Public sector body launches tender for AI supercomputer platform
Contract notice seeks a supplier to deliver, install and maintain an AI-focused supercomputer, highlighting growing public demand for high-performance computing.
Contract notice seeks a supplier to deliver, install and maintain an AI-focused supercomputer, highlighting growing public demand for high-performance computing.
Follow Tenderlake on LinkedIn for concise insights on public-sector tenders and emerging procurement signals.
The European High-Performance Computing Joint Undertaking has gone to market for a new AI-focused supercomputer to anchor an AI Factory, signalling how shared high-performance computing for artificial intelligence is becoming core public infrastructure across research and government.
Published on 21st May 2026, the AI Supercomputer Acquisition and Maintenance contract notice sets out plans to select a contractor to acquire, deliver, install and maintain a high-performance AI supercomputer. The system is intended to support an AI Factory, indicating that the buyer sees it as shared infrastructure within that environment.
Although the summary text does not give detailed technical specifications, it is explicit about the scope of services. The contract covers the full lifecycle of the installation, from delivery and set-up through to ongoing hardware and software maintenance. The buyer also asks for proven experience in similar environments and advanced computing capabilities, pointing to a requirement for suppliers with track records in large-scale high-performance and AI deployments.
That combination of AI-specific performance and supercomputing scale sets this project apart from many smaller AI hardware procurements now appearing across the public and research sectors. It aligns the AI Factory with a growing group of facilities where artificial intelligence workloads are treated as mainstream high-performance computing tasks, rather than experimental sidelines.
Recent notices show public bodies procuring AI infrastructure at very different scales. At one end, some contracts focus on a handful of servers. The AI GPU Servers tender from Ústav organické chemie a biochemie AV ČR, published in December 2025, is for two AI GPU servers plus accessories, including transport, installation and full operational set-up. The April 2026 AI Server Delivery and Support notice from Szkoła Główna Gospodarstwa Wiejskiego w Warszawie similarly concentrates on a single AI server, installation, configuration and support.
A step up from that are tenders such as the March 2026 High-Performance Computing Server Acquisition for Instituto de Geografia e Ordenamento do Território, which seeks a single high-performance computing server for HPC/AI workloads, and the March 2026 Computer Equipment Delivery contract for Wojskowa Akademia Techniczna im. Jarosława Dąbrowskiego, which combines GPU/AI computing servers with network disk arrays.
At the other end of the spectrum sit supercomputer-scale projects. In March 2026, University of Galway (ID 1400) launched the CASPIr Supercomputer Hardware and Software procurement to acquire, deliver, install and maintain hardware and software for the CASPIr supercomputer, supporting advanced computing services for Irish and European researchers. In February 2026, Dirección General de la Fundación Computación y Tecnologías Avanzadas de Extremadura went to market with its Infrastructure and Equipment Supply contract to expand storage, acquire a GPU cluster and grow a distributed memory computing cluster as part of the LUSITANIA supercomputer upgrade.
The HPC Datacenter Supply notice issued by SURF B.V. in December 2025 – for a high performance computing datacentre facility for an AI supercomputer – and the January 2026 AI and Computing Servers Supply contract from the State Agency for the Spanish National Research Council, which extends the Drago supercomputer, underscore this trend towards full-scale AI-ready supercomputing infrastructure. The AI Factory tender now joins this small but strategically important group.
Many universities and research institutes are also ramping up AI-specific compute. In December 2025, Fraunhofer IIS published a HPC and AI Cluster Components procurement for a directly water-cooled GPU computing cluster to train generative AI, including the required infrastructure and maintenance services. Around the same time, Vilniaus Gedimino technikos universitetas set out plans in its December 2025 Data Processing Platform Expansion contract to add CPU computing subsystems, GPU computing nodes, dedicated AI computing nodes and data storage infrastructure to an existing high-performance platform.
Early 2026 has seen a string of similar moves. Libera Università di Bolzano’s January 2026 NVIDIA GPU Server Procurement aims to acquire an NVIDIA GPU platform with CUDA support for high-performance computing in research projects and AI training at its Ai-Lab. Uniwersytet w Białymstoku’s January 2026 Computing Cluster Delivery for Research focuses on delivering a computing cluster for data analysis and AI, while Università degli Studi di Foggia’s February 2026 AI/HPC Infrastructure Supply notice bundles GPU-based compute, storage, networking and orchestration software for its Intelligent Technology Transfer project.
