Self-custodial stablecoin wallets reached the payer — and stopped at your counter

The news: a self-custodial stablecoin wallet shipped, and it spends through a card

On September 1, 2026, Ethena launched the iOS beta of Ethena Pay, described by the company as an "internet money neobank" and built exclusively on Avalanche. Avalanche's own announcement — "Ethena Pay Shows How Avalanche Is Powering the Next Generation of Neobanks" — frames the chain as the settlement layer sitting invisibly beneath a consumer product, and notes that users "don't select a network."

The mechanics, as reported and as described by the two companies:

• Balances are held in USDe, Ethena's synthetic dollar. Deposits by bank transfer or crypto convert into USDe; withdrawals settle in the recipient's local currency.
• Custody stays with the user. This is the product's headline claim and the reason it is interesting.
• Yield on the balance: a free Standard tier at 5% on balances up to $5,000; a Pro tier (gated on locking $2,000 of ENA or 10 referrals) at 6% up to $15,000; a VIP tier (locking $10,000 of ENA or 50 referrals) at 6% up to $50,000.
• Spending is via a virtual Visa card across "more than 130 million merchants," with cashback of 4% / 4.5% / 5% by tier, paid in AVAX. The card is not available to US persons.
• Rollout: roughly 400 early-access users at launch, expanding weekly through September, across a beta country list that excludes the US, the EU and South Korea.
• Scale context from Avalanche: Ethena has processed more than $30 billion through its mint and redeem systems, and USDe is integrated across more than 100 platforms and protocols.

A note on the numbers, because the reporting genuinely conflicts. Outlets published 48, 49 and "50+" for the country count on the same weekend. On cashback, The Block reported up to 10% at selected brands for higher tiers, while The Defiant pointed out that Ethena's own launch post and pricing page advertise up to 5%. Founder Guy Young said USDe's rate funds the savings yield and declined to disclose the funding source for the rest of the rewards. We are reporting the conservative, issuer-published figures and flagging the spread rather than picking the flattering number — and you should apply the same discipline to any acceptance economics quoted at you, including ours.

None of that is the interesting part for a business. The interesting part is one sentence buried in every version of the story: the customer holds their own keys, and then pays you with a Visa card.

What the card wrapper costs the business on the other side of the counter

Run the transaction from your side of the till.

A customer walks in holding self-custodial digital dollars on Avalanche. They tap. What arrives at your business is a card authorisation. Which means, unchanged from any other card sale:

• You pay a merchant discount rate. Roughly 1.5–2.5% in developed markets, and commonly 3–3.5% plus a cross-border surcharge in Latin America, Africa and Southeast Asia — which is to say, most of the countries in this beta.
• An acquirer holds your money. Not the customer, not you. Settlement lands T+1 to T+3, sometimes T+30 in higher-risk categories, and a rolling reserve may sit on top of it.
• The payment stays reversible. Card chargeback windows run to roughly 120 days. The fact that the customer funded the card from a self-custodial wallet changes nothing about your dispute exposure — the reversal is a network rule, not a funding question.
• Your acceptance can still be withdrawn. The underwriting decision that governs whether you may accept cards at all is entirely untouched by how your customer stores value.

So the payer's balance sheet changed and yours did not. The customer graduated from "a bank holds my dollars" to "I hold my dollars." You are still at "an acquirer holds my dollars, and may give them back on a schedule, minus a percentage, unless someone disputes."

That asymmetry is the whole story, and it is not an accident.

Why every consumer stablecoin wallet ships a card: acceptance is the hard half

There is a structural reason this keeps happening, and it is worth stating plainly because it explains the last two years of product launches.

Issuing is a deal. Acceptance is a build. Putting a card into a consumer app is a partnership with a BIN sponsor and a processor — weeks of work, and on completion you can claim 130 million merchants. Getting 130 million merchants to display a payment address is not a deal you can sign with anyone. It is a per-merchant integration, one business at a time.

So every consumer wallet takes the shortcut, and the shortcut is always the same: wrap the on-chain balance in a card so it can be spent at businesses that have no idea any of this happened. We wrote about the scale of this in crypto payment cards crossing $759M/month in stablecoin spend, and about the consumer-side build in Chime's stablecoin wallet for 10 million ordinary Americans. In both cases the observation was the same one we are making here, and Ethena Pay is now the cleanest example of it: the merchant side is the half nobody is building.

It is worth being precise about what the card wrapper actually is. It is a translation layer. It takes a settled, final, self-custodied on-chain payment and converts it, at the point of sale, into a promise presented now and funded later — the exact instrument whose properties created interchange, acquirers, rolling reserves and chargebacks in the first place. Every cost on the list above is a cost of the translation, not a cost of the dollars.

And here is the part that makes this launch different from a card programme: Ethena Pay already does the untranslated version. The app's own feature list includes sending money domestically and internationally, settled on Avalanche. The rail that would pay a merchant directly is shipped, working, and in the customer's hand. The only missing component is on your side of the counter: an address to pay.

