Skip to main content
Back to AI NewsNews

Bull Selected to Build Europe's 387.8 Million Euro LUMI-AI Supercomputer

French computing firm Bull will deliver a 387.8 million euro AI supercomputer in Finland for EuroHPC.

cueball EditorialMonday, 31 August 2026 3 min read

What Happened

French advanced computing company Bull has been selected by the EuroHPC Joint Undertaking and the LUMI AI Factory to design and deliver Europe's LUMI-AI supercomputer, to be hosted in Finland. The contract is valued at 387.8 million euros, making it one of the largest single supercomputing procurement awards in European history.

The announcement was made on August 31, 2026, from Paris.

Who Is Involved

Bull is a French technology company specialising in high-performance computing and AI infrastructure. The EuroHPC Joint Undertaking is a European Union body established to coordinate supercomputing investment and capability across member states. The LUMI AI Factory is an initiative connected to the existing LUMI supercomputer, which is operated by CSC, the Finnish IT center for science, and has previously ranked among the most powerful supercomputers in the world.

Finland serves as the host country for the LUMI infrastructure, located at CSC's facilities in Kajaani.

Background

The EuroHPC Joint Undertaking was created in 2018 to reduce European dependence on non-European supercomputing infrastructure and to build competitive capacity for research, industry, and public sector AI workloads. The original LUMI system, procured as part of that initiative, was built using AMD processors and became operational in 2022. At peak, it ranked among the top five fastest supercomputers globally.

The LUMI AI Factory designation refers to a next-generation extension of that infrastructure, specifically oriented toward large-scale AI model training and inference workloads. AI factories, as defined by the European Commission, are supercomputing environments configured to support the development and deployment of foundational AI models at scale.

Bull, which operates under the Eviden brand within the Atos group, has been a longstanding supplier to European public-sector high-performance computing programs. The company has previously delivered systems to the French Alternative Energies and Atomic Energy Commission and other European research institutions.

The Contract and System Details

The total contract value is 387.8 million euros. The wire report does not provide a detailed breakdown of system specifications, including processor architecture, memory capacity, or peak compute performance measured in exaflops or petaflops. Those details had not been publicly released as of the announcement date.

The selection was made through a competitive procurement process administered by EuroHPC JU. The announcement did not name other bidders or provide terms of the selection criteria.

The system will be installed in Finland, consistent with the existing LUMI hosting arrangement.

What It Means in Practice

The LUMI-AI supercomputer is intended to provide shared computing capacity for AI research and development across EuroHPC member states. Researchers and organisations from participating countries will be able to apply for access to the system through EuroHPC allocation processes, as they do with existing EuroHPC machines.

The scale of the investment reflects the European Union's continued emphasis on building sovereign AI infrastructure, an objective outlined in the EU AI Act and related policy frameworks. The Commission has previously stated a goal of deploying AI factories across multiple member states to enable training of large AI models within European jurisdiction, subject to European data governance rules.

No timeline for system delivery or operational launch was included in the wire report issued on August 31, 2026.

What Comes Next

Bull and EuroHPC JU are expected to publish further technical specifications and a project delivery schedule in subsequent official communications as the procurement agreement is formalised.

Get our editors' take on what it all means. Read the Editor's Blog →