Other research-focused buyers include Idryma Technologias & Erevnas (ITE) in its May 2026 HPC and AI Research Infrastructure tender, which combines high-performance servers, user access servers, a storage server and a high-speed network with a five-year operational guarantee, and HRVATSKA AKADEMSKA I ISTRAŽIVAČKA MREŽA CARNet’s May 2026 High-Performance Computing Platform notice for a local platform to support compute-intensive operations in scientific, educational and infrastructural contexts.
Semmelweis Egyetem’s National Medical Innovation Training Center adds a medical research angle with its November 2025 Artificial Intelligence Server Procurement, combining acquisition of an NVIDIA-compatible AI server with delivery, installation, training and a comprehensive warranty. Together, these projects show universities and research networks converging on similar design choices, even when their disciplines differ.
Beyond the research sector, AI compute is moving into a wide range of public and quasi-public organisations. Land Baden-Württemberg’s police technology, logistics and service body, for example, is procuring a GPU server for training AI models under its December 2025 GPU Server Procurement for AI notice, with explicit requirements around data protection regulations and ethical standards.
The business community appears in the April 2026 AI Computing Infrastructure for Ceuta tender from Cámara Oficial de Comercio, Industria, Servicios y Navegación de Ceuta, which will implement a computing centre for artificial intelligence at the Chamber of Commerce of Ceuta. That contract covers hardware, storage, networking, software and related installation and support services, echoing the full-stack approach seen in many research procurements.
Education and skills provision also feature. In February 2026, East Durham College published its Infrastructure Refresh Proposal to secure a comprehensive refresh of infrastructure and end user computing equipment under a three-year lease, with an emphasis on meeting public sector standards, sustainability objectives and enhanced warranty and security features. While not focused solely on AI, that notice shows how colleges are upgrading the surrounding infrastructure that AI workloads depend on.
Specialised AI systems for healthcare and life sciences appear in MEDIAN Unternehmensgruppe B.V. & Co.KG’s December 2025 AI Hardware tender for a scalable GPU compute platform to enhance AI-supported medical data processing, and Universitatea Craiova’s April 2026 Procurement of AI Server and UPS for a parallel-processing AI server with GPU acceleration and UPS to support a deep learning project on fetal morphology. These contrast with the AI Factory supercomputer, which is not tied in the notice to a single domain.
Finally, the national-level January 2026 AI Platform Delivery and Integration prior information notice from Národní agentura pro komunikační a informační technologie sketches a hardware platform that brings together computing power, data storage and network infrastructure for AI systems, designed to integrate with both cloud and on-premise environments. It sits conceptually close to the AI Factory idea: a platform approach rather than a standalone server.
Across this wave of AI and high-performance computing tenders, several themes recur. First is scope: most buyers are procuring not just equipment but end-to-end solutions. The AI Factory supercomputer contract wraps acquisition, delivery, installation and maintenance into a single package. So do notices such as Ceuta’s AI computing centre, Foggia’s AI/HPC infrastructure, University of Galway’s CASPIr supercomputer and Semmelweis Egyetem’s AI server, all of which combine hardware with installation, configuration, training or ongoing support.
A second theme is convergence between AI and traditional high-performance computing. Many of the projects – from Fraunhofer IIS’s water-cooled GPU cluster and CARNet’s local high-performance platform to the LUSITANIA and Drago supercomputers – explicitly mix AI workloads with broader scientific or data-intensive computing. The AI Factory tender fits that blended model by specifying both high-performance and AI capabilities in its supercomputer.
Governance and resilience also stand out. Land Baden-Württemberg’s police notice spells out data protection and ethical requirements. East Durham College builds sustainability objectives and security features into its infrastructure refresh. Idryma Technologias & Erevnas requires a five-year operational guarantee, and many other buyers emphasise extended warranties. Against this backdrop, the AI Factory contract’s demand for proven experience in similar environments signals a preference for suppliers able to demonstrate robust operations over time, not just theoretical performance.
The summary information available in the AI Factory supercomputer notice does not detail technical specifications, budgets or timelines, so the precise scale of the investment is not clear from the published text alone. Even so, its framing as a high-performance AI supercomputer for an AI Factory places it among a small cluster of flagship AI and high-performance computing infrastructure projects now moving through procurement.
With institutions ranging from universities and research councils to police services, chambers of commerce and national technology agencies all investing in AI-capable compute, the European High-Performance Computing Joint Undertaking’s tender is a notable addition. It will be important to see which suppliers can evidence the kind of proven experience in similar environments that the buyer is seeking, and how this new system will sit alongside supercomputers such as CASPIr, LUSITANIA and Drago as they come online.
Follow Tenderlake on LinkedIn for concise insights on public-sector tenders and emerging procurement signals.