The quiet reallocation of where banking actually earns its money.
Back in 2022 I argued that embedded financial services (the trend of ordinary companies offering banking and payments inside their own apps) would start to dissolve the line between banks and the digital platforms we use every day. A few days ago I picked that thread back up. This is the fuller version I promised.
That shift is now well underway. But the part that actually matters is not the blurring of boundaries. It is where the money goes.
Let me put that plainly, because it is the whole argument. For a long time, banks earned their money at the edges of everyday commerce. You bought something; a bank or card network quietly took a small cut of the transaction somewhere behind the scenes. Embedded finance changes who takes that cut. Increasingly, the company running the app, the shop, the platform, keeps it for itself. The margin is migrating from the banks to the businesses closest to the customer.
A few numbers to frame the size of this, though the cleanest ones are not from home. The UK embedded finance market is worth roughly $26 billion in 2026, heading toward $35 billion by 2030. The figure that really shows the scale is American: the consultancy Bain estimates that embedded finance handled around $7 trillion of US transactions in 2026, roughly a tenth of everything that moves through the US financial system. I use that as a yardstick, not a forecast for here; UK and European data is thinner. The exact decimal matters less than the direction, and the direction is not in doubt.
And those revenues will not be shared out evenly. They will flow to whoever controls the chain of activity the customer moves through.
This is no longer just an industry talking point. In its first Emerging Technology Horizon Scan (a report in which the regulator sets out the technologies it is watching), the Financial Conduct Authority described the big technology platforms as “dominant intermediaries within the global financial value chain”. It went on to sketch a near future in which the plumbing of finance, things like instantly settling a transaction, or money that can be programmed to move under set conditions, gets absorbed into the banks we already have rather than replacing them. When the regulator itself reaches for that framing, the question stops being whether this is happening. It becomes a sharper one: are you built to capture the value moving through you, or are you laying the track and letting someone else collect the fare?
The economics are moving upstream
For decades the arrangement was stable and simple to describe. Businesses generated the activity: the sales, the bookings, the journeys. Banks and card networks earned the money on the financial layer underneath it. That money came from things like the small fee charged every time a card is used, the gap between the interest a lender charges and what it pays, the margin on converting currency, and the interest earned on money held in transit. Almost all of it sat outside the business that actually created the activity.
Embedded finance turns that around. Platforms like Shopify and Uber have built payments, lending and digital wallets directly into their own ecosystems. Stripe has turned financial capability into something any company can plug in through software. Visa increasingly works with fintechs and “banking-as-a-service” providers: firms that let a non-bank offer bank-like products by renting the licence and infrastructure of a real bank behind the scenes, so that a card can be issued inside an app you would never think of as a bank.
It is not only banking and payments. Embedded insurance is growing just as fast: cover offered to you at the exact moment you need it, inside the thing you are already buying, whether that is a flight, a parcel or a car journey, priced on the spot.
None of these companies is trying to become a bank or an insurer. They are simply keeping the financial economics of activity that was already happening on their platforms. That distinction is the heart of it.
Experience is the way in, not the strategy
Embedded finance usually gets sold on the experience: the checkout with no friction, the credit offered at the right moment, the instant payout, the journey that just flows. All of that is real, and all of it is good.
But the experience is the Trojan horse. What it smuggles in is structural. A company that embeds finance well diversifies its income, holds more cash within its own system, lowers its cost of serving each customer, gains far richer data to judge risk and price with, and, crucially, makes itself harder to leave.
That last point is the one leaders underrate. This is not a new feature being switched on. It is the operating model of the business being quietly redesigned.
The real risk: enabling the flow without capturing it
Here is the trap a lot of organisations are already in. They enable financial activity inside their ecosystem without ever capturing the value of it. They process the payments. They pass their customers to a lending partner. They generate all the transaction volume. And the actual financial reward lands in someone else’s accounts.
