Data Gaps Are Stalling AI in Banking

The core of the problem lies in the financial sector's heavy reliance on legacy mainframe systems. Over 90% of financial services institutions, including the top 100 banks, still use mainframes to process the vast majority of non-cash and credit card transactions worldwide. While these systems are reliable and secure, they were not designed for the real-time data flows and unstructured data analysis that modern AI applications require. This reliance on older infrastructure creates significant data silos. Information is often fragmented across different legacy systems, making it difficult to get a unified view of a customer or to feed comprehensive, real-time data into AI models for things like fraud detection or personalized product recommendations. Integrating new AI technologies with these decades-old systems often requires costly and complex custom solutions. The scale of the data challenge is immense, as 80-90% of a bank's data is unstructured, including emails, call transcripts, and documents. Legacy systems are ill-equipped to handle this type of information, which is crucial for training effective AI models. Without access to high-quality, real-time, and varied data, the performance of AI in areas like identifying complex fraud patterns or understanding customer sentiment is severely limited. To bridge this gap, some financial institutions are turning to "data fabric" architecture. This approach creates a unified data layer that can access and integrate information from various sources, including legacy mainframes and modern cloud platforms, without needing to overhaul the entire underlying infrastructure. This allows banks to begin leveraging their vast data assets for AI while incrementally modernizing their systems. The financial incentive to solve this data dilemma is substantial. Generative AI alone is projected to add between $200 billion and $340 billion in annual value to the banking sector. Furthermore, AI is expected to drive up to a 20% reduction in net costs for banks. Global spending on AI in banking is forecast to hit $85 billion by 2030, a more than 1,400% increase from 2024. However, the cost of inaction is also high. Maintaining outdated systems already consumes a significant portion of IT budgets, with some reports indicating that banks spend nearly 70% of their IT budgets just to keep these systems running. This leaves little room for innovation and can lead to higher operational risks and a diminished customer experience. The cost of a data breach in the financial sector averages over $6 million, a risk that is heightened by aging infrastructure.