Your Company Is Already a Bank. You Just Haven’t Noticed

By Ilya Podoynitsyn, CEO FinHarbor

A founder I work with runs a vertical SaaS platform serving a few thousand small merchants. Going through his payment stack last quarter, I asked, almost in passing, what the average idle balance looked like across his customers’ accounts at the partner bank. The number stopped the conversation. Tens of millions, sitting in third-party infrastructure, generating float for a counterparty he had never met. His platform held the data. His platform held the customer relationship. And his platform was paid nothing for any of it.

He wasn’t running a SaaS company. He was running a bank that happened to bill itself as software.

This is the conversation I keep having in 2026. Embedded finance has stopped being a feature and started being a question of who keeps the economics of a transaction. The numbers are no longer in dispute. Bain & Company projects embedded finance will move $7 trillion through US software platforms by 2026 – roughly ten per cent of all US financial transactions – with platform and enabler revenue more than doubling to $51 billion in the same period. McKinsey expects the category could account for ten to fifteen per cent of European banking revenue pools by 2030.

Look at what this means in practice. Toast closed 2025 on $6.15 billion of revenue, of which the SaaS subscription line was barely a billion. The rest is payments, lending and other financial services delivered to restaurants. Shopify Capital originated $4.2 billion in loans and merchant cash advances to its own sellers in 2025, underwritten on data the platform already had. Neither company is a bank in the regulatory sense. Both behave like one in every other sense.

The interesting question is why so few platforms with the same customer trust, the same data and the same captive distribution have done the same thing. The honest answer is fear of complexity. There is a deeply held belief that “becoming a bank” means a charter, three years of build, an army of compliance hires and a board-level identity crisis. That belief was true a decade ago. It is not true now.

What “becoming a bank” actually means in 2026 – for a SaaS company, a marketplace, a logistics platform, a retail network – is a stack of three things sitting on top of someone else’s regulatory perimeter. A licensed partner that carries the regulatory weight: an EMI in the EU, a sponsor bank in the US, a VASP for crypto rails. A modular infrastructure layer that handles the moving parts: KYC, ledger, IBANs, card issuing, AML, reconciliation. And your own product layer, which is the only piece your customer ever sees. The platform owns experience and data. The licensed partner owns regulatory burden. The infrastructure layer owns the engineering that almost everyone gets wrong when they try to build it themselves.

Over four weeks, the build looks like this. The first week is environment setup, KYC pipelines and IBAN account management – the parts that used to take a quarter on their own. The second adds card issuing and payment rails, SEPA and international, glued to the existing user base through a single API. The third is the differentiator: lending logic, treasury, FX, crypto on/off ramps if the use case calls for it, loyalty integrations.

The fourth is hardening, audit logging, regulator-ready reporting and the user interface that hides all of the above. At the end of the month, the platform that started as software is running its own accounts, issuing its own cards, lending against its own data and keeping the float that used to be someone else’s revenue line. Four weeks is the realistic ceiling for a competent team working with pre-integrated components – not a marketing claim. We have done this often enough to be specific about it.

The economics rearrange quickly. A platform that processes a billion in annual transaction volume and previously earned nothing on customer balances now earns float income, interchange, lending margin and FX spreads – at conversion costs an order of magnitude lower than what a from-scratch build would imply. Fifteen to twenty engineers and roughly €1.5 to €2 million of infrastructure build collapse into a deployment exercise. The competitive moat shifts from “do we have a banking licence” to “do we own the customer relationship better than the bank does.” For most platforms with real distribution, the answer is already yes.

There is a version of this that goes wrong. I have seen teams try to assemble the stack themselves from a dozen vendors and end up with a Frankenstein that breaks at every API update. I have seen others outsource so completely that they lose the data layer that made the bet worth taking. The point of modular infrastructure is to keep what is yours – the customer, the experience, the data – and stop paying someone else for what is no longer hard.

The decision in front of every platform with a meaningful balance sheet is no longer whether your company can become a bank. The infrastructure to make that happen is sitting in the market, priced and ready. The decision is whether you keep paying a third party to do it for your own customers, on your own data, while watching the margin walk out of the building.

Similar read: Crypto Savings vs. Traditional Bank Savings Accounts: What is the Difference?