More than two-thirds of small businesses that apply for a bank loan do so because of an existing relationship with the lender, and this relationship is based on trust. That’s why banks still remain the first call for most businesses seeking credit.
There’s one thing that has changed, though, and it’s a big part of why nonbank credit offers have expanded so quickly. According to Mordor Intelligence, nonbank lenders now handle some $2.9 trillion of the $12.8 trillion global SME lending market, and this share is growing at nearly twice the pace of bank lending.
We asked Dmitriy Wolkenstein, CEO and co-founder of TIMVERO, to share his take on why companies are increasingly turning to alternative credit over traditional bank loans, and what it will take for banks to close the gap. Here’s what he had to say.
Speed is one of the factors pulling businesses toward that exploration. Trust in banks hasn’t declined, but the speed of lending has become one of the deciding factors, and it’s the one where banks are falling behind.
Typical time to a credit decision in SME lending takes from three to five weeks in traditional banks, with cash arriving closer to three months later. The strongest digital lenders reach a decision in minutes. For a business whose capital need is tied to a specific deadline, like a supplier discount, a payroll gap, or a time-limited opportunity, that gap is the difference between financing that works and financing that arrives too late to matter.
Let’s take a closer look at why that gap exists, and what it will actually take for banks to close it.
An Infrastructure Problem, Not an Effort Problem
The gap isn’t explained by bank lending teams working less hard than fintech competitors. It’s all about what each side is building on top of. A 2024 survey of over 200 senior bank IT leaders found that 55% of banks name limitations in their core banking system as the single biggest obstacle to hitting their business goals.
The American Bankers Association’s 2025 research found that 35% of banks are dissatisfied with their core processor, with a wide gap between how core vendors rate their own performance and how the banks using them do.
That dissatisfaction traces back to how these systems were built in the first place. Most core platforms in use today are decades old, designed long before real-time processing or rapid product changes were ever a requirement. Banks have kept extending them.
Fear of change plays a part too. The same survey found that 69% of banks cite migration risk as the reason they hesitate to deploy next-generation systems, since any intervention, any mistake in a system handling live customer accounts, can mean real financial losses. But change becomes inevitable once dissatisfaction with the product gets harder to ignore than the risk and cost of fixing it.
The economics explain why. A full core banking transformation typically costs hundreds of millions of dollars and takes years to plan and deliver. According to McKinsey, only 30% of these transformations fully succeed on schedule and on budget. Betting that scale of risk on launching one new credit product faster is not a decision most banks can responsibly make.
So the safer choice, protecting the core and routing every product change through a long internal queue, becomes the one that quietly costs the bank the deal.
Where the Correction Is Already Happening
The market is already adjusting for this, and the adjustment looks less like banks losing to fintechs than lending infrastructure fragmenting into faster-moving layers within banking itself. British Business Bank states that challenger and specialist banks now account for 60% of gross SME bank lending in the UK, up from 39% in 2012. Since 2013, 28 new providers have entered the UK’s smaller-business lending market. This means that the institutions gaining get defined by how quickly they can change their own product logic.
Globally, the pattern holds. Banks still handle the majority of SME lending volume, but nonbank lenders are gaining share fastest at the margin, among new customers, thinner credit files, and time-sensitive deals, while relationship-driven lending stays largely intact at the core.
One more factor is accelerating that shift: the AI buildout of banking itself. Every lender in this market, bank or fintech, is racing to fold AI into how it operates, and that race is reshaping the competition as much as the infrastructure fragmentation already described. But not all AI use is the same, and what a lender actually does with it matters more than whether it uses it at all.
That distinction gets lost in most coverage of this shift. The competitive lever isn’t “AI will decide who gets a loan faster.” It’s which institutions can build, test, and adjust a lending product in weeks instead of quarters, and which are still measuring that cycle in fiscal years.
Where AI Belongs in This Story, and Where It Doesn’t
This runs against a lot of current AI marketing, but it’s worth stating plainly: generative AI has no place at the moment a credit decision gets made. Lending decisions must be explainable and reconstructible months or years later. A regulator can ask why a specific application was declined, and “the model decided” has never been an acceptable answer. That part of the process should remain deterministic, reviewed, and owned by human underwriters and compliance teams.
Where AI has changed outcomes is upstream of that moment, in how fast an institution can move from deciding it wants to launch a product to having a compliant, production-ready system live. According to McKinsey, implementing a new credit-decisioning model the traditional way takes twelve to twenty-four months, and an agile approach can compress that to under six.
Client deployments on TIMVERO’s own lending infrastructure show similar gains. One bank took a fully bespoke lending product, with guarantor support and multiple refinancing paths, from signed contract to production in six months, after three earlier attempts with other vendors had failed, and cut cost per loan by 60% along the way.
A European fintech client launched banking-grade, multi-country lending in four months instead of the better part of a year, automating 98% of the process while keeping every decision auditable.
None of that involved AI talking to a customer or approving an application. But AI compressed the distance between identifying a needed change and having engineering ship it, while every credit decision continued to run through the same human review and sign-off it always has, just built faster.
Companies are not leaving banks because banks became untrustworthy. They are going wherever the infrastructure lets speed and discipline coexist. For banks, that remains a winnable competition, but only if it’s treated as an infrastructure problem rather than a trust problem.
