Shopify's Stablecoin Protocol Supports Delayed Capture. Crypto Just Rebuilt the Authorization.

The Coinbase-Shopify commerce protocol supports delayed capture, refunds and tax. That is the card message set, rebuilt on-chain — which tells you where the savings actually come from.

Share
Shopify's Stablecoin Protocol Supports Delayed Capture. Crypto Just Rebuilt the Authorization.

For a decade the pitch was that blockchains make payment instant and final. The first stablecoin rail built for real commerce had to undo both.

On 12 June 2025, Shopify announced that merchants would be able to accept USDC on Base, Coinbase's Ethereum layer-2 network. Early access began the same day. Broader availability to merchants on Shopify Payments was promised for later in the year. Shopify said it would return 1% to customers who paid with USDC, also later in the year.

The coverage was about stablecoins arriving in mainstream commerce. That is true and it is the least interesting sentence available. The interesting sentence is in the technical description, and almost nobody picked it up.

The payments run on an open-source smart contract and payments protocol co-developed by Coinbase and Shopify. According to the announcement, it supports delayed capture, tax calculation and refund processing, and it integrates with merchants' existing order fulfillment systems.

Delayed capture.

What that word is doing there

For anyone who has not spent their career inside card processing: a card payment is not one event. It is two.

First comes the authorization. The merchant asks the issuer to confirm that funds exist and to reserve them. Nothing moves. A hold appears on the cardholder's account. The merchant now has a promise with an expiry date.

Then comes the capture. When the merchant ships the goods — which may be hours or days later — it tells the acquirer to convert the reservation into an actual charge. Money moves after that, in a batch, on a schedule.

This two-phase design is not an accident of legacy technology. It exists because commerce is not simultaneous. The customer clicks before the warehouse picks. The order may be cancelled, partially fulfilled, back-ordered, split across shipments, or short-shipped by one unit. And in most consumer-protection regimes, charging for goods you have not yet shipped is somewhere between discouraged and illegal.

So the merchant needs a thing that is not a payment but is better than a promise. That thing is an authorization, and the entire card message set is built around it.

Now consider what blockchains were sold as for ten years. A transfer that is instant. A transfer that is final. No reversal, no intermediary, no waiting. Those three properties were the pitch.

And they are precisely the three properties a merchant cannot use.

Rebuilding what was eliminated

An on-chain commerce protocol that supports delayed capture is, functionally, an authorization layer. It has to be. Something must hold value without transferring it, release it on a merchant instruction that comes later, expire if that instruction never comes, and handle the case where the captured amount differs from the authorized amount.

That is the card model. It has been reimplemented in a smart contract.

The refund support tells the same story. A refund is a merchant-initiated reversal of a completed capture, which is to say a second transaction that undoes a first. On a chain where transactions are final by design, the only way to have a refund is to send money back. That works, and it is exactly what a merchant does today through an acquirer, and it is not what "final settlement" was supposed to mean.

And the tax calculation is almost funny in its ordinariness. An open-source crypto-native commerce protocol has a tax module, because commerce has sales tax and VAT and IVA and the chain does not care but the tax authority does.

I want to be clear that I am not mocking this. I think it is the most encouraging thing that has happened in crypto payments. For a decade, the industry's answer to "commerce needs authorization, dispute, refund and tax" was that commerce was wrong. This is the first serious protocol I have seen that answers the question by building the things instead of arguing with the requirement. Coinbase and Shopify went and read the problem.

But the implication cuts both ways. If a stablecoin rail that works for commerce needs authorization, capture, refund and tax, then the savings cannot come from eliminating those functions. They were never the expensive part.

The merchant does not get a stablecoin

Here is the second thing the coverage skipped. Merchants receive local currency by default, with no foreign transaction or exchange fees. A merchant who wants USDC in its own wallet can elect to take it, but that is the option, not the default.

