WEEKLY WRAP · 2026-09-15
6 quick hits · market context · 53 cards updated
LAUNCH
Nu, Nubank's US-chartered entity, opened American operations between September 10 and 13, and the piece worth your attention is the Nu Global account. Deposits convert into Circle-issued USDC at 3.50% APY or EURC at 2.20% APY, and a virtual Mastercard spends that balance at what Nu calls competitive exchange rates with no markup of its own on top. Then read the small print, because two sentences do a lot of work. Nu has not disclosed where the yield comes from. And the protection behind your balance is a Swiss bank default guarantee, not FDIC insurance, which are different products with different failure modes and only one of them is the thing Americans think they are getting from a bank. The card's launch focus is Europe to LatAm corridors, and US card availability has no firm date. We have not opened an entry for it, and the reason is specific rather than cautious: there is no published fee schedule, no stated KYC terms and no confirmed country list, so three of the four fields we would compare it on do not exist yet. It gets an entry the week those pages appear. Our read: 3.50% on a dollar balance you can spend directly beats most of the cashback ladders we track, once you account for the fact that nobody caps your yield at $25 a month. The undisclosed yield source is the sentence we want filled in before anyone treats it as a savings account.
LAUNCH
This one went into the database on September 13, three days after launch. A USDC balance lives in the MoneyGram app and spends at Visa merchants, or you walk into an agent location and take it out as cash, which is a thing no crypto-native card on our list can do. It is built with Rain on Crossmint wallets and settles over Stellar. No monthly fee, no annual fee, no purchase fee at launch; there is an inactivity fee after three or more months and MoneyGram does not publish the amount. Cashback programme: none, not a token 0.5%, nothing. Digital only for now, with a physical card and a wider LATAM rollout stated for late 2026 and no country list attached. Two things we will say plainly about our own entry. First, moneygram.com/mgo/co/en/card serves the generic US money-transfer homepage when you read it from outside Colombia, and moneygram.com/card is a 404, so the fee wording rests on launch coverage quoting MoneyGram's product page rather than a first-party read, and the entry says so in the note instead of pretending otherwise. Second, Rain. On August 28 an outdated version of a Rain collateral contract was drained of roughly $1.1 million across Avici, Tria and at least one programme nobody has named. Rain said at the time that every programme on that contract version had been upgraded and that it had seen no further unauthorised activity since, which is what we reported on September 1 and is the most recent thing Rain has said about it. That is probably fine and it is also now the plumbing under a MoneyGram card, which is the sort of thing worth knowing before you park a balance on one.
Source: The Block MoneyGram Card → Rain Card → Avici Card →
RATES
On August 11 we made Bitunix's card a hero item off the launch announcement, wrote that its 1% to 8% VIP ladder capped at 1,000 USDT a month was a real number by comparison with Bybit, and held the entry back until we could confirm regional availability and the network from the official help centre. We were right to hold it and wrong about which field would break. The entry went in on September 8, and the official rate table is not the circulated one. Every secondary source republishes 1/2/2/4/4/6/6/8% across VIP 1 to VIP 8. Bitunix's own VIP page pays nothing at all through VIP 2, then 1% at VIP 3, 2% at VIP 4 and 5, 4% at VIP 6 and 7, and 8% only at VIP 8. There is no 6% band anywhere. VIP 3, the first level that pays a cent, wants any one of 25,000 USDT in assets, 800,000 USDT of 30-day spot volume, or 8,000,000 USDT of 30-day futures volume. VIP 8 has no asset-balance route at all: it is 100,000,000 USDT of 30-day spot volume or 500,000,000 USDT of futures, which is not a tier a card user reaches, it is a tier a trading desk reaches. Two more things we did not know in August. Your VIP level is recomputed daily at 00:00 UTC from trailing 30-day volumes and your current balance, so the rate is not stable from one month to the next. And the card can only be funded by internal transfer from a Bitunix Spot, Futures or Earn account, so no bank transfer and no on-chain deposit. Our read: a published floor of 1% against a real floor of 0% is the most misleading shape a cashback table can take, because a skimming reader converts it into "worst case, 1%" and the worst case is actually nothing. The August 11 wrap still carries the old figure and we are not quietly editing a dated page; this is the correction, in the same place we make every other one.
Source: Card Pilled Bitunix Card →
SHUTDOWN
Two of these came out of the September 11 rotating check, which went through 50 cards, and the third out of the regional sweep two days later. None of the three findings is a rate. WhiteBIT's Nova EUR Card has been suspended since June 30, pending the launch of WhiteBIT Europe. The cards sit in accounts and no transaction goes through, with WhiteBIT pointing users at the Nova USD Card in the meantime. Pera Card is not obtainable either: perawallet.app/pera-card now serves a waitlist page with no rate, no fee schedule and no country list, while our entry still carries 2% in USDC. And in Brazil, onilx.com.br's card page returns a 404 while the site's own homepage keeps linking to it, with Livecoins reporting on September 9 that OnilX has ceased operating, citing more than 470 Reclame Aqui complaints, missed withdrawal deadlines, staff departures and closed offices. All three are still marked active. Here is why, including the part that does not flatter us. Our rules require multiple independent sources before we mark a card discontinued, and OnilX has one columnist and no statement from the company, so it stays put and the note now tells you to treat the card as unobtainable in practice. WhiteBIT and Pera are not discontinued at all, they are suspended and withdrawn, which is genuinely a different state. The honest label for all three is unclear rather than active, and switching them touches the section headers our validator counts, the pre-render and the redirect logic for dead cards, so an unattended routine flagged it for a person instead of reorganising the file at 2am. Our read: "active" in any comparison database, ours very much included, does not mean somebody checked that you can get one. It means nobody has proved it is dead. Read the note, not the badge.