With embedded finance revenues running into the hundreds of billions a year this decade, that is not a neutral choice. It is a slow leak of strategic value. And the danger is not that embedded finance fails to take off, and it plainly has. The danger is that a competitor embeds it more cleverly than you, and in doing so makes their customers far more costly to leave than yours.
How the winners actually do it
Across a career spanning retail and commercial banking, regulatory change, treasury operations, digital challenger-bank builds and large-scale transformation, I have seen the organisations that get this right follow the same disciplined sequence. Four steps.
1. Map where the money already flows. Before anything else, find every point where financial value already moves through your business: payments, lending exposure, customer deposits, currency conversion, insurance risk. On the UK’s first mobile-only challenger bank, this meant tracing every flow across lending, mortgages, payments and collections before a single one was turned into revenue. You cannot capture what you have not first located.
2. Draw the regulatory and risk boundary. Be clear about what you are on the hook for: your obligations to customers, the safety capital you must hold, and the resilience your operation has to prove. This is exactly where the banking-as-a-service market is being tested right now. Banks that lend their licence to fintech partners face materially more regulatory risk than those that do not, and the collapse of Synapse (a US banking-as-a-service provider whose failure left partner funds frozen) showed how fast a badly drawn boundary can bring the whole thing down. The boundary is not a legal formality. It is part of the design.
3. Decide what you keep and what you share. Work out which financial flows to run yourself, which to hand to a partner, and how the capital you commit lines up with the return you want. The leverage here is real: I have seen a £20,000 feasibility study set the structure for over £1.4 million of follow-on work. Get this decision right early and it compounds in your favour. Get it wrong and you can sign away the economics for good.
4. Build in the reasons to stay. Embed the financial capability deeply enough that leaving would genuinely cost the customer something: the convenience, the data, the connected services, the accumulated value. Defensibility is not a feature you bolt on at the end. It exists when a customer would lose something real by walking away. The deeper the embed, the higher the cost of leaving.
Incumbents and challengers start from opposite ends
Those four steps run in the same order whoever you are. What changes is which step is cheap and which is painful, and that depends entirely on where you start.
An established bank usually already has the financial flows: payments, deposits, lending, currency, but they are scattered across old, separate product systems that were never designed to be seen as a single chain. So for an incumbent, step one, mapping the flows, is the expensive one. It is less a discovery exercise than an archaeology dig: reconciling the profit-and-loss of different product lines, old ledgers and long-forgotten supplier contracts just to establish what is actually flowing where. But once that map exists, steps two and three go quickly, because the bank already holds the licence and the balance sheet that a newcomer would have to build or rent.
A challenger or platform business has the opposite problem. Mapping the flows is almost free, because there is no legacy tangle and the data can be designed for clarity from day one. For them, the expensive step is drawing the regulatory boundary: acquiring or renting the permissions, the capital treatment and the operational resilience that an established bank already has sitting on its licence. This is precisely the banking-as-a-service dynamic: a challenger borrowing a regulatory standing it has not yet earned, at a cost that only becomes visible when something breaks.
The practical lesson is: do not borrow a playbook from the other side of that line. Incumbents told to “move like a fintech” often skip the mapping and jump straight to deciding how to make money, then discover mid-build that two product lines were quietly counting the same currency margin twice. Challengers under pressure to grow often treat the regulatory boundary as a one-off legal sign-off rather than a structural design decision, which is exactly the failure the Synapse collapse laid bare. Know which step is expensive for you before you commit budget to the others.
The decision is being made now
By 2030, this reallocation of financial margin into digital ecosystems will be plainly measurable and material. The competitive question is no longer whether embedded finance will scale. It already has.
The real question is whether your organisation is built to capture the financial value moving through your own value chain, or whether it will carry on generating activity for someone else to earn from. That is not a technology decision. It is a strategic one. And it is being made now, whether or not it is being made deliberately.
Francis Hellawell
I solve hard problems in banking.