What can AI do for my bank?

It’s a question every bank CEO is asking today, and for good reason. Banking sector economics have been under pressure for decades (see Figure 1). Over the next five years, agentic AI may further disrupt the industry by altering which banking businesses remain structurally attractive, which competitive moats endure, and which customer relationships prove worth defending. It has the potential to lower barriers to entry, reset costs, strip friction from profitable products, and upend customer relationships. But viewing AI only through the lens of productivity and disruption misses another side of the opportunity: It can also help build genuinely superior propositions that take share with increased speed.

Figure 1

Banks with No. 1 or close No. 2 positions in their business lines have measurably higher return on equity (ROE) than those with weaker leadership positions (see Figure 2). Setting the right strategy remains essential. AI may change how defensible those leadership positions are and where new ones can be built. So before asking what AI can do for the bank, leaders need to ask where and how the bank should compete in an AI-led future.

Figure 2

Banks can pressure test their current strategy for an AI future by exploring a series of high-gain questions: Where can we build and defend positions that will be difficult to displace? Where should we rethink how we participate as existing positions become more exposed? And where can AI help create propositions that offer customers distinctive value? 

Those choices are what should inform any bank’s AI agenda and the allocation of capital and resources behind it.

Four hard truths for an AI-led future

1. Profit pools built on inattention and inertia are the most exposed

Many traditional banking advantages—physical footprint, brand, customer primacy, trust, and product integration—work because they help keep customers in place. That has proved remarkably durable. Customers do not continually compare every deposit rate, loan, fee, or financial product available to them. Switching takes time and effort.

Banks have been hearing about disruption from digital attackers for years. But even when challengers—whether upstart or incumbent rival—offered compelling products and experiences, customers still had to take action to switch. That friction has helped incumbents.

AI agents could change that equation. If comparing offers, identifying better terms, or switching providers becomes dramatically easier, businesses that benefit from customers staying put could become more vulnerable. Services and tools already exist to monitor deposit rates and help move cash toward higher-yielding accounts, and to redirect payroll and recurring bills. Agents could make finding and acting on better offers easier.

This could give digital attackers a better opportunity to move beyond areas where they have already made strong inroads, such as payments, into money management, simpler forms of lending, and potentially, over time, other parts of the banking relationship. Revolut, which started with international payments and currency exchange, has grown to more than 80 million customers globally by expanding progressively into savings, lending, and investment products, establishing banking relationships that traditionally took years for incumbents to build.

Nor will the opportunity belong only to attackers. Incumbent banks that remove friction faster could use the same dynamic to take share from traditional competitors.

2. Efficiency gains will be table stakes but may accrue to customers, not shareholders

Banks are racing to use AI to lower operating costs, and those efforts could have significant implications for expense levels, expense ratios, and customer experience.

But there is a catch. Productivity gains may provide a significant advantage, particularly for banks that move early, creating a window during which those banks can out-invest slower competitors, building new propositions and taking share. Over time, however, as banks adopt similar technologies in markets with excess capacity and comparable products, many of those gains could be competed away through lower prices or better products rather than retained as incremental profit.

Capturing shareholder value will therefore require banks to redeploy some of those gains into differentiated propositions. Those that do so early could take share from traditional competitors and compound that advantage over time.

3. Trust must be earned and converted into value others can’t match

As capital, core technology, and credit capabilities become more widely available or standardized, customer relationships remain a critical source of differentiation relative to insurgents. But the old playbook for building intimacy and loyalty may become less effective.

Banks enjoy a wealth of strengths that can support the durability of their relationships, particularly regulatory oversight and licensing and balance sheet capacity that reinforce trust, as well as a wealth of proprietary data about customers and their needs. To harness these advantages, they will need to find new ways to turn what they know about customers into proactive and anticipatory advice; personalized engagement and relevant, targeted offers; and services customers value enough to choose.

Trust belongs at the center of that effort—reliability, transparent value, proactive protection, and advice that clearly serves the customer. Banks enter this transition with advantages rooted in their role as regulated institutions with established infrastructure and protections. But that advantage cannot simply be assumed. As AI raises new questions around fraud, cybersecurity, and privacy, preserving and strengthening trust will become an active strategic priority.

