Author: Thejaswini M A Translator: Shan Ouba, Golden Finance

There is an old joke about the planned economy: a planner walks into a store and the shelves are empty, and he says: "Look, there is no demand." This is a joke that has been circulated among economists for a long time to make fun of the Soviet Union.

Today, the digital banking industry is falling into exactly the same cycle of death. More than 500 start-up companies around the world have launched debit card current accounts, attracting a total of 1.4 billion users to actually use them. Making money is extremely difficult.

76% of digital banks are still unprofitable; a single digital bank generates an average of $45 per user per year, compared with $350 for traditional banks.

The root lies in the product track initially selected by the start-up company, and whether there is any real profit potential in this track.

To understand the original choices of each company, we have to see clearly the ills of the old industry that they wanted to escape from.

In the early years, traditional banks exploited users at all levels. Even when withdrawing their own wages, ATM machines charged handling fees. If you come from an ordinary background and have little savings, your experience will only be worse. After the first batch of digital banks launched accounts with zero handling fees and no minimum deposit threshold, users naturally voted with their feet.

In just a few years, hundreds of millions of users have poured in. Today Nubank covers more than 60% of Brazil’s adult population. Traditional banks have always treated ordinary customers in a perfunctory manner, and it is almost inevitable that digital banks can achieve explosive growth.

But this business model will never work.

When you spend $40 with your debit card at a coffee shop, the merchant pays a transaction fee. According to the Federal Reserve's "Regulation II", the upper limit of this fee is about 22 cents, and the revenue is divided among the three parties of the card organization, the card issuing bank, and the payment clearing agency.

The profits shared by digital banks are minuscule. Millions of users only use it as a daily consumption account, and their mortgage and financial management are all placed elsewhere. The meager handling fees cannot support a business at all.

Traditional banks do not make money by swiping cards, and transaction fees are only a meager supplement. The core profit pillar of the banking industry is always credit: interest generated from various types of loans such as home loans and car loans. The payment business is only the daily channel for banks to reach users, and the lending business is the core of reaping profits.

This is also the crux of the long-term losses of digital banks: without a banking license, it is impossible to lend on a large scale and collect interest. The vast majority of digital banks were initially just technology platforms built with traditional banking licenses, and were subject to legal constraints in their credit business.

In 2013, Nubanco started in Brazil, focusing on free credit cards. At that time, the loan interest rates of the local traditional major banks were ridiculously high, and this product quickly emerged from the circle. By 2026, the number of users exceeded 131 million, and the company's valuation reached US$60 billion.

Free accounts are just tools to attract customers, and all real profits come from credit business. Last year, NU Bank's annual revenue was US$15.8 billion, with the vast majority of revenue coming from credit cards, personal loans and interest income. Among them, personal loan business has the fastest growth rate and has become the largest profit segment. NuBank can survive not by new technology but by lending; the smooth and beautiful APP is just a bait to attract users to join the game.

Revolut has found another path to profitability. The company's net profit will reach 1.3 billion pounds in 2025, and revenue will increase by 46% year-on-year to 4.5 billion pounds, achieving profitability for five consecutive years. Profits mainly come from foreign exchange fees, membership subscriptions, cryptocurrency business and credit assets. Credit volume grew 120% year over year to total $2.9 billion. Relying on the stable cash flow brought by its early foreign exchange and membership businesses, Revolut has enough time to lay out the credit sector in a low-key manner.

Chime took the longest to figure out this logic. Previously it relied almost entirely on card processing fees to survive. The cost of acquiring customers in the United States is extremely high, and the profit from single-card processing fees is extremely thin. Once users stop spending, revenue will directly return to zero.

Chime’s revenue will exceed US$2 billion in 2025, but it will still lose US$1 billion, mainly due to the huge equity incentives from the listing. The company was listed at a valuation of US$11 billion, and its stock price plummeted within a few months of listing. A turnaround occurred in the first quarter of 2026. The company made a profit of US$53 million for the first time in 12 years. The core driving force was the explosion of credit products: the annual salary advance business revenue is expected to exceed US$400 million, and the scale of the instant small loan business has skyrocketed.

In June 2026, a developer of Banco Nuo mistakenly triggered the liquidation process during a routine software update. The system sent push notifications and emails to a large number of users, falsely claiming that the Brazilian Central Bank had liquidated the bank and informing users how to apply for funds through the Deposit Insurance Fund. Co-founder Christina Junqueira had to publicly apologize via Instagram, saying that this was just an outrageous operational error and that both bank and user funds were safe.

In just a few minutes, a simple mistake led users to mistakenly believe that the company was on the verge of bankruptcy. Objectively speaking, traditional major banks often make similar mistakes, such as input errors leading to misdirection of billions of dollars. However, a century-old bank like Citigroup was established in 1812 and has a deep foundation. Even if the system fails, users will only regard it as an ordinary corporate error. However, once risks are reported in a start-up digital bank, users will immediately panic and run.

