Banks have a problem with the ATM. Customers are using them less, digital payments are taking a growing share of everyday transactions, and maintaining thousands of cash machines is becoming harder to justify. Closing them should be an easy efficiency decision. Except cash has refused to disappear. Millions of consumers still withdraw it, some communities depend heavily on physical access to it, and cash remains a useful fallback when digital payment systems fail. Banks are therefore confronting an awkward infrastructure question: how do you dismantle an ATM network built for another era without removing access to something customers still need?
The ATM network is already getting smaller
The decline is particularly visible in mature banking markets.
In the UK, the number of free-to-use ATMs fell from 54,599 in 2017 to 33,710 by the end of 2025, according to LINK, the country’s main cash access and ATM network. The total number of cash machines declined another 5% during 2025, from 44,569 to 42,403.
Usage is falling as well. UK consumers made 832 million ATM withdrawals in 2025, 9% fewer than a year earlier. The average adult visited a cash machine around 15 times during the year.
The direction is similar in the US. Federal Reserve data shows that ATM withdrawals have fallen substantially over the past decade, while cards, mobile payments and other digital methods have absorbed an increasing proportion of everyday transactions.
This is not simply a story about machines being removed. It reflects a much bigger change in how banks allocate capital to physical infrastructure that fewer customers regularly use.
The economics are becoming harder to justify
An ATM carries costs regardless of how frequently customers use it.
Banks and independent operators have to pay for hardware, software, connectivity, maintenance, security, cash replenishment, insurance and physical locations. Cash itself also needs to be transported, counted, stored and protected.
When transaction volumes fall, those costs are spread across fewer withdrawals.
That creates an obvious incentive to consolidate networks, particularly when banks are simultaneously reducing branch footprints and moving more customer activity towards digital channels.
The calculation becomes increasingly difficult for machines in low-volume locations. An ATM that once supported hundreds of transactions every day may no longer make economic sense when customers can complete most routine banking activities from a smartphone.
For banks, this is becoming part of a broader infrastructure optimisation exercise: deciding which physical capabilities still justify their cost and which can be absorbed by digital banking.
The smartphone has replaced much of what the ATM used to do
Cash withdrawals are only part of the story.
ATMs became successful because they moved routine banking transactions outside the branch. Customers could check balances, transfer money, make deposits, obtain statements and access cash without waiting for a teller or visiting during banking hours.
Much of that functionality now sits inside a mobile banking application.
Balance enquiries happen instantly. Transfers can be initiated from anywhere. Cards can be frozen, replaced or added to digital wallets. Bills can be paid digitally and, in some markets, even cheque deposits can be completed remotely.
The transformation of payments has accelerated the change.
Cards and digital wallets have reduced the number of occasions when consumers need physical currency in the first place. Real-time payment systems are making account-to-account transfers easier, while mobile wallets increasingly allow customers to leave both cash and physical cards at home.
The broader shift towards Payment Innovation in Banking: Real Time, Cross Border and Wallet Driven Evolution is therefore changing more than the payment experience. It is also changing the infrastructure banks need to support it.
But there is an important distinction between using cash less frequently and no longer needing access to it.
Cash has proved surprisingly difficult to eliminate
Cash usage has fallen dramatically in many economies, but the decline has not followed the straight line towards extinction that some predictions once suggested.
Federal Reserve research found that US consumers made an average of seven cash payments per month in 2024, unchanged since 2020. Cash still represented 14% of consumer payments and remained the third-most-used payment method behind credit and debit cards.
More than 90% of US consumers also said they intended to continue using cash either for payments or as a store of value.
The UK presents a similar contradiction. Despite declining ATM transactions, adults still withdrew an average of £1,352 from ATMs during 2025, while cash machines accounted for the overwhelming majority of cash withdrawals.
The pattern suggests that cash is changing its role rather than simply disappearing.
For many consumers it is no longer the default way to pay for groceries, transport or everyday purchases. But it remains useful for budgeting, smaller transactions, emergencies and situations where electronic payments are inconvenient or unavailable.
That makes cash increasingly resemble a secondary payment rail: used less frequently, but still expected to be available.
A digital banking system still needs a physical fallback
The more banking becomes digital, the more significant resilience becomes.
