles/nvidia-35-ai-supercomputers-europe) (canonical) > Author: iCharles News — Botensten, https://botensten.com > Published: 2026-06-22 ## TL;DR Nvidia announced on June 22, 2026, that 35 AI supercomputers are under development across 23 European countries. The systems run on Nvidia's Blackwell and Hopper platforms and will serve more than 3 million researchers. Key installations include Germany's first AI factory HammerHAI, Barcelona's MareNostrum5 upgrade, BavariaAI's Blue Swan, and Italy's IT4LIA. Nvidia says its infrastructure now powers over 90% of Europe's AI factory buildout, with 800 AI exaflops deployed or announced since last year. ## What did Nvidia announce at ISC High Performance 2026? Nvidia announced that 35 AI supercomputers are under development across Europe, spanning 23 countries. The announcement came on June 22, 2026, from Hamburg, Germany, where Nvidia also launched its Vera Rubin platform at ISC High Performance 2026. The systems are installed at national supercomputing centers, AI factories, and academic research institutions. The company says its infrastructure now powers over 90% of Europe's AI factory buildout. Since last year, 800 AI exaflops have been deployed or announced across the continent, [according to Nvidia's press release](https://investor.nvidia.com/news/press-release-details/2025/NVIDIA-Powers-Europes-Fastest-Supercomputer/default.aspx). ## Which hardware platforms power these systems? All 35 systems are built on Nvidia's **Blackwell** and **Hopper** GPU platforms. Blackwell and Hopper are Nvidia's data center GPU architectures designed for large-scale AI training and inference workloads. Together, the installations will serve more than 3 million researchers across Europe. Here's what we know so far about the scale of each major installation: | System | Location | GPUs | AI Training | AI Inference | |---|---|---|---|---| | IT4LIA | Italy | 8,000+ | 82 exaflops | 164 exaflops | | MareNostrum5 upgrade | Barcelona | — | ~20 exaflops | ~33 exaflops | | BavariaAI Blue Swan | FAU Erlangen / LRZ | 1,000 | Up to 11 exaflops | Up to 22 exaflops | | HammerHAI | Germany | 850+ | ~8 exaflops | ~15 exaflops | ## What is HammerHAI and why does it matter? **HammerHAI** is Germany's first AI factory, deployed by HLRS (High Performance Computing Center Stuttgart). It will run over 850 GPUs and deliver approximately 8 exaflops of AI training performance and 15 exaflops of AI inference performance. The EuroHPC Joint Undertaking signed the contract to deploy HammerHAI in March 2026, [as reported by EuroHPC JU](https://www.eurohpc-ju.europa.eu/eurohpc-ju-signs-contract-deploy-ai-supercomputer-hammerhai-2026-03-16_en). HammerHAI is part of the broader EuroHPC initiative to build AI computing capacity within Europe. The system is intended to support European digital sovereignty and research innovation. ## What is MareNostrum5 getting upgraded with? Barcelona Supercomputing Center will expand its MareNostrum 5 system with Nvidia **GB300 NVL72** and **GB200 NVL4** hardware. The upgrade will deliver approximately 20 exaflops of AI training performance and 33 exaflops of AI inference performance. MareNostrum 5 is one of Europe's flagship supercomputing systems. This upgrade is part of Nvidia's broader [Blackwell platform expansion](/articles/nvidia-coherent-texas-factory-groundbreaking) across its global infrastructure buildout. ## What will IT4LIA be used for? IT4LIA is Italy's entry in the buildout and is the largest single system announced by GPU count. With over 8,000 GPUs, it will deliver 82 exaflops of AI training and 164 exaflops of AI inference performance. The system will support applications across agritech, cybersecurity, meteorology, climate science, and manufacturing. ## What research areas will these supercomputers support? Nvidia says the systems will support research in: - Climate science and clean-energy decarbonization - Healthcare - Quantum computing - Agritech, cybersecurity, and meteorology (IT4LIA specifically) On the climate side, Siemens Energy is using Nvidia technologies to develop