In April 2024, the intermediary service provider Synapse declared bankruptcy, completely exposing the fatal loopholes of the no-own-license model.

Digital banks are essentially just software companies. If they want to launch current accounts, they must connect an entire cooperative supply chain behind them. Synapse is an intermediary that connects hundreds of digital banks with traditional banks that hold funds, and is responsible for accounting, compliance review, and asset confirmation.

After Synapse collapsed, customer fund records were lost, and approximately $265 million in user funds were frozen. The cooperative bank was unable to distinguish the ownership of the funds. Afterwards, verification revealed that US$95 million in funds were missing. The entire system completely lacked an accountability mechanism. Mainstream digital banks such as Yotta and Juno have been suspended for several months, and some users are even unable to repay their mortgages normally.

If your digital bank does not have its own banking entity and relies on third-party intermediaries, the system may seem perfect, but in fact it is fragile and may collapse at any time.

In the final analysis, the only trump card that can resist this type of systemic risk is a bank license. In the past, digital banks always claimed that they did not need a license at all.

In October last year, I wrote an article and analyzed that encrypted digital banks have real potential for implementation: the regulatory framework is gradually becoming clearer, and users hold on-chain assets and hope to use them directly for consumption. This set of judgments has been established so far, but I seriously underestimated the chain risks brought by "depending on the underlying structure of third-party banks".

The response plan for the encryption industry is to completely abandon the OEM cooperation model and apply for compliance qualifications independently. From December 2025 to May 2026, the U.S. Office of the Comptroller of the Currency (OCC) has "conditionally approved" ten national trust licenses for encryption and financial technology, which exceeds the total number of the past ten years. Paxos, BitGo, Fidelity Digital Assets, Ripple, Circle, and Bridge, which Stripe spent $1.1 billion to acquire, all applied for a U.S. national trust license that digital banks initially scorned.

The national trust license is the ultimate way out of the middleman trap: the company directly obtains official federal qualifications, can independently custody user assets, handle payment and settlement, and has unified compliance standards in all 50 states across the United States. There is no need to look at the faces of traditional cooperative banks, and they will not bet all their business on invisible intermediary service providers like Synapse.

Encryption companies have finally recognized the reality: if they want to circulate billions of funds without being constrained by the underlying structure of traditional banks, they must be included in the federal formal financial system and obtain compliance licenses.

Payward (the parent company of Kraken) has now gathered three types of US compliance qualifications: Wyoming bank license, approval of the Federal Reserve master account in March 2026, and submission of OCC national trust license application in May 2026.

SoFi acquired Golden Pacific Bank as early as 2022 and obtained an OCC banking license; in December 2025, it issued a U.S. dollar stable currency, becoming the first nationally licensed bank in the United States to issue stable currency based on an unlicensed public chain. By May 2026, 14.7 million platform users can hold, consume, and exchange the stablecoin in the APP, and Mastercard is its clearing partner.

Coinbase relies on the Base public chain Morpho protocol to carry out Bitcoin pledge loans, and the scale of mortgaged BTC exceeded US$1.4 billion in early 2026.

SoFi’s development path is very representative: student loan service provider → digital bank → licensed formal bank → stablecoin issuer, completing the entire industry chain.

The current total scale of centralized and decentralized crypto-collateralized lending is US$67.42 billion, while the volume of DeFi unsecured lending is only US$24 million. Protocols that once laid out a mortgage-free track (Goldfinch, early Maple, TrueFi) have either fully shifted to a fully mortgaged model or are on the verge of closure.

Today, Maple, the leading DeFi lending protocol, has a collateralization rate as high as 160%.

In the anonymous public chain environment, unsecured lending lacks an effective default recovery mechanism. In reality, if a user's loan is overdue, the bank can report the credit report and file a lawsuit; DeFi does not have a credit report agency or asset collection channel. The borrower directly abandons the wallet address, and the loaned funds can never be recovered. Many protocols have tried to build a reputation scoring system, but in the end they still encountered large-scale bad debts - lack of realistic legal constraints, and anonymous users have no incentive to repay.

Nubank lends to 131 million users without credit records and relies on user transaction behavior for risk control and approval. This model has real commercial value, but the operating costs are extremely high and it is difficult to implement. If you want to implement similar credit services on a large scale on the blockchain, companies will almost certainly need a banking license. The U.S. Office of the Comptroller of the Currency license application queue will only get longer in the future.

Last October I wrote that encrypted digital banks are retracing the path taken by the banking industry a hundred years ago: technological iterations never stop, but the underlying logic of human beings dealing with money has never changed. I originally thought that this set of rules had a sense of industry destiny, but now it shows a more straightforward reality:

Traditional banks always charge high fees for lending. The digital banks that broke through in the early years promised to end this chaos; and the surviving companies eventually embarked on the road of lending - only with more friendly interest rates, smoother product interfaces, and the same underlying business logic.

Every change remains true to its roots.