Almost every business in crypto or next to it issues a card. The cards are all alike: debit, sometimes prepaid. The customer brings money, swaps it for crypto and spends it. He already has a card like that from his own bank.

I think this is a mistake in the very idea of the product. A bank makes money on credit. A debit card collects deposits, and a crypto service does not lend them out.

How banks grew on the credit card

Sberbank doesn’t count: it inherited its branch network and deposits from the Soviet Union. The rest were built from scratch.

Russian Standard started with point-of-sale loans in stores, then mailed cards to people who had paid off their loan. In 2006 it issued 99 billion rubles in card loans and 79 billion in consumer loans. The card rate was 79% a year. Tinkoff was built from 2007 with no branches, on credit cards sent by mail. By the end of October 2010 it had sent 45 million letters and issued 740,000 cards, 1 card per 60 letters. In April 2011 it issued its millionth. The Tinkoff Black debit card appeared in September 2012, on top of the base the credit cards had gathered.

From roughly 2005 to 2015 you could get a credit card in Russia on every street corner. Then the model ran into the retail credit crisis: Russian Standard restructured its debt and never got back to its former size, Tinkoff came through the crisis. Both bases were built before it.

Nubank in Brazil started in 2014 with a no-annual-fee credit card managed from an app, and there was a waitlist for the card until 2019. It added an account in October 2017, when it had 2.5 million cardholders. The card brought in the first customers, the account grew from there, and in May 2024 the customer count reached 100 million.

Why people take a credit card

People need money. A credit card gives it to them, and that is what they come for. A debit card from yet another service gets opened for the cashback and dropped when the cashback runs out. Nobody publishes retention for crypto cards, so this is my own observation. On demand there is one figure: when Binance shut down its card in Europe in December 2023, it said about 1% of its customers had used it.

There is a line I like: nobody cares about your ability to make a pile of money, but plenty of people care about your ability to change the world. A debit card is the service’s ability to make money, a percentage of each purchase made with the customer’s own money. A credit card changes the customer’s position: he now has money he did not have. Russian Standard and Tinkoff made credit an everyday thing in a country that had not had it, and got their customer base in return.

A debit card lives on interchange. In Europe it is capped at 0.2% per purchase; in the US, for smaller banks, it is about 1.4%. Revolut and Chime grew on debit. Revolut on currency exchange with no markup, and the card brings it about 22% of revenue. Chime on US interchange. A crypto card in Europe has neither.

A credit card earns interest on the outstanding balance minus the cost of risk. Russian Standard’s 79% is an extreme, and in 2015 its cost of risk reached 26% and ate everything. But even 20% a year on the balance, with normal losses, brings in more than 0.2% of turnover.

What crypto issues

Crypto.com, Coinbase Card, Binance, Gnosis Pay, dozens of smaller cards. Unsecured credit cards exist too. I know of 3, one is already closed, and each is built differently. BlockFi issued a credit card from July 2021, with 87,000 cardholders. The issuer was Evolve Bank, the risk and the debt sat with the program manager Deserve, and BlockFi was the brand. Since October 2025 Coinbase has issued a credit card on the American Express network through First Electronic Bank and the Cardless platform; who holds the risk is not disclosed. Gemini launched its card in April 2022 through WebBank. According to its IPO prospectus and filings, it buys the entire portfolio from the bank, keeps the risk on its own books and funds it with a credit line. Without a banking license you cannot lend out customer deposits.

So Gemini has spent 4 years doing what I recommend, and the thesis has to be tested on it. Per its Q1 2026 report, the portfolio is $218 million, up from $69 million a year earlier, with 114,000 monthly active cardholders; how many cards have been issued, Gemini does not disclose. For the quarter: interchange 10.3 million, interest 5.8, crypto rewards 8.4, funding 2.5. Then provisions of 8.6, of which 4.1 was one-off fraud, and the bottom line is minus 4.9. Delinquency over 30 days is 3.8%.

Gemini has built the infrastructure. Its interchange is larger than its interest, its rewards are larger than its interest, the card so far works as a debit card with rewards, and Gemini itself calls it an acquisition channel for the exchange. I see 2 explanations. First: it sells the card in the US, to people who already have a credit card in their wallet, and they come for the reward. Second: a 3.8% delinquency rate at this yield is the result of underwriting and the price of the rewards. From the outside you cannot separate the two, and both say the same thing: the card was sold as a reward. Tinkoff and Nubank sold a credit limit where there had been none.

A separate branch is credit against coins. Since 2022 the Nexo card has been backed by a credit line secured by the customer’s portfolio, up to 50% of its value for bitcoin, as in DeFi loans. No scoring needed: when the debt gets close to the value of the collateral, the collateral is sold. A line like this does not bring in new customers: with no coins in the account, it gives no money. In August 2023, 16 months in, Nexo added card payments from the balance, without credit.

What stands in the way

The license. Russian Standard, Tinkoff and Nubank are banks, and they funded credit with deposits. In the US a credit card is issued through a partner bank that sells you the portfolio, but you have to fund that portfolio with a credit line: for Gemini it is 2.5 million a quarter, about 5% a year on the portfolio by my arithmetic. That is the price of working without a license. In Europe consumer credit is regulated separately, and a credit card there is issued only with a license or through a partner that has one.

Risk. In 2022 the lending businesses of Voyager, Celsius and BlockFi went bankrupt, followed by Genesis. They fell on deposits: loans to funds, their own trading, yields promised to depositors. Card debt is not on that list. But when BlockFi froze deposits, holders of its credit card stopped paying on it until they got their money back. Deposits kept separate from credit, or a reserve for a deposit freeze. Of the 3 crypto credit cards, only Gemini holds the risk itself; BlockFi passed it to the program manager along with the interest; Coinbase does not disclose.

A question for whoever is building a card

Does the customer bring money to this card, or get money from it? If he brings it, I don’t know of a crypto card the size of Tinkoff or Nubank.

If he gets it, based on scoring and with the risk on your balance sheet, this is the product Tinkoff and Nubank grew on. For a team with a debit card that means: a partner bank with portfolio buyback, credit bureau scoring, a growing limit, provisions from day one, people on collections, and deposits separate from credit. And a market where people line up for a limit.

Where this thesis may be wrong. If your market is the US or Western Europe, the customer already has a credit card and will not take a limit from you: that is the Gemini case. If the regulator closes unsecured credit to companies without a license, what remains is collateral, and the card becomes a product for holders. It will grow at the speed of their inflow, like Nexo.