BDVA at EuroHPC Summit 2025: AI factories set to make Europe an AI continent
BDVA is honoured to have contributed to the EuroHPC Summit 2025. We are proud of our association’s and our members’ contribution, as we have helped shape the event’s programme.
The Big Data Value Association is honoured to have contributed to the EuroHPC Summit 2025, an event that took place from 18-20 March 2025 in Kraków, Poland, under the Polish presidency. The EuroHPC Summit is an annual event organised by the JU that brings together key European stakeholders in supercomputing, both from the public and private sectors, alongside decision-makers. It provides a platform to share the latest technological developments, explore synergies, express current and future needs and help shape the future of European supercomputing.
The event’s programme was built on burning topics, relevant to the entire community, where the AI Factories took the stage on day 1. Discussions gravitated around AI Factories and European Competitiveness, Industrial Innovation and Scientific Research, the Applications Ecosystem and its future growth, AI Factories and Cloud Integration, The Future of Benchmarks in HPC, AI and Quantum Computing, the System Software Stack for Next-Gen HPC Applications and Architectures, the European Supercomputing Infrastructure, the future frontiers in Supercomputing, industry demands and Quantum Computing and ignited conversations, obtaining a valuable input that gave a thorough perspective on the status quo of the field.
BDVA and its members contributed to sessions related to AI factories, as our association is committed to collaborating with the European Commission, the EuroHPC JU and the future hosting organisations to support the implementation of the AI factories, as we will bring our community’s in-depth knowledge and contribution in the fields of Data, Data Spaces and Industrial AI.
AI Factories and Europe’s Competitiveness in the Global AI Race
During the EuroHPC Summit 2025, BDVA President and Siemens Chief Expert Software Thomas Hahn took the stage in the plenary session titled “AI Factories and Europe’s Competitiveness in the Global AI Race.” The session featured discussions on the role of AI in industrial innovation and the application of generative AI for industrial foundational models.
Hahn joined a high-level panel, which included Gustav Kalbe, Krzysztof Kurowski, Mariona Sanz Ausàs, Dennis Hoppe and Mateusz Maj.
Highlighting the significance of the AI Factories, Hahn emphasised their role in bridging research and industry, leveraging data to improve business operations and societal impact. He pointed out the necessity of seamless access to high-performance computing infrastructure to support the growth of emerging data ecosystems. Furthermore, he stressed that AI Factories are instrumental in ensuring that Europe remains a leader in trustworthy Industrial AI. European public supercomputers play a key role in fostering a collaborative AI development environment by integrating interdisciplinary AI ecosystems.
Calling for broader cooperation, Hahn highlighted the need for AI Factories to work together with industry data ecosystem initiatives and research organisations. He advocated for a governance framework that ensures standardised, fair and efficient access to AI services for enterprises across Europe. Thomas Hahn urged the AI community to place AI Factories at the core of Europe’s AI strategy to drive industrial competitiveness and development, which would end up having a positive economic impact and an immense societal growth and numerous benefits.
AI Factories as Catalysts for Industrial Innovation and Scientific Research
Another highlight of the EuroHPC Summit was the session titled “AI Factories as Catalysts for Industrial Innovation and Scientific Research,” moderated by Ana Garcia Robles, Secretary General of BDVA. The discussion was packed with attendees, reflecting the growing importance of AI Factories in Europe’s digital transformation.
Esteemed panelists included prof. Jan Hajic, Sébastien Bratières, Andre Martins, Vangelis Karkaletsis and Thomas Mayerhofer.
The session covered a wide array of critical issues and strategic opportunities, including the role of AI factories in achieving AI European Sovereignty e.g offering the necessary access and computing capabilities to strategic European projects such as OpenEuroLLM and LLMs4EU, advancements in machine translation and the necessity of adaptable frameworks for experimental AI research. Panelists also emphasised the importance of funding mechanisms such as small, fast grants to accelerate AI projects, reducing the costs and challenges of migration. Participants highlighted the importance of small, fast grants and advocated for multi-stage, iterative funding schemes. Grant allocation should prioritise empirical evidence over mere promises, ensuring that large grants are awarded based on demonstrated success in smaller-scale projects. How will we assess value in the years to come, as we will have to update models created through AI factories.
The value and the overall assessment of value was also touched upon. Discussion further delved into key user needs in experimental AI research, emphasising adaptable usage conditions, effective project selection and improved funding mechanisms.
The discussion further highlighted the importance of leveraging existing investments in HPC, AI and data infrastructure, such as AI TEFs, data Spaces, EDIHs, EuroCC, DeployAI etc. and also foster collaboration in between initiatives. Experts stressed the need to empower startups and easy and fast access to Pharos dedicated ΑΙ partition, build, maintain and join sectoral AI communities, AI Talent nurturing – capitalise on academia network to co-create sector-specific AI training programs and curricula, language and sustainability sectors, fostering a robust European tech ecosystem.
The session emphasised the crucial components of AI Factories—data, AI infrastructure and human capital—as well as the need for seamless access to high-performance computing resources. Discussions on next-generation AI models such as EuroLLM, TowerMoE and Multimodal EuroLLM showcased the potential of large-scale AI advancements, while concerns over delays caused by hardware issues and slow peer review processes for funding applications were also raised.
Discussions delved into hardware front, networking improvements for multi-GPU and multi-node AI workloads and also cultural shifts were also highlighted, with calls to support teams experienced in tools (such as Axolotl, Deepspeed,and Hugging Face Accelerate).
The session concluded with key recommendations, including identifying and sustaining successful projects to ensure uninterrupted progress, defining clear measures of success and streamlining the grant allocation process—potentially through automatic renewals. Some participants stated that open science principles, including open models, data and code, should be encouraged or mandated. Additionally, participants stressed the need to separate AI training from inference, recognising them as distinct use cases. Supercomputers should be dedicated to AI model training, while inference could be efficiently handled by outsourced cloud providers.
Through its speakers, the session voiced the need for efficient access and funding mechanisms, strong AI infrastructure and enhanced cooperation among academia, industry and policymakers and a close cooperation with Industrial ecosystems. With AI evolving at an unprecedented pace, the AI Factories aims at becoming essential to securing Europe’s leadership in artificial intelligence and to achieving the profound digital transformation of society. The EuroHPC Summit 2025 reaffirmed Europe’s commitment to fostering innovation, ensuring ethical AI development and maintaining its competitive edge on the global stage.
As a private member of the EuroHPC JU, BDVA congratulates the team for the organisation of the EuroHPC Summit 2025. We are proud of our association’s and our members’ contribution, as we have helped shape the event’s programme. We are committed to keep providing value to the AI Factories and other areas of the JU through our Task Forces, member activities and collaboration partners.
BDVA is active player in the field through its dedicated HPC – Big Data – AI Task Force, which promotes the integration and convergence between HPC, Big Data and AI technologies, nurturing and growing the ecosystem and with our Data Spaces Task Force, where we concentrate on securing data sharing realise the practical potential of sectorial European data spaces.