Why do 76% of digital banks continue to lose money?

Written by: Thejaswini M A

Translated by: Saoirse, Foresight News

A planned economy planner walks into a store, and the shelves are completely empty. He says, "You see, there is simply no demand." This is a long-standing joke among economists used to poke fun at the Soviet Union.

Today, new banks are trapped in the same vicious cycle. Hundreds of startups have launched checking account services, with a total of 1.4 billion people actually using them, but turning a profit from this business is as difficult as climbing to the sky. 76% of new banks are still operating at a loss. On average, each new bank earns only $45 per user per year, while traditional banks can earn up to $350.

The root cause lies in the products companies initially chose to create, as this type of business has almost no profit margin.

To understand the choices made by the practitioners, one must first acknowledge the flaws of the old system they wanted to escape.

Traditional banks constantly squeeze user benefits, even charging fees to withdraw their own salaries at ATMs. If you do not have much savings, the experience is even worse. When the first batch of new banks introduced accounts with no fees and no minimum deposit requirements, users naturally flocked to them.

Soon, hundreds of millions of users rushed to the platforms. Now, Nubank's services cover over 60% of Brazil's adult population. Local traditional banks have always viewed ordinary customers as nuisances, which has made the explosive growth of new banks inevitable.

However, these new banks find it difficult to sustain themselves.

When you swipe a debit card at a coffee shop, the merchant needs to pay a small transaction fee. According to the Federal Reserve's Regulation II, for a $40 transaction, the fee cap is about 22 cents, which is shared among the card organization, the bank, and the payment processing agency.

The profits that new banks receive are pitifully low. Millions of users treat their new bank accounts merely as everyday spending wallets, keeping their mortgages and investments with other institutions; the minuscule transaction fees accumulate but do not sustain a business.

The core of profitability in traditional banking has never been user everyday spending, as the revenue from transactions is merely a drop in the bucket.

The true profit pillar of the banking industry is credit, specifically the interest generated from loans such as mortgages and auto loans. Payment services are merely the daily point of contact for banks with users, and lending is the core means of profiting from users. This is also the root of the continued losses for the vast majority of new banks: without a banking license, they cannot issue loans at scale or collect interest. Most early new banks were built as technology platforms attached to the licenses of other banks and faced legal restrictions when it came to large lending operations.

Nubank started in Brazil in 2013 by launching a free credit card to open the market. At that time, large local traditional banks had outrageously high loan rates, providing Nubank with an opportunity for growth, and by 2026, it had accumulated 131 million users.

Today, Nubank is valued at $60 billion. The free account is merely a tool to attract users to download the app, with the true profits coming from lending operations.

Last year, the company generated $15.8 billion in revenue, with most coming from interest on credit cards and personal loans. The personal loan business has rapidly grown and become the largest profit segment. Nubank survives not due to revolutionary new technology, but due to lending; the user-friendly app is just bait to hook users.

Source: @sec.gov

Revolut has taken another profitable path. By 2025, the company’s net profit reached £1.3 billion, with revenue increasing 46% year-on-year to £4.5 billion, achieving five consecutive years of profitability. Profits mainly come from foreign exchange transaction fees, subscription memberships, cryptocurrency assets, and credit asset portfolios. The scale of lending grew by 120% year-on-year, totaling $2.9 billion. The early revenue from foreign exchange fees and subscription memberships gave it ample time to steadily expand its lending business.

Chime took the longest time to realize this truth. In its early years, it relied almost entirely on swipe fees for survival. With customer acquisition costs extremely high in the U.S., profits from card swipe divisions are minimal, entirely dependent on users continuously swiping their cards; once users reduced their spending, revenues plummeted.

By 2025, Chime's revenue exceeded $2 billion, yet it still incurred a loss of $1 billion, mainly due to high equity expenses associated with going public. The company was valued at $11 billion at the time of its IPO, but its stock price plummeted shortly after. It wasn't until the first quarter of 2026, after 12 years of operation, that it finally achieved profitability, with a net profit of $53 million. The turning point was the explosion in lending products: revenue from early wage access products was expected to exceed $400 million, while the scale of instant microloans skyrocketed.

In June 2026, a developer at Nubank accidentally activated a clearing process during a routine system update, causing mass notifications to be sent to users claiming that the central bank had cleared the bank and informing them how to claim funds through the deposit insurance fund. Co-founder Cristina Junqueira had to publicly apologize on Instagram, stating that it was merely a bizarre operational error and that both the bank and user funds were safe. However, in just a few minutes, the erroneous notification led users to believe that the platform was on the verge of collapse.

To be fair, traditional large banks often encounter such technical blunders, such as mistakenly transferring a billion dollars due to input errors. However, institutions like Citibank, which was established in 1812, have a solid foundation; even when issues arise, users simply consider them regular business mistakes. Yet once rumors of a collapse spread about a newly established digital bank, users will immediately panic and withdraw funds. Established banks may lag in technical capabilities, but emerging online platforms have yet to learn how to operate stably like actual banks.

In April 2024, the intermediary service provider Synapse declared bankruptcy.

