Embedded Finance: When Every App Becomes a Bank

Disclosure: this page contains affiliate links. If you sign up through them we may earn a commission, at no extra cost to you. It does not change what we write.

What is embedded finance? It is a financial product — a payment, a card, a loan, an account — offered inside the app of a company that is not a bank, with a licensed bank or payment firm behind it doing the regulated part. Shopify lending to its merchants, Uber paying drivers onto a debit card, and "pay in four" at a checkout are all embedded finance. The convenience is real. The risk is that the brand you see, the firm that keeps the record of your money, and the bank that actually holds it can be three different companies — and deposit insurance only covers one of the three failing.

That invisibility is the whole subject. The same structure produced one of the worst consumer outcomes in recent fintech history, so it is worth understanding the plumbing before you keep a meaningful balance in an app that is not a bank.

The three-party structure

Almost every embedded-finance product is three companies wearing one brand.

• The platform — Shopify, Uber, an airline, a retailer. It owns the app, the brand on the card and the customer relationship. This is the only one you interact with.

• The licensed bank — the entity that actually holds the deposits, carries the charter and answers to a regulator. Its name is usually in small print on a terms page.

• The middleware, or Banking-as-a-Service (BaaS) provider — the software layer between the two. It moves the money, handles onboarding and compliance tooling, and keeps the ledger of who owns which part of the balance.

Your convenience comes from the first. Your legal protection, if any, comes from the second. And the record of what you personally own frequently lives in the third — which is the part almost nobody checks, and the part that failed.

Three kinds of balance, three kinds of protection

The FDIC's own guidance sets the US conditions out plainly: pass-through coverage applies only if the funds are in fact owned by the customer rather than the third party who set up the account, the bank's account records indicate the agency nature of the account, and the records indicate the identities of the owners and their ownership interests. Miss any of those and the protection on the landing page does not reach you.

Where it goes wrong: the ledger, not the bank

Deposit insurance protects you when a bank fails. It does not protect you when the software company sitting between you and the bank fails. Those are different events, and the marketing rarely distinguishes them.

Synapse was a BaaS middleware provider serving consumer apps including the savings app Yotta. It filed for Chapter 11 bankruptcy in April 2024. The court-appointed trustee later reported that customers of the fintechs using Synapse had roughly $265 million in balances, while the partner banks held roughly $180 million against those accounts — a shortfall of up to $96 million, as reported by NBC News. Around 85,000 Yotta customers lost access to their money. Debit cards stopped working, transfers stopped processing, and some people waited more than a year for partial recovery.

No bank failed. Deposit insurance was never triggered, because it insures the failure of a chartered institution, not the collapse of a bookkeeping layer. The funds were pooled in accounts at real, solvent, insured banks — but only Synapse held the ledger saying which dollars were whose, and its bankruptcy destroyed the ability to reconstruct that. CNBC's account of the episode is worth reading in full if you keep savings in an app.

The lesson generalises well beyond one company: "FDIC-insured" on a fintech landing page describes where the money is parked, not who is obliged to prove it is yours.

The FDIC's response was a proposed rule in September 2024 requiring banks to keep their own records of each beneficial owner's balance in custodial accounts with transaction features, and to reconcile them daily. We have not found a final rule on the FDIC's site, so treat it as proposed, not in force.

Worked example: how much of a US app balance is actually insured

Say you hold $300,000 in a fintech app whose partner bank holds the money in a pooled custodial account, and the pass-through conditions are met.

• Your share is insured up to $250,000, leaving $50,000 uninsured if the bank fails.

• If you also hold $100,000 in your own name at that same bank, in the same ownership category, the two are added together: $400,000 against one $250,000 limit, so $150,000 is uninsured.

• If the middleware fails and the records of who owns what cannot be reconciled — the Synapse case — the $250,000 limit is irrelevant, because no bank has failed. What you get back, and when, depends on reconstructing the ledger.

The UK and EU version: safeguarding, and why the rules just changed

In the UK and EU, many embedded-finance balances are held by e-money or payment institutions rather than banks. Their protection is safeguarding — keeping customer money separate from the firm's own — not a deposit guarantee scheme.

The FCA has published how that worked in practice. Payment and e-money firms that became insolvent between the first quarter of 2018 and the second quarter of 2023 had shortfalls of 65% of their customers' funds on average, per the FCA's August 2025 announcement. In response, new rules took effect on 7 May 2026: daily checks that the right amount is safeguarded, monthly reporting, annual audits by qualified auditors, and better planning for returning money if a firm fails, set out in PS25/12.

Worked example: what a 65% shortfall means

Suppose you held £2,000 with an e-money firm that failed with a shortfall equal to that historical average. Customers would share what is left: 35% of £2,000 = £700. The average says nothing about your particular firm — some failed firms returned more, some less — but it shows why "safeguarded" and "insured" are not the same word.