The country list is the tell

Look at where this beta went live: Brazil, Mexico, South Africa, Kenya, the Philippines, Singapore, Japan, the UAE, Australia — and explicitly not the United States, the European Union or South Korea, which are pending regulatory requirements.

The exclusions are regulatory. But the inclusions are commercial, and they describe a very specific customer: someone for whom holding dollars is itself the feature. Someone whose local currency loses value, or whose bank makes holding USD difficult, or who gets paid across a border and would rather not convert twice.

Now overlay a merchant map on that same list. Those are, with remarkable consistency, the markets where:

• Card MDR runs 3–3.5% rather than 1.5%, and cross-border cards carry a surcharge on top.
• Settlement is slowest and rolling reserves are most common.
• A meaningful share of paying customers are foreigners, remote clients or diaspora — precisely the cohort whose card is issued somewhere your acquirer treats as risk.
• The "we'll convert it to local currency for you" product — the model Japan priced at 0.98% — either doesn't exist or isn't licensed.

In other words: the consumer wallet went live in exactly the geography where direct acceptance is worth the most to the business, and it arrived there wearing a card. If you operate in one of those markets, a growing number of your customers now hold spendable on-chain dollars, and you are meeting all of them through the most expensive possible interface.

Yield is the first honest reason a normal person holds stablecoins

There is a second thing in this launch that merchants should register, and it isn't the custody model. It's the 5–6%.

Until now, the case for a non-technical consumer holding a stablecoin balance was defensive — inflation, capital controls, a bank that won't hold dollars. Real reasons, but narrow ones. A yield-bearing self-custodial balance is a different proposition: it gives an ordinary person a positive, arithmetic reason to keep spending money in a stablecoin rather than a checking account, and to top it up rather than draw it down.

For a business, the consequence is simple and it is the only forecast in this article: the pool of customers who can pay you on-chain grows. Not because they became crypto enthusiasts, but because a rate box in an app made it the default place their money sits.

Two honest qualifications, because a rate quoted at launch is a marketing instrument:

• The rate is tiered, capped and partly gated on locking a volatile token. 6% applies up to $15,000 or $50,000 depending on tier, and reaching those tiers requires locking $2,000 or $10,000 of ENA — or recruiting 10 to 50 referrals. That is a growth mechanic as much as a savings product.
• Yield sources change. Ethena states USDe's rate funds the savings yield; the funding source for the remaining rewards was not disclosed. Rates funded by market conditions move with market conditions.

You do not need the rate to persist for the merchant conclusion to hold. You need it to have worked once — long enough to move a cohort of customers into on-chain dollars. Distribution changes are stickier than the promotions that cause them.

The part the launch coverage under-weighted: USDe is not USDC or USDT

This matters more to a business than to a consumer, so it deserves its own section.

USDe is a synthetic dollar. Per Ethena's own technical documentation, it is "a synthetic dollar, backed with crypto assets and corresponding short futures positions" — spot crypto held against offsetting short derivatives to hold the portfolio delta-neutral, alongside allocations to liquid stablecoins, staked ETH, basis trades, lending positions and tokenised real-world assets. It is not a T-bill-and-cash reserve token in the mould of USDC or the model the GENIUS Act frames.

Ethena is admirably explicit about the risk categories that follow, and we will quote the categories rather than characterise them: funding-rate risk (persistently negative funding erodes the revenue that supports the peg mechanics), liquidation risk, custodial risk via off-exchange settlement providers, exchange-failure risk from holding derivatives positions at centralised venues, and margin-collateral risk. For context on scale, USDe's circulating supply sits around $4 billion, down from roughly $15 billion at its September 2025 peak.

None of that is an accusation. It is a description, published by the issuer, of a genuinely different instrument. The merchant lesson is the one we made in stablecoin payments and counterparty risk and it generalises well beyond this launch:

This is exactly why Payzum's optional auto-conversion to USDC or USDT is an issuer-selection control and not a volatility feature. Your customer pays with what they have. You hold what you chose to hold. Those are two separate questions, and a payment system that conflates them is handing your treasury policy to an app store.

How Payzum closes the half that's missing

Payzum is a non-custodial, crypto-only payment processor. The design principle is a single sentence: the settlement is the payment. There is no Payzum balance, because funds move directly to wallets the merchant controls. Nothing is pooled, held, scheduled or released.

Set against the four costs of the card wrapper listed earlier:

• No acquirer holding the money. The payment lands in your wallet. There is no processor float, no payout schedule, no rolling reserve and no balance for anyone — including us — to freeze.
• No chargebacks. An on-chain payment is final on confirmation. There is no 120-day window and no card-network dispute process, because there is no network rule granting one.
• Seconds, not days. Typical confirmations: ~0.4s on Solana, ~2s on Base and Polygon. Payzum settles across Bitcoin, Ethereum, Solana, Polygon, Base, Arbitrum, Optimism, BNB Chain and Avalanche — the same network Ethena Pay settles on.
• Cents, not percentages. No acquirer and no card-scheme fees; on the low-cost chains the network cost of moving a stablecoin is measured in cents. // confirmar pricing actual
• Your treasury stays yours. Optional auto-conversion to USDC or USDT means you decide which dollar you hold, regardless of which one walked in.

