Europe has placed a major bet on its own technological sovereignty with the order of a €387.8 million (roughly $449 million) AI supercomputer destined for Kajaani, northern Finland. The system, named LUMI-AI, will be built by Bull, the French supercomputer manufacturer that the French state acquired outright five months ago. The contract marks the sixth AI Factory procurement signed by the EuroHPC Joint Undertaking, the coordinating body behind Europe's publicly funded supercomputing infrastructure.
The decision to buy from a state-owned supplier is the clearest signal yet of how much Europe's strategic approach has shifted. Only a few years ago, government ownership of a major supercomputer vendor would have been an unusual detail in any procurement story. Now France owns the manufacturer, six governments are helping finance the machine, and the accelerators at its core are still made by an American company. That combination reflects both the progress Europe has made toward technological sovereignty and the parts of the supply chain where dependence on foreign suppliers remains unavoidable.
Sharing the cost and the capability
The cost of LUMI-AI is being split evenly between EuroHPC and the LUMI AI Factory consortium, a grouping that includes Finland, Czechia, Denmark, Estonia, Norway, and Poland. Each partner government is contributing to a machine that will sit on Finnish soil but will be available to researchers and companies across the entire consortium. This pooling of resources has become a defining feature of Europe's AI infrastructure strategy, especially as individual nations struggle to match the spending power of the world's largest private technology companies.
In practical terms, the new system is expected to deliver roughly ten times the AI capacity of the existing LUMI supercomputer, while nearly doubling its conventional high-performance computing capability. Deployment is scheduled for the second half of 2027. That timeline may seem distant for a field moving as quickly as AI, but it is fairly standard for a major supercomputer procurement, where design, integration, testing, and installation typically stretch over years.
Hardware with a European identity
The hardware choices reveal more than the headline price tag. LUMI-AI will be built around AMD Instinct MI430X accelerators and sixth-generation EPYC processors with 256 cores, housed in Bull's BullSequana XH3500 architecture. While the chips are American-designed, two crucial system components give the project a stronger European identity than merely assembling a US-designed machine on European soil.
Bull is supplying its BXI interconnect, which links the nodes together, as well as its patented warm-water direct liquid cooling technology. Both were developed in France and represent some of the most advanced cooling and fabric technologies available in the market today. The wider supplier list also reflects a deliberate effort to keep more of the infrastructure within Europe. IBM Storage Scale will provide the storage layer, while Nokia is supplying the data centre networking, bringing a Finnish company into a system that will be installed in Finland.
Emmanuel Le Roux, Bull's chief executive, described the project as a milestone for European AI and for Bull's leadership in energy-efficient, complex AI infrastructures. Anders Jensen, who runs EuroHPC, framed the purchase as the sixth step in a broader sequence of AI infrastructure investments rather than an isolated project, underscoring the coordinated nature of Europe's public sector push into high-performance computing.
The strategic importance of Bull
Bull's ownership is central to understanding the deal's significance. Atos completed the sale of its Advanced Computing business to the French state on 31 March, in a transaction with an enterprise value of up to €404 million, including €104 million in earn-outs. This made the French government the company's sole shareholder. The business generated around €700 million in revenue in 2025 and operates Europe's only supercomputer manufacturing plant, located in Angers.
Beyond commercial supercomputing, Bull also builds the systems used to model France's nuclear deterrent. That gives the business a strategic importance that goes well beyond market share, and helps explain why the French government was unwilling to see it fall into foreign ownership. The Angers plant is not merely a factory; it is a key piece of national and European technological infrastructure, capable of producing systems that underpin both civilian research and national security.
Kajaani and Finland's natural advantages
Kajaani, a small city in central Finland, might seem an unlikely hub for cutting-edge AI infrastructure, but it has distinct advantages that many potential European sites cannot easily replicate. Finland enjoys abundant, relatively cheap, low-carbon electricity and a cold climate that makes cooling large computing installations considerably easier. Over the years, the country has invested heavily in the infrastructure needed to turn these conditions into a competitive edge for data centres and supercomputing.
Kimmo Koski, who runs CSC, the Finnish research organisation that will host LUMI-AI, described the new system as a continuation of what the existing LUMI consortium has already built rather than a project starting from scratch. That existing track record is part of the reason six countries are prepared to pool funding around a facility located in a country that only one of them actually hosts. The original LUMI system has become one of Europe's most established large-scale supercomputers, and with Germany's Jupiter now joining the landscape, LUMI remains a flagship of the continent's high-performance computing ecosystem.
The broader European AI infrastructure mosaic
The LUMI-AI order fits into a pattern that has been taking shape across Europe for the past two years. EuroHPC selected its first seven AI Factory sites in December 2024, followed by six more in October 2025, and procurement contracts have been signed progressively since then. These AI Factories are designed to provide European researchers and startups with access to the massive computing power required to train and run advanced AI models.
The gigafactories were supposed to take that effort to another scale, with a €30 billion programme covering seven much larger sites. Progress on that part of the strategy has been slower, however. The gigafactory programme has been struggling with delays that have frustrated some of the partners it needs, while the projects themselves are considerably more complicated than building an individual supercomputer at an established site. The AI Factories now being procured remain relatively small compared with what the biggest private technology companies are building. A €387.8 million supercomputer represents only a fraction of the investment behind a single hyperscaler campus, so Europe's response to that imbalance has so far relied heavily on pooling resources between governments rather than trying to match the private sector project for project.
Dependence and sovereignty in the AI stack
One of the most telling aspects of the LUMI-AI procurement is that it relies on AMD for the core accelerators. Even as Europe works to build a more self-sufficient AI ecosystem, the cutting-edge hardware at the heart of most AI supercomputers still comes from a small number of American firms. The European Commission has recognised this gap and has launched several initiatives to foster domestic chip design and manufacturing, but those efforts are still in their early stages. The reality is that for now, any European AI supercomputer will depend on US-made processors and GPUs, even if the surrounding infrastructure is increasingly European.
The decision to buy from Bull, a state-owned champion, is a bold statement of intent. It signals that European governments are willing to use public procurement to support strategic industrial capabilities. It also creates an interesting dynamic: a company owned by one EU member state is now supplying a system co-funded by several others. This could serve as a model for future pan-European procurement, or it could raise questions about how to ensure fair competition and access across the bloc.
For the researchers and companies that will eventually use LUMI-AI, the ownership structure matters less than the machine's performance. The combination of AMD's latest accelerators, Bull's interconnect and cooling, IBM's storage, and Nokia's networking is designed to deliver a system that can handle some of the most demanding AI workloads in the world.
The contract is now signed, the site is established, and the hardware has been chosen. The real test will come when LUMI-AI is deployed in the second half of 2027, when Europe will find out how much of its AI infrastructure strategy can translate from coordinated procurement into computing capacity that researchers and companies can actually use.