The partner banks came under pressure too

The bank side of the model has also tightened. Blue Ridge Bank built a large BaaS business — around 50 fintech partnerships — and entered a consent order with the US Office of the Comptroller of the Currency in January 2024 over anti-money-laundering programme failings, which barred it from adding new fintech relationships without permission. It cut the partnerships back, rebuilt its risk function, and exited the business entirely; its chief executive Billy Beale put it as "we just threw BaaS out the door", per Banking Dive. The consent order was later terminated and the bank returned to conventional community banking.

That matters to you as a customer for one practical reason: when a partner bank exits, the apps built on top of it have to migrate to a new one, and migrations are exactly when access interruptions and record-keeping gaps appear.

What embedded finance is genuinely good at

None of this makes the category bad. The products that work best share a shape: the financial service is an extension of data the platform already holds, not a general-purpose bank account.

• Merchant and seller lending — a marketplace can see your sales history, so it can underwrite a working-capital advance faster and with less paperwork than a bank that starts from nothing.

• Instant payout of earnings — for gig and platform workers, being paid at the end of a shift instead of at the end of a fortnight is a real cash-flow improvement.

• Point-of-sale instalments — transparent, fixed-term instalments on a large purchase can be cheaper than revolving credit, provided the term is genuinely fixed and interest-free.

• Business accounts inside the tool you already run on — reconciliation is far less painful when payouts and bookkeeping share one system, which is why platform-native business accounts have stuck. Our fintech business account vs bank account guide covers what changes when one of them fails.

What tends to work badly is the general savings account with an unusually attractive rate, offered by a brand with no banking history, several intermediaries deep. That is the exact shape that failed.

Five questions worth asking before you keep money there

• Which regulated entity holds the deposit? It should be named, in plain text, on the app's own site. If you cannot find a bank's legal name in under two minutes, treat that as the answer.

• Is this a deposit or a stored balance? An e-money or payments balance is a different legal product from a bank deposit and usually carries safeguarding rules rather than deposit insurance.

• Who keeps the ledger? If the platform and the bank do not each hold a record of your individual balance, you are relying on a third party's bookkeeping surviving that party's insolvency.

• What is the recovery route if the app disappears? A good provider can tell you which bank to contact directly. If the only route back to your money is the app itself, the app is a single point of failure.

• How much would you be comfortable losing access to for six months? Not losing — losing access to. That was the real experience in 2024, and it is the more likely failure mode.

Frequently asked questions

What is embedded finance, with examples?

It is a financial product offered inside a non-bank's app, with a licensed bank or payment firm behind it. Examples are a marketplace lending to its sellers, a ride-hailing app paying drivers onto its own debit card, buy-now-pay-later at an online checkout, and a business account built into accounting or e-commerce software.

Is money in a fintech app FDIC insured?

Only in a specific sense. If the money is held at a partner bank in a custodial account that meets the FDIC's pass-through conditions, it is insured up to $250,000 per ownership category against that bank failing. It is not insured against the fintech or its middleware failing, which is what happened with Synapse in 2024.

What is the difference between embedded finance and Banking-as-a-Service?

Embedded finance is what you see: a financial product inside a non-bank's app. Banking-as-a-Service is the plumbing that makes it possible: a licensed bank, often through a middleware company, renting its licence and infrastructure to the platform.

Is an e-money account as safe as a bank account?

No. An e-money balance is safeguarded, not covered by a deposit guarantee scheme, and the FCA found that payment and e-money firms that failed between 2018 and 2023 had average shortfalls of 65% of customer funds. New UK safeguarding rules in force since 7 May 2026 are designed to reduce that.

Where to put the balance instead

For money you need reliably, the boring version is safer: an account with a bank that holds its own licence, where the protection scheme and the account provider are the same institution. Our guides on how to judge a neobank's safety, on what happens when a neobank fails and on what actually separates digital banks from traditional ones go through how to verify that, and the bank account switching guide covers moving without breaking direct debits.

If you are running a business rather than a household, the trade-off is different and often favours the platform-native option — see the best business neobanks for where that holds up.

The verdict

Embedded finance has made financial products more accessible and much better fitted to the moment you need them, and it is not going away. The consumer risk is not that the technology is flimsy; it is that the brand on the app and the institution holding the money are different, and the chain between them can be longer than it looks.

Use it for what it is good at — credit and payouts attached to a platform that already knows your business — and keep the balance you actually depend on with an institution whose name is on both the licence and the account. Understanding how neobanks make money and what open banking permits makes the rest of these arrangements much easier to read.

This is general information, not financial advice.