Neobank vs Traditional Bank: How the Business Models Differ

A neobank is an app-first financial company that usually is not itself a bank: it owns the brand and the customer, while a chartered partner bank holds the deposits. A traditional bank holds its own charter and deposits, and earns most of its money from the spread between what it pays depositors and what it earns on loans and securities. The largest US neobank, Chime, earns mostly from debit-card interchange paid by merchants; JPMorgan Chase earns mostly from net interest income. The two can serve the same customer and make money from that customer in almost opposite ways.

The difference shows in the results. Chime reported about $2.2 billion of revenue in 2025 and its first GAAP-profitable quarter in the first three months of 2026. JPMorgan reported about $182.4 billion of revenue and about $57.0 billion of net income in 2025. Varo, the first consumer fintech to win a full national bank charter, lost about $92 million in 2025. This comparison explains why.

Neobank vs traditional bank at a glance

How does a neobank make money?

Chime's founders, Chris Britt and Ryan King, set out in 2012 to build an account for people living paycheck to paycheck: no monthly fee, no minimum balance, no overdraft fee. A traditional bank earns from such customers mainly through overdraft and maintenance fees. Chime would earn from merchants instead.

The mechanism is a rule. Under the Durbin amendment, debit interchange at banks with $10 billion or more of assets is capped at 21 cents plus 0.05% of the purchase plus a one-cent fraud adjustment. Smaller banks are exempt and earn roughly two to three times the regulated rate on a typical purchase (practitioner estimates). Chime's partner banks, The Bancorp Bank and Stride Bank, were both under $10 billion, so they could earn the higher rate and share it with Chime. The book's Chapter 13 on fintech and neobanks calls this "the Durbin arbitrage."

Everything else follows from the arithmetic. Interchange is earned only when customers spend, so Chime needed to be their primary account, the one their pay lands in. It paid wages up to two days early, launched the SpotMe fee-free overdraft cushion in 2019 and offered a secured card to build credit. By the end of 2025 it had 9.5 million active members and about $257 of revenue per active member a year.

How does a traditional bank make money?

A traditional bank takes deposits and lends them out or invests them. The US industry earned a net interest margin of about 3.32% on its earning assets in the second quarter of 2026, according to the FDIC, and a return on assets of about 1.37%. Our explainer on net interest margin explains how that spread works.

Scale matters enormously. JPMorgan opened about 1.7 million net new checking accounts in 2025, as reported, more than the entire customer base of most regional banks. A mobile app, a fraud system and a national brand campaign cost about the same whether they serve 5 million customers or 80 million. Community banks, 3,818 of the 4,238 insured institutions in mid-2026, run efficiency ratios of roughly 60–70%, against the low to mid 50s at JPMorgan. The book's commercial and retail banking study describes the tiers.

The traditional bank's great advantage is the deposit franchise: the ability to pay depositors less than market rates for years and lend the money at a spread. The neobank on a partner bank does not usually get that spread; the partner bank keeps the deposit income unless the contract shares it.

Which is safer: a neobank or a traditional bank?

For an ordinary depositor, the difference is mainly in the plumbing. At a traditional bank, deposits are insured directly by the FDIC up to the limit. At a neobank, the money sits at the partner bank, often in pooled "for benefit of" accounts, and pass-through insurance applies only if the records identify each customer and amount.

That condition matters. When the middleware company Synapse collapsed in 2024, more than 100,000 users of apps built on it lost access to about $219 million, and the trustee found a gap of $65–96 million between what users were owed and what the banks held. No bank failed, so deposit insurance paid nothing. Our case study on what happened to Synapse tells the story. Chime always contracted directly with its banks and used no middleware layer, which is a different structure.

Traditional banks have their own failure mode: a run. Silicon Valley Bank lost about $42 billion of mostly uninsured deposits in one day in March 2023. Our list of the biggest bank failures in US history shows that the large failures came from banks with concentrated, uninsured funding and long-dated assets, not from size alone.

Should a fintech get a bank charter?

This is the central strategic choice in US fintech, and the book's cases point both ways.

The lesson is not that charters are wrong. A charter brings obligations a partner bank would otherwise carry: capital ratios, an anti-money-laundering program, consumer compliance, Community Reinvestment Act duties and examinations by up to three federal supervisors. Those costs are fixed. Varo's customers kept an average of about $30–40 per insured account, as reported, and its on-balance-sheet deposits were about $160 million in the third quarter of 2025 for a bank with more than 6 million accounts. A bank with so few deposits cannot fund much lending, so the spread income that justified the charter never arrived at scale.

SoFi had the opposite profile: a loan book large enough to earn the spread, and members who set up direct deposit to earn a higher savings rate, which made its deposit base mostly insured and relatively sticky. In the book's phrase, a charter is a cost before it is a moat.

How do the unit economics compare?

The book's research models two made-up neobanks over three years with the same 10,000 customers. "Neobank A" is a partner-bank model that keeps 80% of interchange and none of the deposit income. "Neobank B" is licensed: it keeps all its interchange, earns a spread on deposits and lends to 30% of its primary customers.

The licensed model earns more per account and survives a cut to interchange, at the price of capital, compliance and credit risk. The partner-bank model is cheaper to launch and depends on the exemption. Our guide to evaluating fintech unit economics walks through the sensitivities.

What could change the comparison?

Three things were in motion as of late 2026. First, the Durbin cap was vacated by a North Dakota federal court in August 2025, and the Eighth Circuit affirmed on 26 October 2026; a Sixth Circuit case was argued on 22 October 2026. If the cap is removed or reset, the gap between large and small banks' interchange, on which most US neobanks are built, would change. Second, regulators granted a run of charter approvals: Mercury in April 2026, Upstart in July 2026, and conditional approval to Nubank in January 2026, with Revolut winning preliminary conditional approval in September 2026, as reported. Third, after the 2022–24 wave of consent orders against partner banks, sponsor banks are keeping the customer ledger themselves.

The likely result is convergence. Neobanks are buying or applying for charters, and the largest traditional banks keep building better apps.

The takeaway

A traditional bank earns mainly from the spread on deposits it holds under its own charter; a US neobank has mostly earned from debit interchange rented through a small partner bank. The neobank model is fast and cheap to launch but depends on an exemption and a partner, while the bank model earns more per customer but carries capital, compliance and run risk. In money-equation terms, neobanks have competed on the third factor, reaching customers banks served poorly at a cost those customers' interchange could repay, and the charter question is about whether they can also own the fourth. For the wider picture see our guide to how financial companies make money, and read Chapter 13, "Fintech and Neobanks" for the full cases.