Skip to content
Velloreth
Financial Technology 9 min read

The Rise of Neobanks: What They Get Right

Neobanks spent a decade proving that a better app is not a bank. Now that the survivors are profitable, the actual lessons are clearer — and fewer than the pitch decks claimed.

By

The first decade of neobanking produced a simple story that turned out to be half wrong: banks are slow, apps are fast, so app-first banks will eat the incumbents. The second decade produced the correction. Some app-first banks — a few, in specific markets — did become genuinely large and profitable. Most did not, and the reasons why are more instructive than the hype that funded them.

This post is about what the survivors actually got right, separated from what the pitch decks claimed.

What a neobank is, precisely

The term covers three structurally different things that happen to ship similar-looking apps.

Licensed digital banks. Nubank in Latin America, Monzo and Starling in the UK, Revolut (eventually) in the UK and EU. These hold banking licenses, take deposits, and carry the full regulatory stack — capital requirements, deposit protection, prudential supervision.

Charterless fintech fronts. An app and a card, running on a partner bank’s license through banking-as-a-service (BaaS) infrastructure. The customer sees a fintech; the deposits legally sit at a partner bank; compliance and settlement run through middleware providers. This was the majority model by count in the US and parts of Europe, and it produced both the best UX of the era and the most spectacular failures.

Specialized deposit-takers. Business banking (Mercury, Brex’s evolution, Qonto in Europe), teen accounts, freelancer banking — neobanks that narrowed the segment until the unit economics worked.

The profitable end of the category is dominated by the first and third types. The BaaS middle has largely been consolidating or dying since the mid-2020s, after several high-profile infrastructure failures — including a settlement breakdown at a major BaaS middleware provider that froze customer funds and took dependent fintechs with it — made the dependency structure impossible to ignore.

What the survivors got right

1. Free, honest core accounts

The durable consumer proposition was never a novel feature. It was: a current account with no monthly fee, no minimum balance, transparent FX, and a card that arrives quickly. For customers in markets where incumbent banks charged monthly fees for the privilege of holding money, or refused accounts to people without salary credits, this was not a nice-to-have.

Nubank’s original product — a no-fee credit card with an app that actually worked — succeeded in Brazil because the incumbents’ card fees and branch-centric service were genuinely bad for most of the population. Monzo and Starling won UK current accounts on the same logic against high-street banks with legacy IT.

The lesson generalizes: neobanks win where incumbent pricing or service is broken for a mass segment, and lose where incumbents are already decent and the switching cost is habit.

2. Onboarding measured in minutes, not visits

Every serious neobank treated account opening as a funnel to be optimized — document scan, selfie, decision, virtual card ready to spend before the physical card arrives. Incumbents, where digital onboarding existed at all, routinely lost half their applicants to multi-day verification loops.

This mattered most in underbanked markets. Getting a first bank account in Brazil or Mexico historically meant branch visits, paperwork, and credit history most people did not have. Nubank and its peers onboarded tens of millions of people who had never had a bank relationship, and the growth numbers — Nubank crossed 100 million customers years before most analysts predicted — came from that unbuilt segment, not from stealing incumbents’ best customers.

3. Distribution economics that scale with software

A branch acquires customers at fixed physical cost. An app acquires them at near-zero marginal cost, with referral loops, social sharing, and product-led growth baked in. When the product is genuinely better, word of mouth compounds. Nubank’s marketing spend per customer has historically undercut incumbents’ acquisition costs by a wide margin, and its efficiency ratio — a rough proxy for cost per unit of revenue — has sat in territory most banks consider impossible.

4. Product velocity as a moat

Shipping features weekly against incumbent release cycles measured in quarters is a real advantage, when the features matter. Real-time spending notifications, instant card freeze, savings “pots,” salary advance, FX at transparent rates: none individually is a defensible moat, but the aggregate creates a product experience gap that customers notice every day. The moat is organizational — the ability to keep shipping — more than any single feature.

5. Credit, carefully, at scale

The neobanks that reached durable profitability all built credit businesses. Nubank’s Q2 2026 numbers make the point cleanly: 139 million customers, quarterly gross revenue approaching $6 billion, and over $39 billion in credit outstanding. Roughly two-fifths of gross profit came from credit. Net income crossed $1 billion in a single quarter for the first time, with return on equity at 33% — numbers that put it in the territory of the most profitable traditional banks, not fintechs.

The important qualifier: profitable neobank credit is underwriting against transaction data the bank itself sees. Nubank’s investment in its own underwriting models — the company publicly attributes much of its risk performance to a proprietary model trained on its transaction graph — is the quiet engine behind the headline growth. Neobank credit books that skipped that discipline and underwrote against marketing cohorts instead took losses that ended them.

What the survivors did not prove

Several widely repeated claims from the 2015–2022 era have since been contradicted by results.

“Incumbents cannot respond.” They can and did. Incumbent mobile apps improved, free-account offerings matched, and switching incentives kept most profitable customers in place. Neobank growth in saturated markets has been real but slower than early projections, because the pool of people badly served by incumbents turned out to be smaller in the US and UK than in Brazil or Mexico.

“The app is the moat.” It is not. Apps are replicable; regulatory licenses are not. Every serious neobank either obtained a banking license or accepted permanent dependence on someone who did — and dependence turned out to be existentially fragile, as the BaaS failures showed. Deposits, charter, and capital are moats. UX is a feature.

“Fees are dead.” Zero-fee core accounts survived, but monetization crept back: premium subscription tiers, overdraft and interest-adjacent products, FX spreads on the free tier’s less-transparent paths. The pure no-fee-everything model needed a revenue engine elsewhere, and every survivor found one — usually credit.

