Banks rethink legacy IT costs to fund AI

Banks are rethinking legacy technology spending and vendor strategies to free capital for AI whilst keeping operations resilient.

Asian banks are under pressure to free technology spending for artificial intelligence as maintenance, security, and rising software costs consume most IT budgets.

Bank technology budgets have grown to between 8% and 12% of revenue, but up to 90% can go towards maintaining existing systems. Luc Grimond, Managing Director and Senior Partner at Boston Consulting Group, said banks need to rebalance spending towards transformation.

“It’s very important to find the right balance and free up some capital to reinvest in the growth,” Grimond said.

Michael Perica, Chief Financial Officer at Rimini Street, said banks can stabilise existing systems whilst shifting IT teams from continually buying vendor upgrades towards building solutions with newer technologies.

Reducing vendor lock-in presents another challenge because switching critical systems carries execution risks and may not be economically justified.

Grimond said banks need vendor autonomy strategies that determine what to build and buy, establish alternative suppliers for critical systems, and consider open-source solutions. He also warned that aggressive software bundling can make existing vendor relationships harder over time.

Perica said consolidating vendors and using third-party support for existing systems could release capital without immediately replacing critical infrastructure.

Banks also face pressure to demonstrate returns as they scale AI. Grimond estimated that fewer than 30% of financial institutions are currently delivering substantial value from their AI investments.

He said banks should assess the economic return from AI by considering both human intelligence and token costs rather than measuring deployment scale alone.

Regulators, however, will prioritise operational safety, resilience and governance as AI becomes more widely deployed.

For Perica, banks’ data-rich operations offer compelling AI opportunities, but that does not justify enterprise-wide deployment. Banks instead need to identify specific processes where AI can generate sufficient returns whilst directing capital towards use cases that justify the investment.