News
DealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanSovereignNeysa Deploys AMD GPU Cluster in Mumbai for AI Training and InferenceInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to NvidiaDealsCorvex agrees USD 103 million take-or-pay deal to perform AI inference with Mid-Atlantic GPUsSovereignNeysa scales AI infrastructure in India with AMD Instinct MI350X GPUs on TytanSovereignNeysa Deploys AMD GPU Cluster in Mumbai for AI Training and InferenceInferenceMicrosoft Foundry adds Baseten, DeepSeek AI inference (MSFT:NASDAQ)ClustersDataOne and Rowan University sign engineering co-op program agreement for DataOne's Vineland AI factoryClustersLook inside Cal Poly's $3 million AI Factory (Video)DealsOpenAI Taps AMD Chips for 457 MW of AI Capacity in Six-Year TensorWave DealClustersCal Poly launches an AI data center on campus. Take a look insideInferenceArista takes aim at the next AI network battleground: distributed inferenceDealsOpenAI Signs 457 MW AI Data Center Deal as AMD Expands Its Challenge to Nvidia
EveryNeocloud
EuroHPC Joint Undertaking

EuroHPC Joint Undertaking

EU public-private partnership building a pan-European supercomputing and AI infrastructure

EuroHPC JU is a public-private partnership pooling EU, national and private money to build a pan-European supercomputing infrastructure and back research and innovation. It began operating in November 2018, became autonomous from the European Commission in 2020, and runs a fleet of supercomputers plus a growing set of AI factories across member states.

Share on WhatsApp

  • Budget of around €7 billion for 2021-2027, matched by participating countries
  • LUMI measured 151.9 petaflops on HPL, third on the June 2022 TOP500 list
  • AI gigafactories planned with at least 100,000 GPUs in each location
  • 19 standard AI factories in Europe as of March 2025, up to 25,000 GPUs each
  • Became autonomous from the European Commission on 23 September 2020

What it is

EuroHPC JU is a joint undertaking of the European Union, set up as a public-private partnership in high-performance computing. It combines money and resources from the EU, participating member states, countries associated to Horizon Europe and Digital Europe, and private bodies. Its two aims are to build a shared supercomputing infrastructure for Europe and to support research and innovation around it. Headquartered in Luxembourg City, it buys, deploys and operates supercomputers and AI systems, then opens them to scientific and industrial users, including small firms. It also funds research grants and works on European processor technology.

How it came about

EU leaders called for closer coordination on high-performance computing in June 2016 under the Digital Single Market strategy. A European Declaration on the subject was signed in Rome in March 2017 by seven states: France, Germany, Italy, Luxembourg, the Netherlands, Portugal and Spain. The Commission proposal was endorsed by the Council in June 2018, backed by the European Parliament on 3 July 2018 and formally adopted on 28 September 2018. Operations began in November 2018. An Executive Director was appointed on 15 May 2020 and the body became autonomous from the Commission on 23 September 2020. Council Regulation (EU) 2021/1173 later reviewed it.

The compute

The fleet mixes machines from several suppliers. LUMI in Finland is an HPE Cray EX system, while Vega, MeluXina, Discoverer and Leonardo were delivered by Atos, and Deucalion combines Fujitsu PRIMEHPC ARM hardware with Atos x86 partitions. Performance ranges from petascale systems of a few petaflops up to LUMI, which measured 151.9 petaflops on HPL and has a theoretical peak of 550 petaflops, and Leonardo at over 250 petaflops. EuroHPC also runs the European Processor Initiative, whose SiPearl Rhea1 chip is destined for JUPITER, and the DARE project on RISC-V designs.

Where the GPUs are

In June 2019 the governing board picked eight host sites: Sofia, Ostrava, Kajaani, Bologna, Bissen, Minho, Maribor and Barcelona. Three were earmarked for pre-exascale machines above 150 petaflops and five for petascale systems of at least 4 petaflops. Five more sites followed in 2022, including Germany's JUPITER, and two further ones in 2023. From December 2024, locations were chosen for AI data centres, with more added in March 2025, bringing the count of standard AI factories to 19. Hosting partners include CSC in Kajaani, CINECA in Bologna, the Barcelona Supercomputing Center and LuxProvide in Bissen.

