Financial institutions are investing billions in artificial intelligence, yet many struggle to move beyond the pilot phase. Despite promising technical results, enterprise-wide impact often remains elusive. This disconnect sets the stage for the core argument of the article.
In their contribution to the Journal of Financial Services, Kristof Meganck and Bart Claeys contend that the real constraint is not algorithmic sophistication, but the way data is organised, governed, and delivered across the bank. Building on this premise, they show how legacy data operating models have reached their structural limits and argue for data mesh as a pragmatic foundation for scaling AI in a way that is safe, sustainable, and commercially viable.
This shift in perspective reframes the AI challenge fundamentally. Looking at AI through a data lens exposes a central paradox in financial services: while the value of AI is well established, scaling it across the enterprise remains difficult. European financial institutions already operate hundreds of AI use cases supporting client interactions, compliance, and operational efficiency, generating recurring value and productivity gains. Yet the impact remains uneven. Most of these successes stay confined to narrowly defined domains rather than extending across the organisation.
The limitations become more visible as AI matures. As systems move from experimentation into live production, data-related issues that were manageable during pilots become critical. Challenges around data quality, lineage, accessibility, and accountability intensify and translate directly into concerns about trust, regulatory compliance and model risk concerns that are particularly acute in financial services.
While the value of AI is well established, scaling it across the enterprise remains difficult.
These challenges are compounded by the limits of existing data architectures. Centralised and hybrid data operating models, while historically effective, are now reaching their breaking point. Enterprise data warehouses and lakes have delivered stability, but increasingly act as bottlenecks, stripping away business context and slowing AI delivery. As a result, central data teams struggle to keep pace with the growing demand for high-quality, well-contextualised data needed to support AI at scale.
It is against this backdrop that data mesh enters the discussion. Rather than being presented as a buzzword, data mesh is framed as a pragmatic shift in operating model. It promotes domain-oriented data ownership, treats data as a product with explicit quality and contract expectations, relies on centrally provided self-service infrastructure and applies federated governance to balance autonomy with control.
Importantly, the authors ground this model in real-world experience. At BNP Paribas Fortis, the introduction of domain-level data products, a Data Product Board and operational data stores aligned to business domains demonstrates how data mesh principles can be applied incrementally even within a complex and highly regulated environment.
The implications for leadership are clear. For financial services executives, the message is that AI initiatives will stall unless data ownership, funding and accountability move closer to the business. In this context, data products that support multiple analytical, regulatory, and AI use cases become strategic assets. Correspondingly, the role of central data teams shifts from delivery to enablement, while change management and leadership sponsorship emerge as factors just as critical as technology choices.
This strategic reframing becomes even more urgent in light of regulatory and technological change. As AI capabilities continue to advance and regulations such as the Digital Operational Resilience Act (DORA) and the EU AI Act raise expectations around data quality and oversight, institutions need data operating models that support both safe experimentation and industrialisation. Data mesh offers a way to strengthen resilience, governance and innovation speed without requiring large-scale system replacements.
Finally, the article connects data strategy to broader shifts in AI-driven operating models.
For complementary perspectives on how AI is reshaping financial institutions, see the following JoFS summaries:
The Projective Group Institute’s Journal of Financial Services (JoFS) provides structured insights on developments in the European financial sector. Each edition brings together contributions from practitioners, academics and regulatory experts to help readers understand key changes in the industry.
This edition examines how data and Artificial Intelligence are influencing financial services. It looks at how modern analytics and evolving regulations such as GDPR, DORA and the EU AI Act are raising expectations for data quality, governance and oversight. Through concise, thoughtful articles, the edition highlights the practical implications for decision‑making, risk management and operational resilience.
