SwipeStables  
All posts

Reports

What a year on a crypto card costs

Six of the 57 live cards in release 2026-09-02.1 publish enough for us to price a year of ordinary use. For the other 51 at least one fee is missing from the issuer's own documents, so no yearly number exists — and we do not invent one. This report gives the six numbers, the arithmetic behind them, and a named list of what the other 51 leave unpublished.

Written by SwipeStables · last updated · computed from release 2026-09-02.1, which holds 1598 recorded values across 77 cards. Every date below is the day we observed the document.

The six cards that can be priced

Cost over 12 months at $1,000 a month, of which 20% is spent in a foreign currency and none is withdrawn at an ATM. A negative number means the base cashback exceeds the fees: the card pays you. Where a fee is published as a range, the year is a range too.

Coinbase One Cardearns $240$120 at $500/mo · $720 at $3,000/mo. No issuance, no annual fee, 0% foreign transaction fee, 2% base cashback in BTC.coinbase.com2026-08-22
Fold Bitcoin Credit Cardearns $180$90 at $500/mo · $540 at $3,000/mo. No issuance, no annual fee, 0% foreign transaction fee, 1.5% base cashback in BTC.foldapp.com2026-08-23
Crypto.com Visa Signatureearns $108$54 at $500/mo · $324 at $3,000/mo. 1.5% base cashback capped at $250 a month, less a 3% foreign transaction fee — the FX fee eats two thirds of the reward at this foreign share.comenity.net2026-08-23
MetaMask Virtual Cardearns $96 to costs $9−$48 to +$4.50 at $500/mo · −$288 to +$27 at $3,000/mo. The spread is which asset you spend: MetaMask publishes 0% conversion on a stablecoin denominated in your own currency, 0.5% on a stablecoin in another currency, and 0.875% on other crypto such as wETH. At 0% the 1% cashback wins; at 0.875% it does not.support.metamask.io2026-08-22
Binance Cardcosts $60 to $84$30–$42 at $500/mo · $180–$252 at $3,000/mo. Brazil programme: 0.9% top-up plus 0.9% conversion on everything spent, 1–2% FX, against 1.5% base cashback.binance.com2026-08-22
RedotPay Standardcosts $159 to $249$84–$174 at $500/mo · $456–$546 at $3,000/mo. The $90 spread is the card itself: RedotPay publishes $10 for a virtual card and $100 for a physical one. On top of that, 1% conversion on everything spent and 1.20% FX. RedotPay does not publish a current cashback rate for the Standard card, so none is subtracted.redotpay.com2026-08-22

The three cards at the top are US credit cards, and a credit card only stays free if the statement is paid in full. Fold publishes a purchase APR of 20.24–29.24%; carry a balance and the interest is larger than anything in this table. Interest is not a card fee, so it is not counted here.

One gap we are carrying openly: on the Fold and Crypto.com cards we have recorded the APR but have not yet read the per-transaction fee schedule, so those two rows assume no per-purchase fee. If one exists, both numbers move against the card. Every other cell in the six rows was read from the document linked beside it.

How the number is built

Issuance + subscription + top-up + crypto conversion + FX on the foreign share + per-purchase fees, minus base-tier cashback with its published cap, over 12 months. It is arithmetic on published fees, nothing else:

Horizon12 months
Monthly spend$500, $1,000 and $3,000 — the $1,000 figure is the headline, the other two are in each row's note
Foreign share20% of spend in a currency other than the card's ownThe FX fee only applies to this part. Spend nothing abroad and the four cards with a non-zero FX fee get cheaper; spend everything abroad and they get worse.
ATM$0 withdrawnATM fees are excluded from this report. Only 42 of the 77 cards have an ATM fee cell at all, so including cash would shrink the comparable set further.
Cashbackbase tier only, cap appliedAn advertised "up to 4%" never enters the arithmetic. Coinbase One publishes 2%–4% and is counted at 2%; Fold publishes 1.5%–4% and is counted at 1.5%; Crypto.com publishes 1.5%–6% and is counted at 1.5%.
Non-USD feesconverted with an approximate rate table, dated in the engine output

Why 51 cards have no number

A yearly cost is only published when every fee that enters it is published. One missing fee and the card drops out of the table — it does not get a zero. Across the 51 live cards we could not price, these are the cells the issuer leaves open:

Crypto conversionmissing on 48 cardsThe spread taken when your crypto becomes spendable money. It is the single largest gap in the base and, on a card used daily, usually the largest fee.
Top-upmissing on 26 cards
Subscriptionmissing on 14 cards
Foreign currencymissing on 13 cards
Per-purchasemissing on 7 cards
Issuancemissing on 6 cards
Cashbackpublished but not in a comparable form on 4 cardsA rate that varies in-app, or a rotating reward asset chosen monthly, cannot be turned into a yearly figure.

Sixteen cards are one fee short, fourteen are two short, fifteen are three short, and six are missing four or more. "Missing" here means the issuer does not publish the value, or publishes it in a form we have not yet been able to normalise — in both cases the honest output is a gap, not an estimate.

The pattern is not random. Cards issued against a US credit line publish a full schedule because they must; exchange and wallet cards publish the fee they advertise and leave the conversion spread out of the documents entirely.

The data

Every live card in this release, priced or not, with each fee cell, its issuer source URL and the date it was observed: card-costs-2026-09-02.csv (57 rows). The computable column says whether a yearly number exists, and missing_fees names the cells that stop it.

The three cost columns are empty for the 51 cards without a complete schedule. That is deliberate: an empty cell is a fact, a zero would be a guess.

What this report does not claim

It is not a ranking. Six cards is too few to rank a market of 77, and the six are not comparable in the way a ranking implies — three are US-only credit cards, one is a Brazilian programme.

A missing fee is not a free fee. A card absent from the table may well be cheaper than everything in it. We do not know, and neither does anyone reading the issuer's page.

The numbers are dated. They describe release 2026-09-02.1. Fees move; each row carries the day we read the document so you can check whether it has since changed. The method is in one page, here.