Berlin, October 21th, 2026
Bristol Centre for Supercomputing (BriCS) and GENCI today announced a joint federated learning plan, developed under the UK-France ConcordAI partnership, to connect their national AI supercomputers, enabling researchers in both countries to train larger and more capable AI models together. The announcement was made at the GTC Berlin technology conference, with technical support from NVIDIA.
The partnership brings together three national supercomputers to start with, the UK's Isambard-AI, based at the University of Bristol and powered by NVIDIA Grace Hopper Superchips, and France's Jean Zay and Adastra supercomputers. The project is designed as a genuine two-way partnership, pooling computing power so that both countries can do more together than either could alone, while each keeps full control of its own machines and data.
For the UK, this means access to a larger pool of AI computing power to support British research and start-ups. For France, it means the same in return, plus the chance to combine two supercomputers built on different hardware, as Jean Zay runs on NVIDIA accelerated computing, whilst Adastra runs on other hardware, proving that this approach can work even when the underlying technology differs from site to site.
Federated learning is a technique where each supercomputer trains its own copy of an AI model on the data held locally at that site, and only the 'lessons learned' are shared between sites, instead of copying vast amounts of data between countries. It could be compared to several students each working through their own notes, then comparing what they've learned, rather than sending the notes to each other.
This means sensitive data in fields like healthcare, defence, or industries such as aerospace, never has to leave the country, where it originates. That opens the door to joint AI research in areas where data protection rules or commercial sensitivity would otherwise make collaboration very difficult, while still letting both nations benefit from shared computing power and expertise.
Federated learning itself isn't new but new open-source tools, such as NVIDIA FLARE, have made it far more practical to use at the scale of national supercomputers. With NVIDIA's support, teams within the UK, at Bristol and Cambridge, and within France, at the IDRIS and CINES centres, have already been testing the approach by gradually increasing the size and complexity of the AI models involved and checking that results match models trained on a single machine.
Once this national testing phase is complete, BriCS and GENCI will move to connect their supercomputers directly with one another, targeted for completion by the end of 2026.
This UK-France partnership sits alongside similar thinking elsewhere in Europe. Germany's Forschungszentrum Jülich (FZJ) and GENCI are separately interested in connecting Jupiter AI Factory and AI France Factory compute infrastructure, starting with JUPITER at Jülich Supercomputing Center and Jean Zay/Adastra at the beginning and later with Alice Recoque at GENCI to give European researchers early access to the computing power needed for major AI-for-Science projects.
In France, GENCI is also planning to connect its supercomputers with those of French hyperscalers, creating a more direct route from public research computing to commercial use for French start-ups and industry. BriCS and GENCI see this kind of public-private link as part of a broader effort to make advanced AI computing power more accessible to businesses on both sides of the Channel.
BriCS and GENCI describe this as a first step toward a much closer AI computing relationship between the UK and France and say they hope to extend the approach to more computing centres and partners over time, building AI capability that benefits researchers, businesses and the public in both countries.
Quotes
- Prof. Simon McIntosh-Smith, Director at BriCS, said, ‘This joint venture is about the UK and France doing more together than either of us could do alone, without compromising on data security. It's an exciting moment for AI research on both sides of the Channel, with a realistic prospect of making breakthroughs that could benefit the world beyond our countries.’
- Prof. Michael Krajecki, CEO of GENCI, said “By federating our AI supercomputers using open-source tools like NVIDIA FLARE UK and France are pioneering at European scale new AI services for scaling the ambitions of scientific and industrial AI communities. This is a first step toward engaging with additional public HPC centers as well as private partners, paving the path to joint services operated by AI Factories and future AI GigaFactories”.
- “Connecting national supercomputing capabilities will help researchers tackle scientific challenges beyond the reach of any one institution,” said John Josephakis Global VP HPC/Supercomputing at NVIDIA. “With NVIDIA FLARE, BriCS and GENCI are laying the groundwork to train models collaboratively across borders while keeping sensitive data within each country.”
____
About BriCS
The Bristol Centre for Supercomputing (BriCS), part of the University of Bristol, is a leading UK host for advanced high-performance computing and AI infrastructure. BriCS built and operates Isambard-AI, the UK's most powerful AI supercomputer and a cornerstone of the government funded AI Research Resource (AIRR), enabling breakthroughs in areas such as drug discovery, climate research, and robotics across academia and the private sector. BriCS also built and manage Isambard 3, one of the UK's most powerful CPU-based systems. Combining deep HPC expertise with rapid, modular deployment capability, BriCS positions Bristol and the UK as an international hub for AI-driven scientific discovery.
Website | YouTube | LinkedIn | Instagram | Bluesky
About GENCI
Created by the public authorities in 2007, GENCI (Grand Équipement National de Calcul Intensif) is a major research infrastructure, a public operator aiming to democratize the use of digital simulation through high-performance computing associated with the use of artificial intelligence and quantum computing to support French scientific and industrial competitiveness.
GENCI pursues three missions:
- Implement the national strategy of equipping France with high-performance computing, storage and massive data processing resources associated with AI (Artificial Intelligence) technologies and future quantum computing technologies for the benefit of French and European open scientific research, drawing on the three national computing centers;
- Support the development of an integrated ecosystem on a national and European scale
- Promote the use of numerical simulation, artificial intelligence and quantum computing among academic and industrial teams through open research.
GENCI is a civil company owned 49% by the French State represented by the Ministry in charge of Higher Education, Research and Space, 20% by CEA, 20% by CNRS, 10% by the Universities represented by France Universités and 1% by Inria.
Media Contacts
BriCS: Emily.coles@bristol.ac.uk
GENCI : nicolas.belot@genci.fr