Customers and deals

Access is aimed at European scientific and industrial users across more than 800 application fields, with small and medium-sized enterprises explicitly included. Support is handed out through open, competitive calls, either as procurement or as research and innovation grants. Machines serve workloads such as numerical simulation, data analysis and artificial intelligence. The United Kingdom lost observer standing when it left the EU on 31 January 2020 but rejoined EuroHPC in May 2024.

Money and ownership

Members fund the undertaking jointly, with a budget of roughly €7 billion for 2021-2027. Of that, €3 billion comes from the EU's Multiannual Financial Framework: €1.9 billion via the Digital Europe Programme for infrastructure and services, €900 million via Horizon Europe for research, and €200 million via Connecting Europe Facility for interconnections. Participating countries match the EU share, and private members add €900 million. Separately, the InvestAI initiative announced in February 2025 carries a headline budget of €200 billion, including a €20 billion fund for data centres.

Its place in AI

EuroHPC JU sits at the centre of European efforts to hold its own computing capacity. Its machines have placed highly in global rankings: LUMI was third on the June 2022 TOP500 list, KAROLINA ranked 69th worldwide in 2021 and eighth on the Green500 efficiency list. Through the European Processor Initiative it fostered the SiPearl Rhea1 processor and the EPAC accelerator test chips, and the DARE project pushes RISC-V designs for accelerators, inference and general-purpose HPC. Its AI gigafactory plan targets facilities of at least 100,000 GPUs each.

Questions and criticism

Announcements around InvestAI have drawn questions about whether the promised scale will be delivered. The programme's ambitions also rest on large amounts of public funding and on matching contributions from participating states, alongside private money.

Questions people ask about EuroHPC Joint Undertaking

What is EuroHPC JU?

A public-private partnership set up by the European Union to build shared supercomputing infrastructure and support research. It pools EU budget money, contributions from participating states and funds from private members, and it buys and runs supercomputers open to European users.

Where is it based?

Its headquarters are in the Drosbach Building on rue Guillaume Kroll in Gasperich, Luxembourg City. The machines themselves are spread across member states, from Kajaani in Finland to Barcelona in Spain and Guimarães in Portugal.

How is it funded?

Members contribute jointly, with a budget of about €7 billion for 2021-2027. €3 billion comes from the EU long-term budget through the Digital Europe Programme, Horizon Europe and the Connecting Europe Facility, matched by participating countries, plus €900 million from private members.

What are the AI factories?

Data centres for artificial intelligence selected from December 2024 onwards. Standard sites hold up to 25,000 GPUs, and the plan includes up to five larger gigafactories with at least 100,000 GPUs each, backed by the InvestAI initiative announced in February 2025.

Who can use the supercomputers?

Scientific and industrial users across more than 800 application fields, including small and medium-sized enterprises. Access is allocated through open, competitive calls as either procurement or research and innovation grants.

Good reading and sources

Where this page comes from

The facts on this page come from the Wikipedia article "European High-Performance Computing Joint Undertaking", written by its contributors and used under the CC BY-SA 4.0 licence, and from its Wikidata record (CC0). EveryNeocloud rewrote the text in its own words, with AI assistance, and compares every page with the article to make sure no passage is copied. It may be reused under the same CC BY-SA 4.0 licence with a link back.

It may contain mistakes, and figures change over time. Nothing here is investment or buying advice. If something is wrong, tell us. Last refreshed 2026-11-14.

Photo credits on this page

The small pictures above come from Wikimedia Commons under free licences. Some are cropped to fit.

The weekly digest

Every Monday: the provider of the day, new pages on EveryNeocloud, and the conferences and anniversaries of the week ahead.