New banks are essentially software service providers. To offer checking accounts, they must connect a whole cooperative industrial chain behind the scenes. Synapse acted as an intermediary, connecting hundreds of new banks with traditional banks that actually hold the funds, responsible for account management, compliance review, and asset title registration.

After Synapse's collapse, all business records were lost, and around $265 million of user funds were frozen. The cooperating banks could not determine the user ownership for each fund, and post-checks discovered that $95 million of funds were missing, with the entire system completely lacking an accountability mechanism. Several popular digital bank apps like Yotta and Juno saw users unable to operate their accounts for consecutive months, with some even unable to repay mortgages.

If a bank app relies entirely on third parties for fund custody and intermediate clearing processes that are not under their control, then the system is essentially a house of cards, destined to collapse.

Ultimately, the only safeguard against such systemic risks is a banking license. However, all the new banks previously claimed that they did not need a license at all.

Last October, I wrote that digital banks in the crypto field have genuine development potential. At that time, the regulatory framework was becoming increasingly clear, and many users held on-chain assets intending to use them directly for daily payments. This viewpoint still holds true today, but I gravely underestimated one thing: the foundational infrastructure built on cooperative banks will also carry all potential risks from their partners.

The crypto industry's response was to stop disguising itself and face reality. From December 2025 to May 2026, the Office of the Comptroller of the Currency (OCC) conditionally approved around ten national trust licenses aimed at crypto and fintech companies, exceeding the total number granted in the past decade. Paxos, BitGo, Fidelity Digital Assets, Ripple, Circle, and Stripe's $1.1 billion acquisition of Bridge and Crypto.com all submitted similar license applications – credentials that the new banks once scoffed at as unnecessary.

The national trust license is the ultimate solution to escape the intermediary trap. Holding such a license means receiving direct endorsement from the federal government, allowing companies to autonomously hold user assets and process payment clearing under a unified regulatory framework across all fifty states in the U.S. They no longer need to rely on the favor of traditional cooperative banks for survival, nor gamble their entire business's lifeline on invisible intermediaries like Synapse.

Crypto enterprises finally understand: if they want to circulate billions of dollars in assets without being constantly restricted by the underlying systems of traditional banks, they must obtain formal regulatory qualifications from the federal oversight system.

Kraken's parent company Payward currently holds three layers of regulatory qualifications in the U.S.: a Wyoming financial license, a Federal Reserve master account approved in March 2026, and an OCC national trust license application submitted in May 2026. SoFi acquired Golden Pacific Bancorp in 2022 to obtain the OCC license. In December 2025, SoFi launched a dollar-pegged stablecoin, the first stablecoin issued by a national bank in the U.S. built on an unlicensed public blockchain. By May 2026, 14.7 million users on the platform can hold, spend, and exchange this stablecoin within the app, with Mastercard becoming its clearing partner. Coinbase is conducting Bitcoin staking and lending business through the Base public chain, with collateralized Bitcoin exceeding $1.4 billion in early 2026.

SoFi's development trajectory is highly representative: student loan service provider → digital new bank → licensed formal bank → stablecoin issuer, completing the full evolution process of the industry.

Currently, there remains a major shortcoming in the industry: unsecured lending. The total scale of collateralized lending in CeFi and DeFi reaches $67.42 billion.

However, the actual scale of unsecured lending in the entire decentralized space is only $2.4 million. Protocols that once ventured into the unsecured lending sector (Goldfinch, early Maple, TrueFi) either fully transitioned to fully collateralized models or gradually shut down. Today, the largest DeFi lending protocol, Maple, has a collateralization ratio as high as 160%.

Blockchain addresses have anonymity features, and unsecured lending lacks viable default recovery mechanisms. In the real world, if a user defaults on a loan, banks can report to credit agencies or file lawsuits; in the decentralized realm, there are no credit agencies or asset recovery channels. Once a borrower escapes with unsecured assets, they only need to abandon their wallet address, and the funds can never be recovered. Some DeFi protocols have tried to manage risks using on-chain reputation data, yet major bad debts still emerged, leading practitioners to recognize: without real-world legal constraints, anonymous users have almost no motivation to repay debts.

Nubank issues loans to 131 million users, many of whom have no traditional credit records, and the platform relies on user transaction behavior to complete risk control and credit granting. This type of business has real commercial value, but the operational costs are extremely high, and the practical implementation is challenging. If one wants to replicate similar credit products on the blockchain at scale, companies will almost invariably need to obtain a banking license. It is expected that more and more companies will submit license applications to the OCC in the future.

Last October, I wrote that crypto digital banks are reproducing the development patterns of the banking industry from a century ago. Technology is always iterating, but the underlying logic of humans using and managing money remains unchanged. At the time I wrote this, I felt there was a beauty in the patterns, but looking back now, it presents another reality.

The essence of the banking industry is always to profit from lending by charging interest. The surviving new banks initially promised to break this model, yet the players that truly survive ultimately walked the path of lending - just with friendlier interest rates and smoother product interfaces; the underlying business logic remains unchanged.

In the end, one sentence sums it up: as the world changes, the essence remains the same.