Not yet. And probably not today.
But the question is becoming worth asking seriously, because the line between the two is starting to blur.
For years, the thinking has been straightforward: spend less keeping the lights on, free up more money for transformation, and growth follows. Shift the ratio, unlock the future.
It's a comfortable story. It also rarely works that cleanly.
Spending more on innovation doesn't guarantee results. Nor does cutting operating costs automatically create capacity to change. Banks have spent decades wrestling with this balance while managing technology estates that have accumulated complexity and cost over time.
So let's set that argument aside.
The more interesting question is what AI, including emerging agent-based approaches, might change about the economics of running a bank.
Not the AI apocalypse or the AI nirvana versions of the story. Banks operate somewhere in the middle, where investments need to pay off, controls matter, and the infrastructure you build is the infrastructure you have to live with.
And that's where things could get interesting.
The easy AI story is that it makes run-the-bank cheaper.
Automate selected tickets and manual processes, potentially supporting productivity and creating opportunities to manage operating costs. There's certainly value there, but it's not a fundamentally new idea. Banks have been automating operations for decades.
The more interesting possibility is that AI changes how those operations are actually executed and managed.
Think about selected activities such as reconciliation, exception handling, provisioning, monitoring, or parts of compliance-related workflows. Today, these involve combinations of people, applications, processes, and controls. Over time, agents may be able to assist with more than individual tasks, coordinating parts of selected workflows under defined controls.
Not simply assisting an employee, but potentially executing or coordinating parts of processes under human oversight and defined governance.
Of course, that doesn't remove accountability. A regulated bank should not assume that responsibility for operational risk, controls, or outcomes can be delegated to an AI agent. Human oversight, auditability, and governance become more important, not less.
But where the work gets done, and how much human intervention it requires, could change substantially.
Push that idea further.
What happens when the systems responsible for running the bank are also continuously improving how the bank operates?
The distinction between maintaining something and changing it becomes less obvious. Some improvements that once required separate projects, teams, and investment cycles might increasingly be supported within the operating environment itself.
That doesn't mean every change becomes autonomous, or that major modernization projects disappear. Banks will still replace systems, launch products, and make big technology bets.
But it does raise a question about whether run-the-bank and change-the-bank remain useful ways to think about where technology creates value.
The budget categories may survive. Finance will still need to distinguish recurring operating expenses from investments. But the operating model underneath those categories could look very different.
There's also a catch. AI introduces additional considerations, including costs associated with models, inference, monitoring, security, governance, and resilience. Layering agents over legacy processes without simplifying anything could actually make run-the-bank more expensive.
Potential value may emerge when banks redesign how work gets done, rather than simply automate pieces of existing processes.
And that leads to a potentially bigger opportunity.
Amazon is an interesting example, not because banking looks like retail, but because of what happened when Amazon solved its own infrastructure problems.
As Amazon scaled, its technology architecture had to change. Tightly coupled systems and bespoke integrations weren't going to support the business indefinitely. It needed reusable services, cleaner interfaces, and infrastructure that could scale without rebuilding everything for each new requirement.
Those internal capabilities eventually contributed to the creation and expansion of Amazon Web Services.
AWS became far more than an internal efficiency initiative: it became a business in its own right.
The lesson isn't simply that Amazon faced infrastructure constraints or that spending less automatically led to something bigger.
It's that solving an operational problem with the right architecture created options that weren't obvious at the outset.
And one of those options turned out to be an entirely new business.
Now apply that thinking to banking.
Imagine a bank building an AI-enabled operating capability to handle reconciliation, exceptions, compliance workflows, or other core operational processes.
At first, it's an internal capability. It could support efficiency, potentially reduce manual intervention in appropriate scenarios, and may support the bank's resilience objectives.
But what if those capabilities are designed as reusable services from the beginning?
An agent that helps resolve exceptions for one application may be useful. A platform designed to support governed exception handling across business lines, connect with different systems, support policy-based controls, and provide audit records could be more valuable.
That could potentially be valuable to other institutions, too.
Of course, there's a big difference between making something work internally and making it a product someone else will trust, integrate, and pay for. Particularly in banking.
That takes standardized interfaces, security, isolation, resilience, governance, and a commercial model that works across institutions. It also raises an obvious question: would a bank really want a competitor running part of its operations?
Those are significant design, governance, and commercial questions, not necessarily reasons to dismiss the possibility.
And not every bank needs to turn its technology into a commercial business for this to matter. Building reusable operational capabilities internally could itself be a meaningful competitive advantage.
The broader point is that run-the-bank infrastructure could become more than a cost to manage. It could become a reusable capability that supports scale and may create value beyond the institution.
It's tempting to assume the institutions spending the most on AI will come out ahead.
JPMorgan is one example of a large institution investing substantially in technology and AI. Capital One and other institutions have explored ways to commercialize internally developed technology capabilities.
These are useful examples, but they're not the only ones.
JPMorgan's Kinexys business is an example of a financial institution offering selected infrastructure or technology capabilities to external institutional clients.
And the idea of selling run-the-bank capabilities isn't new.
Companies such as Broadridge and SS&C provide technology and operational services to financial institutions.
A market for these types of technology and operational services already exists.
What AI could change is how these capabilities are delivered, packaged, and potentially extended.
Which raises another question: does agentic AI give banks an opportunity to capture more of that value themselves, or does it make the existing technology and infrastructure providers even stronger?
I don't think the answer is obvious.
Scale matters. So do access to data, distribution, regulatory experience, and the ability to invest over long periods.
But spending alone won't determine the winners.
There's a difference between building an expensive internal solution that works for one institution and building a reusable platform that can scale across multiple businesses or customers.
That's where the Amazon lesson becomes relevant again.
The advantage may come less from how much an institution spends and more from the discipline behind what it builds.
And there's no guarantee the winners are the biggest banks. It could be a mid-tier institution that approaches its operating model differently, a technology company already embedded in banking operations, or a new provider that builds around agentic capabilities from the start.
What's interesting is that AI could change the economics for banks, technology companies, and service providers in different ways.
No. Banks will still need budgets, controls, investment priorities, and accountability. They'll still have significant operating costs, and they'll still need to fund major technology change.
But I'm increasingly less convinced that the traditional run/change distinction tells us enough about where value is being created.
If operating infrastructure becomes more adaptive, reusable, and better able to support ongoing improvement, the relationship between running a bank and changing it starts to look different.
And if some of those capabilities can be turned outward, the implications go beyond operational efficiency. They reach into business models, competitive advantage, and potentially new sources of revenue.
None of this requires believing that banks will become autonomous or that every institution will build the next AWS.
It does require reconsidering an assumption that's been around for a long time: that running the bank is primarily a cost to be contained, while changing the bank is where the future value gets created.
What if, increasingly, the two are connected investments in the same operating capabilities?
And what if some valuable financial-services offerings are eventually built from infrastructure originally designed to support internal operations?
©2026 Nutanix, Inc. All rights reserved. Nutanix, the Nutanix logo and all Nutanix product and service names mentioned are registered trademarks or trademarks of Nutanix, Inc. in the United States and other countries. All other brand names mentioned are for identification purposes only and may be the trademarks of their respective holder(s).