When we published our 2026 outlook a couple of weeks ago, the second shift we named was that embedded finance is leaving its easy phase. That line did more work than it looked like it did. For about a decade, putting a bank account, a card or a payment inside someone else's app was treated as an integration problem. Find a provider, wire up the API, ship the feature. The hard parts - the license, the safeguarding, the liability when something goes wrong - sat quietly behind a partner and rarely surfaced in the product roadmap.
That arrangement is now under strain on both sides of the world. So this is the second of seven follow-ups, and like the first, it pushes past the headline into the part that actually decides whether a business survives the next supervisory cycle.
Fintech Trends 2026 - Deep-dive #2: embedded finance leaving its easy phase.
The growth was real. Bain & Company estimated that embedded finance moved roughly 2.6 trillion dollars in US transactions in 2021, nearly 5 percent of the total, and would pass 7 trillion by 2026 - over 10 percent of all US transaction value. Whatever the precise number turns out to be, the direction is what matters, and the direction was straight up. Software companies discovered they could earn more from the financial flows running through their products than from the software itself.
The easy phase was the period when that opportunity met almost no friction. A non-financial brand could offer accounts, cards or lending by leaning on a licensed partner, and the regulatory weight stayed with the partner. The distribution side got the upside; the licensed side carried the risk it was not always pricing correctly. That imbalance is the thing that is now correcting.
The clearest signal came from the United States. In July 2024, the Federal Reserve, the FDIC and the OCC jointly issued a request for information on bank-fintech arrangements - the very structures that make most embedded finance work. The agencies were explicit about the risks they were watching: third-party, liquidity, compliance and operational risk, sitting inside arrangements where a fintech, not the bank, faces the end customer. That request did not arrive in a vacuum. It followed a run of consent orders against banks whose programs had grown faster than their oversight of them.
Europe is moving on its own track but in the same direction. The European Commission's 2023 payments package - PSD3, the Payment Services Regulation and the Financial Data Access framework - tightens authorization, safeguarding and conduct rules, folds the e-money regime into a single payment-institution framework, and pushes account access well beyond payments. The detail still has to settle, but the intent is plain: the regulated perimeter around embedded products is being drawn more firmly, not loosened.
The lesson from both jurisdictions is the same. The bottleneck in embedded finance is no longer the API. It is the license, the safeguarding and the accountability that sit underneath it - and supervisors are now reading those underneath parts much more closely than the front-end experience on top.
The trend article raised the partner-bank versus standalone-license choice, and India is the sharpest illustration of why it matters. In June 2022, the Reserve Bank of India told non-bank prepaid issuers that prepaid instruments could not be loaded from credit lines, and that any such practice had to stop immediately. A single clarification reshaped a whole category of buy-now-pay-later and credit-on-wallet products overnight, because those products had been built on an interpretation the regulator did not share. Firms relying on a partner's license found the rules of their own business rewritten without notice.
That is the real trade in the build-versus-partner decision, and it is not mainly about cost. A partner model is faster to launch and lighter to run, but you inherit the partner's risk appetite, their supervisory relationship, and their exposure to a rule change you do not control. Your own EMI or payment-institution license is slower and heavier, but it puts the regulatory relationship - and the ability to defend your own model - in your hands. Neither is right for everyone. What has changed is that the decision can no longer be deferred. The easy phase let platforms avoid choosing. The harder phase makes the choice for you if you wait too long.
The easy phase let platforms avoid choosing. The harder phase makes the choice for you if you wait too long.
| Model | What it gives you | What you take on |
|---|---|---|
| Partner-bank model | Faster to launch and lighter to run | You inherit the partner's risk appetite, their supervisory relationship, and their exposure to a rule change you do not control |
| Your own EMI or payment-institution license | The regulatory relationship and the ability to defend your own model in your hands | Slower and heavier, and the substance has to be genuinely yours |
Neither is right for everyone. What has changed is that the decision can no longer be deferred - the harder phase makes the choice for you if you wait too long.
The correction lands differently depending on where you sit in the stack.
For payment institutions and EMIs, the opportunity is that platforms now need licensed partners who can demonstrate real control, not just connectivity. The pressure is that you are the one supervisors examine when a distribution partner's program goes wrong. Onboarding, monitoring and safeguarding have to be yours in substance, not outsourced in all but name.
For SaaS platforms embedding finance, the temptation is to treat the financial layer as another feature. It is not. The moment customer money moves through your product, you have taken on obligations around segregation, complaints, fraud and conduct that a software business has never had to carry. The platforms that prosper will be the ones that decided early whether to hold a license themselves or to partner with eyes open to the risk they keep.
For BaaS providers, the squeeze is structural. The model only works at scale if every downstream program is supervised properly, and supervisors now expect the provider to evidence that oversight program by program. Strong banking-as-a-service infrastructure in 2026 is judged less on the breadth of its integrations and more on whether it can show clean onboarding, monitoring and reconciliation across every brand it powers.
For early-stage founders, the trap is assuming the easy phase is still on. It is closing. Building a financial product on a partner whose risk you have not examined is borrowing a problem you cannot see.
Our advice to clients is consistent and unglamorous. Decide, deliberately, whether you are a distributor of someone else's regulated product or the holder of the license - and structure for that honestly. If you partner, examine the partner's supervisory standing and risk appetite as carefully as their API documentation, because their rule change becomes your outage. If you hold the license, make sure the substance is genuinely yours: real onboarding, real monitoring, properly segregated client-fund accounts, and an audit trail you can put in front of an examiner.
None of this requires exotic technology. It requires a platform where accounts, payments, AML and KYC, safeguarding and reporting already sit together with the controls attached. A capable core banking platform is what turns "we embedded finance" into "we can show exactly how it is controlled, and who answers for it." The interface was always the part anyone could copy. The underneath is the part that earns the right to keep operating.
The headline - embedded finance is leaving its easy phase - sounds like a slowdown. It is not. The flows keep growing; what is ending is the period when distribution captured the upside while someone else absorbed the risk. The firms that win the harder phase are not the ones with the slickest embedded experience. They are the ones who decided, on purpose, where the license and the liability sit, and built the controls to match. The API was never the hard part. Own the underneath.
This is the second in a seven-part series expanding on our 2026 fintech trends. If you are weighing partner-bank against your own license for an embedded product, talk to our team - we have done it across 100+ licensing and platform builds.