4. Being large everywhere will matter less than being difficult to displace somewhere

The nature of scale advantage is changing. Focused competitors now have greater access to advantages once reserved for institutions with large technology budgets, wide distribution, and capabilities serving many products. Today, cloud and AI-based development reduce both the cost and time required to build and improve products; digital channels lower distribution barriers; and partnerships provide access to capabilities banks once had to own. 

Scale still matters where fixed costs, network effects, or proprietary data advantages are truly large. But those benefits accrue not simply to the bank with the biggest balance sheet, but to the leader in a given business. So the key questions become where and how banks can build and reinforce leadership positions that will be difficult to displace.

Banks of any size can strengthen their position by concentrating capital and capabilities in the few arenas where they can build such defensible leadership. Italy’s Intesa Sanpaolo is deepening its already-leading position in wealth management and insurance through its specialized advisory networks, digital capabilities, and integrated advice model. Capital One is using M&A to reinforce its position in payments. Its acquisition of Discover added a global payments network and direct merchant relationships to its card franchise, while its acquisition of Brex added AI-native business payments capabilities. Both build on Capital One’s long-standing investments in cloud, data, and AI.

Pressure test every business against an AI future

Those hard truths have different implications across a bank’s portfolio. For each business, leadership teams should consider three potential strategic responses (see Figure 3).

Build and attack where AI makes new propositions possible

Look beyond defense. AI could combine with banks’ existing assets to make previously uneconomic propositions viable:

Advice backed by a financial commitment. Recommendations coupled with guarantees, such as holding a rate or covering a shortfall against a stated benchmark. 

Credit before the customer asks for it. Facilities offered proactively based on information the bank already observes, such as receivables, payroll, or seasonality. 

Greater customization at smaller ticket sizes. Bespoke hedges, insurance, structured products, or credit made economical for customers previously too costly to serve.

Defend and deepen where profit pools remain structurally attractive

Look for businesses where the bank has an advantage AI is unlikely to erase.

The bank is a necessary part of the value chain. Custody, servicing, and settlement could remain attractive where switching is costly or regulated permissions matter.

Judgment matters more than simple price comparison. Complex, structured, or situation-specific credit may be more defensible where risks are nonstandard and offers cannot be easily compared, though these businesses also face competition from private credit.

The bank sees something others do not. An existing customer relationship can provide proprietary insight into transactions, receipts, payroll, financials, or other information that improves both the offer and customer engagement.

JPMorgan Chase illustrates how relationship depth can become an information advantage. Its Client 360 platform combines more than 150 internal and external data sources to sharpen cash-flow forecasting, pricing, solutions, and credit decisions—a data and analytics foundation that can support specialized businesses such as innovation banking.

Rethink participation where disruption threatens existing economics

Look for businesses whose economics depend on friction that AI could reduce:

Revenue that depends on inattention. Fees, back-book pricing, dormant balances, revolving card debt, and similar sources of revenue could come under pressure as AI agents alert customers to better alternatives.

Products readily compared on price. Deposits, standard mortgages, personal loans, vanilla term lending, retail foreign exchange, and remittances could face greater pressure as comparison becomes easier.

Businesses where others can build a structural advantage. Embedded finance and other models in which a competitor sits closer to the customer or transaction could present a different kind of challenge. 

Navigating tough portfolio choices

AI disruption does not necessarily mean tearing up the strategy. But it should force leadership teams to make explicit choices about where they will compete, where they will invest, and what they will need to win.

Four questions every executive team should consider:

1. How resilient is our portfolio?

Which businesses and profit pools are most exposed to lower switching costs, greater price transparency, or new competition, and which retain durable advantage?

2. Where will we lean in, and where will we reconsider our participation?

Which leadership positions are worth defending or building, and where should the bank narrow its role, partner, or provide infrastructure instead? Those choices will have implications for partnership and M&A strategies.

3. How must our customer proposition evolve?

Which customers and interfaces does the bank need to own, and what will give customers a reason to actively choose it?

4. Are our investments aligned with that strategy?

Where should investment in AI, data, partnerships, and enterprise capabilities concentrate to support those choices?

Start with strategy, then determine what AI must enable

Many banks have approached AI by asking where the technology can improve the business. That matters, but it starts the conversation one step too late. Start instead with how the bank intends to compete in an AI-led future. That requires making explicit choices about where to build and defend leadership, where to rethink participation, and what new value to create for customers, and then focusing AI investment behind those choices.