hydrogen-capable gas turbine burners. Nvidia says this has reduced simulation times by up to 77%. On quantum computing, Barcelona Supercomputing Center, CINECA, Fraunhofer, and Jülich Supercomputing Centre are using Nvidia's **CUDA-Q** platform. CUDA-Q is Nvidia's platform for integrating quantum processors with GPU systems. ## What else did Nvidia announce at ISC High Performance 2026? Alongside the European supercomputer rollout, Nvidia introduced two other products at the Hamburg event. - **Vera Rubin platform** — designed for scientific computing workloads, with over 7 exaflops of AI performance. Developers building on [Nvidia's GPU roadmap](/articles/nvidia-vera-cpu-china-sales-2) will want to track this platform closely. - **Halos for Robotics** — a system for safety in robotics and physical AI, combining Nvidia IGX Thor hardware with the Halos OS software stack. Nvidia also announced a partnership with Fervo Energy and Pacific Northwest National Laboratory to build a digital twin platform for Enhanced Geothermal Systems technology. Separately, Equinix expanded its collaboration with Cisco and Nvidia to deploy AI infrastructure across global data centers through the Cisco Secure AI Factory with Nvidia. ## How does this fit Nvidia's broader chip strategy? Nvidia's European push comes as the company faces scrutiny over [AI chip export controls](/articles/asml-euv-china-export-concern) in other markets. The European buildout, by contrast, represents open deployment at scale. Nvidia reported $253 billion in revenue over the last twelve months, growing 71% year-over-year, [according to Investing.com](https://uk.investing.com/news/stock-market-news/nvidia-announces-35-ai-supercomputers-under-development-in-europe-93CH-4738022). The company's market capitalization stands at $5.1 trillion. Builders and founders tracking [AI infrastructure investment](/articles/amazon-trainium-chips-third-party-sales) should note that Nvidia claims to power more than 90% of Europe's AI factory buildout by exaflops. The EuroHPC JU contract for HammerHAI was signed in March 2026, with deployment planned for 2026. ## Frequently asked questions ****How many AI supercomputers is Nvidia building in Europe?**** Nvidia announced that 35 AI supercomputers are under development across Europe as of June 22, 2026. The systems span 23 countries and are installed at national supercomputing centers, AI factories, and academic research institutions. They are all built on Nvidia's Blackwell and Hopper GPU platforms and will collectively serve more than 3 million researchers. ****What is HammerHAI?**** HammerHAI is Germany's first AI factory, operated by HLRS. It will deploy over 850 GPUs and deliver approximately 8 exaflops of AI training and 15 exaflops of AI inference performance. The EuroHPC Joint Undertaking signed the contract to deploy HammerHAI in March 2026, with the system planned for deployment in 2026. ****What GPU platforms are used in Nvidia's European AI supercomputers?**** All 35 European AI supercomputers announced by Nvidia are built on its Blackwell and Hopper GPU platforms. Specific hardware mentioned includes the GB300 NVL72 and GB200 NVL4 systems, which will be used in the MareNostrum5 upgrade at Barcelona Supercomputing Center. ****How many exaflops has Nvidia deployed in Europe?**** Nvidia says 800 AI exaflops have been deployed or announced across Europe since last year. The company also claims its infrastructure powers over 90% of Europe's AI factory buildout. The largest single system, IT4LIA in Italy, alone delivers 82 exaflops of AI training and 164 exaflops of AI inference performance. ****What research will Europe's new AI supercomputers support?**** The systems will support research in climate science, healthcare, clean-energy decarbonization, and quantum computing. IT4LIA specifically targets agritech, cybersecurity, meteorology, and manufacturing. Siemens Energy is using Nvidia technologies for hydrogen-capable gas turbine burner development, reducing simulation times by up to 77%, according to Nvidia's announcement.