Digital payments depend on multiple layers of infrastructure working correctly: electricity, telecommunications networks, banking systems, card networks, payment processors, cloud platforms and customer devices.
Most of the time, those systems work exceptionally well. But outages happen.
A mobile banking application can become unavailable. A card network can experience disruption. A merchant terminal can lose connectivity. A customer’s smartphone can simply run out of battery.
Cash does not solve every resilience problem, but it provides an alternative that operates differently from the digital infrastructure surrounding it.
This becomes particularly relevant as banks become more dependent on technology providers and interconnected platforms. Finnoex has previously examined this issue in The Biggest Technology Risk Facing Banks Isn’t Legacy Systems. It’s Vendor Concentration. The same concentration of digital infrastructure that makes banking more efficient can also create shared points of dependency.
Cash therefore has value beyond the number of transactions made with it.
It provides optionality.
The risk is creating cash deserts
ATM closures are unlikely to affect every customer equally.
Removing one cash machine from a city centre containing dozens of alternatives may have almost no practical impact. Removing the only ATM from a small community can be very different.
Older consumers, lower-income households, small businesses and people living in areas with limited banking infrastructure can depend more heavily on cash.
That means ATM rationalisation can quickly move from being an operational efficiency decision to a financial inclusion issue.
Regulators are already paying closer attention to cash accessibility in markets where bank branches and ATMs are disappearing simultaneously. Banks may eventually find that the commercial economics of individual ATMs cannot be considered independently from the wider responsibility to maintain reasonable access to financial services.
This creates an uncomfortable dynamic.
The customers who use cash infrastructure most heavily may not be located where operating that infrastructure generates the strongest commercial return.
The future may be fewer but smarter machines
The ATM itself does not necessarily have to disappear.
The basic cash dispenser may simply evolve.
Cash-recycling ATMs can accept deposits and reuse deposited notes for subsequent withdrawals, reducing some of the cost and complexity associated with continuously replenishing machines. More advanced self-service terminals can support deposits, transfers, authentication and other transactions previously handled inside branches.
That could make the ATM more important in locations where traditional branches no longer make economic sense.
Instead of maintaining large networks of relatively simple cash dispensers, banks could operate smaller networks of multifunctional self-service banking points.
This would also fit the direction banking technology is already taking. Routine activity increasingly happens digitally, while physical infrastructure becomes more targeted towards transactions or customers that still require it.
The ATM network of the future may therefore be smaller, but each machine could perform a broader role.
Banks need to measure access, not machines
The wrong question for banks is how many ATMs they can remove.
The more useful question is whether customers can still access cash conveniently after those machines have gone.
That requires a different approach to network planning.
Banks will increasingly need to consider geographic coverage, demographics, local cash usage, nearby alternatives and the availability of branches, retailers or shared banking facilities when deciding where ATM infrastructure remains necessary.
Technology can help make those decisions more precise. Banks already possess enormous amounts of transaction and location data that can show where customers withdraw cash, how frequently machines are used and what happens to behaviour when access points disappear.
The challenge is turning that information into better infrastructure decisions. As Finnoex explored in Banks Don’t Need More Data. They Need Better Decisions, collecting more information creates little value unless institutions can use it to make better operational choices.
ATM rationalisation is a good example.
A smaller network may be inevitable. A badly designed smaller network is not.
What it means for the industry
- ATM networks will continue to shrink, but they are unlikely to disappear. Falling transaction volumes will make low-use machines increasingly difficult to justify while strategically important locations remain necessary.
- Cash is becoming a secondary payment rail rather than an obsolete one. Digital payments dominate growth, but persistent cash usage means banks cannot plan infrastructure around a completely cashless customer base.
- The remaining ATM network will need to become smarter. Cash recycling, deposits and broader self-service capabilities can make fewer machines more useful and improve their economics.
- Cash access will increasingly become an inclusion issue. Older customers, lower-income households and communities with limited banking infrastructure may be disproportionately affected by rapid ATM closures.
- Resilience gives cash continuing strategic value. As banking becomes more dependent on interconnected digital systems, maintaining an alternative payment mechanism provides useful redundancy.
- Banks should measure accessibility rather than machine count. The future question is not how many ATMs a bank operates, but whether customers can still obtain and deposit cash conveniently when they need to.
Image Source: Pexels.com
