How to turn an AI Factory into an engine of innovation?

Francesco Ferrero joined Dr. Lilit Axner (EuroHPC-JU), Mario Grotz (Luxinnovation), Arnaud Lambert (LuxProvide), Simone Niclou (University of Luxembourg) and Bert Verdonck (LNDS) on the panel Orchestrating AI Factories. Together, they represented the full spine of the Luxembourg AI Factory, the consortium co-funded by the European Commission and the Government of Luxembourg to support organisations throughout their AI journey, from concept to deployment.

What makes Luxembourg a credible home for this ambition is not size, but agility. Decision-makers are accessible. Industry, knowledge institutions, and infrastructure can be brought to the same table quickly. That capacity for fast, concrete alignment is precisely what an initiative of this nature requires and what the Luxembourg AI Factory has been built to leverage.

Infrastructure and synergy

The AI Factory is designed to serve all actors, across all sectors. Its particular focus falls on SMEs — the backbone of Luxembourg's economy — and on four priority domains where the country already holds strategic assets: finance, energy, cybersecurity, and space.

At the centre of the offering sits a significant differentiator: MeluXina, the EuroHPC supercomputer hosted in Luxembourg and operated by Luxprovide, providing the high-performance computing power that most organisations cannot access on their own. For a startup testing a computer vision model or a financial institution stress-testing a risk algorithm, that infrastructure changes what is possible. But raw computing power alone is not the answer. 

What gives the Luxembourg AI Factory its distinctive reach and impact is the combination of what its founding members bring together: research expertise, innovation support, industry connectivity, regulatory knowledge, and digital infrastructure — assembled under one roof and oriented toward a shared goal. That convergence is difficult to replicate, and it is precisely what transforms an AI Factory from a label into a lever.

Building the tools for trustworthy AI

Together with the University of Luxembourg's SnT, LIST developed an AI Assessment Sandbox Configurator, an open-architecture platform that streamlines the configuration and deployment of tailored AI assessment environments. Incubated within the AI Factory, it addresses a concrete gap: as the EU AI Act raises the bar for conformity, the tools to meet that bar have not kept pace.

This AI Assessment Sandbox Configurator serves a variety of needs, including AI providers conducting technical self-assessment, competent authorities operating regulatory sandboxes, and notified bodies performing conformity certification and produces a bespoke sandbox deployable on any infrastructure, with no vendor lock-in. 

The longer-term ambition is a European sovereign open-source asset: scalable, extensible across sectors, and capable of feeding machine-readable regulatory intelligence at organisational, national, and European levels.

Proven in the field

Alongside the Configurator, LIST brings its AI Sandbox — a concrete environment where organisations can evaluate their models for technical robustness, regulatory compliance, and ethical behaviour, including bias detection, hallucination analysis, and linguistic inclusiveness. It is not a theoretical framework. The Sandbox has already delivered results: with Banque Internationale à Luxembourg (BIL) on the assessment of banking chatbots, with the City of Luxembourg on AI applications in public services, and with Mistral AI — one of Europe's leading AI developers.

That track record matters. It signals what an open, sovereign, human-centred approach to AI can produce when research infrastructure meets real operational needs.y

(This article was written by LIST and republished on our website)