Will AI agents break the deposit and credit card inertia?

McKinsey authors Ramji Sundararajan and Uzayr Jeenah, with Alana Ellis, argue that agentic AI — "technology that can perform tasks and solve issues largely on its own" — threatens two of retail and SME banking's core "revenue engines": deposits and credit cards. Both "rely heavily on customer inertia and brand familiarity," but, the authors write, "AI agents won't care about brand loyalty. They will optimize for outcomes." They size the stakes at "more than $2.7 trillion in annual revenue" across the global payments industry, with net interest income representing "roughly 30 percent of retail-bank profit" and credit cards generating "$234 billion in 2024" from interest, interchange, fees, and unredeemed rewards.

The mechanism, per McKinsey, is agents substituting active optimization for passive habit. On deposits, the authors note that U.S. banks paid just "0.07 and 0.38 percent" on checking and savings in June 2025 even as "top online saving rates exceed 4 percent," while non-interest-bearing balances grew at a "CAGR of 28 percent" versus 3 percent for interest-bearing accounts — margin that exists only because consumers don't act on the gap. Agentic systems close that gap by monitoring balances, comparing rates across institutions, and sweeping cash to higher-yield accounts and back "in time for bills." McKinsey estimates that if "just 10 to 20 percent of consumers adopt agent-driven cash sweeps," European bank net-interest margins could tighten by "30 to 50 basis points." On cards, the same inertia shows up as unredeemed value: "more than 20 percent of cardholders didn't redeem any rewards" in the past year, and the CFPB's forfeiture rate runs "3 to 5 percent annually" — value an optimizing agent would reasonably claim. McKinsey frames open banking as an accelerant: "the changes will be felt first in markets where open banking has taken root," and points to agentic tools already operating in commerce — OpenAI's Operator Mode, Manus AI, Perplexity, Google's AI Mode shopping — alongside Visa, Mastercard, and PayPal's new agentic-payment capabilities as evidence the infrastructure is arriving.

The claim sits downstream of two surveyed limits on delegation: How willing are people to let AI control their finances? and Does naming a payments brand increase trust in AI transactions? both find that consumers are far less willing to hand agents banking and payment authority than other tasks — precisely the inertia McKinsey's thesis depends on breaking. McKinsey doesn't address that trust gap directly; its argument is about what happens to bank revenue once even a minority of consumers cross it, not about how quickly they will.

The excerpt gives no timeline for when agent adoption reaches the 10-to-20-percent threshold it uses for its own deposit-margin estimate, and its figures are McKinsey's own projections, not independent measurements of agent behavior in the market. If the trust gap found by Public First and Visa persists, McKinsey's inertia-breaking mechanism may compress bank margins far more slowly than the dollar figures in this report suggest.