“Deposits are free growth.” Deposits are funding, and funding has costs and risks: interest expense when rates rise, flight risk when rates on alternatives climb, and prudential obligations that grow with the balance sheet. Several neobanks discovered that deposit growth without a lending business just means paying to hold cash.

What the incumbents actually learned

The standard narrative has incumbents sleepwalking through the disruption. The reality, in most markets, was faster and more competent than that.

High-street banks watched the onboarding funnel data and fixed the worst of it: digital account opening improved, apps were rebuilt (often by the same engineering talent the neobanks had hired), and free-tier accounts appeared where fee pressure demanded. Switching guarantees and current-account switch services made the mechanical part of moving banks easier. None of this produced dramatic market-share shifts — bank customers are sticky in every country — but it removed the easiest points of comparison. A neobank in 2026 has to win on product depth or segment focus, not merely on “our app does not feel like it was designed in 2009.”

The more consequential incumbent response was building or buying the rails-side capabilities: real-time payment features, savings products with competitive rates, and venture arms that invested in or partnered with the fintechs rather than only fighting them. In several markets, the end state is not neobank victory but hybrid: neobank-grade UX attached to incumbent-grade balance sheets — sometimes literally, as regulated banks partner with B2B fintech platforms to power white-label digital offerings.

This is the mature reading of the decade: neobanks did not replace banks. They forced a repricing of attention — incumbents had to spend engineering budget on consumer experience or bleed the demographic that notices — and they proved, permanently, that a banking license plus modern software is a formidable institution. The license was always the point. The app was the packaging.

The market map, honestly

MarketOutcomeWhy
Brazil / MexicoDominant neobanks (Nubank)Large underbanked population, expensive incumbents, card interchange economics
UKTwo profitable banks (Monzo, Starling), Revolut scaled via breadthSaturated but fee-broken segments; licensing eventually obtained
USMixed: profitable specialists, BaaS casualtiesStrong incumbents, regulatory complexity, interchange economics favorable but competition brutal
EUFragmented; strong vertical playersNational licensing, varied regulation, no single mass market
South AsiaWallets and bank apps, not neobanks per seFast-payment rails + wallet licensing created a different path to the same consumer need

The South Asia row deserves a note, because it is the most instructive for anyone wondering where “neobanking” goes in emerging markets outside Latin America. Nepal never needed a neobank to give people a digital account: licensed wallets (eSewa, Khalti by IME, and others) plus bank mobile banking apps, connected through shared infrastructure like NCHL’s Retail Payment Switch and the National Payment Interface, cover most of what a neobank’s core account does — payments, top-ups, bill payments, QR acceptance — without taking deposits under a new charter. The consumer outcome (bank-like digital access without branches) arrived through a different institutional path: shared payment rails and wallet licensing rather than greenfield bank charters.

That path has a lesson for the neobank thesis: the app-first part was right everywhere; the “must become a bank” part was only necessary where the existing rails did not already solve payments. Credit is where the charter still matters — wallets lend cautiously under their license frameworks, but the large credit books of Latin American neobanks came from full banking licenses with deposit funding.

What to take from the decade

If you are building, investing in, or regulating in this space, the distilled lessons are short.

  1. Find a market where incumbents are genuinely failing a mass segment. In Brazil, that was card fees and unbanked adults. In saturated markets with decent incumbents, app polish alone does not move deposits.
  2. Get a charter, or accept fragility. BaaS dependence is a business risk disguised as a cost line. The mid-2020s showed the failure mode clearly.
  3. Credit is the profit engine; treat it like an underwriting company, not a growth feature. The neobanks with 33% ROEs are effectively very efficient lenders with a superior acquisition app.
  4. Interchange is not a strategy. It funds the early stage; it does not fund a bank. Every durable neobank layered net interest income and fee products on top.
  5. The distribution advantage is real but decays. As incumbents improve their apps, the switching reason has to keep being earned — with products, not marketing.

Neobanks did not eat banking. They forced it to get better at the consumer end, proved that licenses plus data plus software produce genuinely formidable institutions in the right markets, and killed a few hundred companies that confused an app with a balance sheet. That is a more useful legacy than the revolution anyone sold in 2018.

Frequently asked questions

Are neobanks actually banks?
Some are — Nubank, Monzo, Starling, and Revolut (after a long licensing haul) hold bank licenses. Many others are fintech apps operating on a partner bank's charter through banking-as-a-service, which means the customer relationship sits in the app but the regulatory stack sits somewhere else. The difference shows up in exactly one place: who holds the deposits when things go wrong.
How do neobanks make money with no fees?
Three engines, in rough order of importance: interchange on card spend, net interest margin on deposits and credit products, and fee-based services (overdrafts, FX, premium subscriptions). The ones that scaled built credit businesses; the ones that stalled treated interchange alone as sufficient.
Why did so many neobanks fail?
Unit economics. A free account with a card costs real money to run — fraud, KYC, support, card issuing — and generates thin interchange. Without credit products or large deposit balances to lend against, growth made losses bigger, not smaller. A wave of banking-as-a-service failures in the mid-2020s took dependent fintechs down with them.
Will neobanks come to Nepal?
Nepal already has the functional equivalent: licensed wallets and mobile banking apps that handle most consumer banking tasks without branches. Full neobank licensing would require NRB authorization, deposit-taking capability, and the capital that implies — a different game from the wallet layer, where the market is already competitive.