And critically for this story, none of it requires the customer to use any particular app. It requires them to be able to send an on-chain payment — which is the capability every one of these wallets ships, card or no card.

How it works, step by step

1. Create your account and connect your own wallets. You supply the destination addresses per chain and token. Payzum never holds keys and never takes custody — the addresses you enter are where the money lands. Enable optional auto-conversion to USDC or USDT if you want a fixed settlement token.
2. Pick how you get paid. In person: POS generates a fresh QR per sale for the exact amount, so any phone or tablet becomes a terminal — no register integration, no hardware purchase, no acquirer. Online: hosted checkout (redirect, modal or inline), no-code payment links and buttons, invoices with expiry and overpayment detection, tip jars and donations, or recurring subscriptions.
3. Set up your team and your books. Cashiers get PIN-level access with per-cashier and per-terminal analytics, so a busy counter reconciles by shift rather than by guesswork. For a developer-led setup, use the REST API with API keys, signed webhooks and the integration playground.
4. Take the payment — and watch it settle. The customer scans or clicks and sends from whatever wallet they use. Confirmation lands in seconds, the funds arrive in your wallet, and the transaction is final. No authorisation hold, no batch, no deposit two days later, no dispute window.

Where this actually lands: concrete situations

Abstractions don't get adopted; specific Tuesdays do. Four situations where the payer-side shift in this launch changes something measurable for a business:

• A café or restaurant in a tourist district — Cartagena, Cancún, Punta del Este, Cape Town, Boracay. A large share of tickets are foreign cards, which is the most expensive category the acquirer sells, and settlement is slowest exactly when cash flow matters most. A QR taped beside the register takes the same customer's on-chain dollars in about two seconds, at cents of network cost, with the money in the owner's wallet before the customer sits down. See accepting crypto payments in person.
• A freelancer or studio invoicing abroad from São Paulo, Bogotá, Lagos or Manila. The client now holds spendable dollars in an app; the current alternative is a wire that arrives late, short, and after two conversions. A Payzum invoice with an expiry and overpayment detection is one link, and the dollars land in the studio's own wallet the same afternoon.
• An online store with a high-value, high-dispute basket — electronics, travel, ticketing, anything shipped internationally. Chargeback exposure is the operating cost that never appears on the pricing page. Finality removes the category. See reducing chargebacks on an online store.
• A business that pays out as well as collects — affiliates, contractors, tournament prizes, creator revenue shares. The same customers holding self-custodial dollars are also the people you owe. Payzum does mass payouts by CSV (BTC/LTC/DOGE) and EVM stablecoin payouts on Polygon, Arbitrum, Optimism, Base, BNB Chain and Avalanche. See crypto mass payouts.

Same customer, two routes: card wrapper vs direct acceptance

This table holds the payer constant. In both columns, the customer is the same person, holding the same self-custodial dollars in the same app. The only variable is how that value reaches the business.

Objections worth taking seriously

"My customers don't hold stablecoins, so none of this matters yet."

For most businesses on most days, correct — and we'd rather say so than pretend otherwise. The reason to read a launch like this is that it is a leading indicator, not a current one: 400 users is not a market. But the direction is unambiguous, the incentive attached to it is now financial rather than ideological, and it is landing first in the exact markets where card acceptance hurts most. The right response is not to migrate; it is to add a second way to be paid, at close to zero fixed cost, and let volume decide. A QR beside the register costs nothing when it goes unused.

"If the card gives them 130 million merchants, why would anyone pay me directly?"

Because the incentives point that way for both sides once the option exists. The wrapper reintroduces a fee the payer's rail didn't have, and cashback is funded out of it. On your side, a direct payment removes 1.5–3.5%, removes the settlement delay and removes the dispute exposure. That gap is real money, and it's the room from which discounts, faster service or simply better margin come. The card wins where you have no address to pay. That's a solvable condition.

"Isn't taking a synthetic dollar like USDe risky for my treasury?"

Holding any single token as a treasury position carries issuer risk, which is why the honest answer is a mechanism rather than a reassurance: separate what you accept from what you hold. Optional auto-conversion to USDC or USDT means your settlement token is your decision. Beyond that, the exposure window on an acceptance payment is minutes, not quarters — you are not running a carry trade, you are taking a payment. Which tokens and chains are enabled on your specific account is a configuration question, and the right place to answer it is a call, not a blog post.

"We already take cards. Why add anything?"

Keep them. We say this in every one of these posts and we mean it here: a card is a transfer instrument fused with a consumer lending product, and stablecoins replace the first, not the second. If your average ticket depends on instalments or revolving credit, cards stay. Add direct acceptance for the segment cards serve worst — cross-border customers, high-dispute baskets, foreign-issued cards, anything where the surcharge and the reserve are eating the margin — and run both.