Source: Livecoins WhiteBIT Nova Card → Pera Card → OnilX Crypto Card →
LAUNCH
Kite went in on September 10: a virtual USD Visa funded from a non-custodial USDC or USDT balance, live in nine APAC countries (Australia, Japan, Malaysia, the Philippines, Singapore, South Korea, Taiwan, Thailand, Vietnam), with the US a prohibited jurisdiction and the wallet reaching about 40 more markets that get no card. No issuance, maintenance or replacement fee, 0.53% to move stablecoins onto the card, 0% on USD spend and 1.8% on everything else. Now the rate, which is the unusual part. Kite has no general cashback whatsoever. Every published rate is one of five 5% category offers, all dated to 30 November 2026: AI subscriptions (ChatGPT Plus, Claude Pro or Max, Perplexity Pro, Midjourney, Cursor Pro, Gemini Advanced), streaming, creator tools, Starlink bills, and travel booked through Travala. Once per platform per person per month. If you pay for three AI subscriptions on a card that costs nothing to hold, that is a decent eleven weeks and a 0% card in December. Kite also forced a fix we should have made months ago. Promotional rates were previously written as prose in three inconsistent shapes across seven cards, which is exactly how Zoomex's promo lapsed with nobody noticing and had to be hand-patched after the fact. Any cashback tier can now carry a dated promotional flag, and the badge relabels itself from "Promo, ends 30 Nov 2026" to "Promo ended" on the day, in the React page, the pre-rendered HTML and the markdown version, all from one function so the three cannot drift apart. Satoshi Tango and Spritz were retrofitted as promotional. Tria, SafePal, Wallib and BingX were checked and their rates are the standing ones. Separately, four bugs in the ROI calculator's tier parser were mispricing five cards, and one was wrong in the direction that actually costs a reader money: HashKey's card was published at minus $3,120 a year, on a $3,600 subscription fee it does not charge, because the parser read "capped at HK$300/mo" as a monthly cost rather than a rebate ceiling. It now reads plus $456. KAST's effective rate was being halved because a "(+ KAST Points)" clause tripped a non-cash-reward test on a rate that pays in USD stablecoin.
Source: Card Pilled Kite Card → HashKey Credit Card → KAST Card → Spritz Card → Satoshi Tango Card → Zoomex Card →
Market Context
Three stablecoin card stories this week and not one came from a crypto company. Nubank's US entity put a virtual Mastercard on a USDC balance paying 3.50%. MoneyGram put a Visa on a USDC balance in Colombia and let you walk into an agent location to take the cash out. Circle agreed to pay $400 million for a Singapore payments firm so USDC settles natively across APAC. Now notice what none of those three sells: a cashback rate. Nu pays yield on the balance instead. MoneyGram pays nothing and sells the cash-pickup leg. Circle is not selling a card at all, it is selling the floor everyone else's card stands on. Our read: the cashback ladder is a crypto-native product habit, and it is not the part the incumbents are copying. They are copying the spendable dollar balance and the absent FX markup, which are the two features people actually use. If you have been ranking these cards by headline rate, the interesting entrants over the next year may not have a rate to rank. Which makes our own findings this week read differently. The Bitunix ladder we published at 1% to 8% pays 0% below VIP 3, which takes 25,000 USDT held or heavy trading volume to reach, and its top band needs nine-figure volume. Kite's five 5% offers all expire on November 30 with no general rate underneath them. HashKey was showing minus $3,120 a year on our own calculator because of a subscription fee it does not charge. Three cards sit marked active that you cannot obtain or use today. Every one of those is a number that looked authoritative until somebody opened the provider's own page. We are at 201 cards and 168 active after adding twelve: eight in the first discovery run with working page fetches since the network policy opened on September 4, Kite on the 10th, and three from the regional sweep. Five existing entries had a field corrected, two of them the kind that moves a card between guides. Honestly, the twelve new entries are the least interesting part of that. The reason they are worth having is that the parser underneath them now knows a rebate cap from a subscription fee, and a promotional rate from a standing one, which it did not on Monday.
Cards updated this week: Bit2Me Card → Aven Bitcoin Visa Card → Shakepay Card → Kolo Card → DeCard → Young Platform Card → WhiteBIT Nova Card → Pera Card → Bitbank EPOS CRYPTO Card → Slash Card → Littio Card → OnilX Crypto Card → Lemon Card → Binance Japan Card → Binance Card (LATAM/Africa) → Fasset Card → HashKey Credit Card → Bitget Wallet Card → KAST Card → Spritz Card → Satoshi Tango Card → Tria Card → SafePal Card → Wallib Card → BingX Card → Crypto.com US Credit Card → Robinhood Gold Card → Robinhood Platinum Card → Floki Card → Cryptomus Card → Plasma One Card → Tuyo Card → SwissBorg Card → Mercuryo Spend Card → Fold Credit Card → Pengu Card → SBI Visa Crypto Card → Rizon Card → GoMining Card → Bipa Card → SolCard →
New cards added: Bringin Card → Fiat24 Debit Card → Wallbit Card → Based Card → Kem Infinity Card → Kripicard → PEXX Card → Bitunix Card → Kite Card → MoneyGram Card → Lulucard → Zaif Card →
← Previous issue: September 1–7, 2026 | All issues: Weekly Wraps archive →