Why AI Isn't Delivering for Banks

Banks face a growing gap between their AI investments and actual results, largely because they rely on outdated systems and processes. The anticipated cost cuts and better customer experiences are not materializing because AI is often added on top of inefficient systems, rather than being used to rebuild them.

AI Investment Falls Short on Old Foundations

Banks are channeling substantial capital into AI technologies like voice bots and analytics, with projections often citing potential cost reductions of 30-45%. However, these benefits are frequently unmet. The main reason AI fails is that it automates current inefficient processes rather than fixing the root causes of customer issues. AI effectiveness is further hampered by regulatory and risk constraints, which are often added later rather than built in from the start. Industry benchmarks show that while some technology-forward banks spend as much as 16.4% of revenue on technology, many struggle to translate this investment into commensurate value.

Legacy Systems Drain IT Budgets, Slow Progress

Analysis reveals that banks spend a large part of their IT budgets—up to 70%—just to maintain old systems, leaving less money for innovation. IT spending can range from 6% to 12% of revenue. Core banking platforms can be 30-40 years old, creating delays and making it harder to launch new products quickly, unlike nimble FinTech firms. Many institutions still use COBOL-based systems, with billions of lines of COBOL code globally, which are becoming expensive and difficult to maintain. These old systems slow down updates compared to modern ones, affecting how fast changes can be made. Some 43% of companies struggle with AI projects because they lack clear business goals and proper data management. Some leading banks, like JPMorgan Chase and Bank of America, are investing heavily in technology and AI talent to integrate AI fully. Bank of America, for instance, is investing $4 billion in technology for 2024, with its virtual assistant Erica handling billions of interactions. However, many others remain stuck in "pilot purgatory," trying out AI on small projects without expanding them across the company.

The High Cost of Doing Nothing: Missed Revenue and Risks

Banks that don't update risk not just missing out, but becoming irrelevant. Old systems cost more to run, are less secure, and can't keep up with digital needs. By 2028, banks failing to modernize could lose over $57 billion, with significant revenue missed in payments alone. Systems pieced together over years through mergers make banks slow to adapt to customer demands and new rules. AI's promise of efficiency is limited when applied to these unstable systems. Unlike competitors building modern systems, banks with old "technical debt" face rising maintenance costs and greater security risks, including slower breach detection. A shortage of AI talent, made worse by competition from tech companies, also hinders advanced AI projects.

Unlocking AI's Power Requires Overhaul, Not Overlays

McKinsey stresses that banks must fundamentally change how they operate, not just add AI. They need to redesign operations fully. Success requires dismantling old structures. This could lead to benefits like a 25-40% drop in customer calls and higher satisfaction. The future of banking depends on treating AI as a transformative force, requiring a complete overhaul of everything from talent and technology to data and risk management.