So the standard experience is this: a customer pays in USDC on Base; somewhere in the middle the USDC becomes local currency; the merchant sees a familiar settlement in a familiar currency in a familiar report. The stablecoin exists, for most of these transactions, for about as long as it takes to convert it.

That is a perfectly good product. It is also the opposite of disintermediation. Count the parties: the buyer's wallet, the Base network, the smart contract, Coinbase, Stripe doing the fiat leg, Shopify Payments, the merchant's bank. That is not a shorter chain than a card transaction. It is a different chain, with one leg — the cross-border leg — that is genuinely cheaper, bolted onto a set of intermediaries that look a lot like the ones we have.

And read "local currency" carefully if your local currency is not the dollar. For a Mexican merchant, local currency means pesos, which means somebody converted USDC into pesos at a rate the merchant did not negotiate and will not see quoted. "No foreign transaction fees" is a precise claim about fees. It is not a claim about the rate. Anyone who has spent a career watching dynamic currency conversion knows the difference between those two sentences, and knows which one the margin lives in.

That said, let me not undersell the real gain. Cross-border card acceptance carries cross-border assessments, a currency conversion spread, and a higher interchange tier, and those costs fall on exactly the merchants least able to negotiate them: the small ones selling into markets where they have no entity, no acquirer and no leverage. A rail that compresses the cost of a cross-border consumer purchase is a real improvement for a real population.

It is just not the improvement being advertised. Nobody got removed. Some prices got cut.

What the 1% is actually buying

Shopify said it would return 1% to customers who pay with USDC.

Any operator reading that should ask the obvious question: who funds it?

In cards, the equivalent exists and the funding is explicit. The issuer gives the cardholder points or cashback, and the issuer pays for it out of interchange, which the merchant pays. The rewards economy is merchant-funded consumer acquisition, and that is why interchange is what it is. You can dislike the arrangement but you cannot call it hidden; it is the central fact of the card business.

On a USDC transfer over Base there is no interchange. There is no issuer earning a margin out of which to fund a reward. So the 1% is coming from somewhere else — from the platform, as customer acquisition spend, to solve the chicken-and-egg problem that nobody holds USDC for shopping because nobody accepts USDC for shopping.

Customer acquisition spend has a characteristic that interchange does not. It ends.

Any merchant building a business case on a world where USDC checkout converts better because buyers get 1% should model what the volume looks like when the 1% stops. That is not cynicism about Shopify; it is how every launch incentive in the history of this industry has worked. The question is not whether the subsidy ends. It is whether, by the time it ends, the rail is cheap enough that the merchant can fund a smaller version of it themselves and still be ahead.

The thing no one mentions

The announcement describes refund processing. It does not describe chargebacks, and the trade press coverage I read does not discuss them either.

That absence is not an oversight in the reporting. It is the architecture.

A refund is something a merchant does. A chargeback is something a cardholder does to a merchant, through an issuer, under a scheme rule, with the merchant bearing the burden of proof and, if it loses, the loss. The entire apparatus of consumer protection in card payments lives on that side of the line. It is also most of the cost: not the interchange itself, but the dispute operation, the representment evidence, the fraud liability, the reserves held against it.

On a chain, there is no issuer. There is nobody to ask for the money back. The buyer who pays USDC to a merchant that never ships has the same recourse as someone who wired cash: a complaint, a platform policy, and hope.

Shopify is a large and well-run marketplace and will have platform-level buyer protections, as marketplaces do. But a platform policy is not a scheme rule and a platform's goodwill is not a regulated dispute right. They are not the same instrument, and merchants should be honest with themselves about which one just got cheaper.

Because this is where the economics actually are. If USDC acceptance is cheaper than card acceptance, a meaningful part of that difference is not technology. It is the consumer's dispute right, which has been removed from the product. The merchant keeps the savings. The buyer carries the risk. And the buyer is offered 1% to accept the trade without anyone describing it as a trade.

I am not saying that is wrong. For plenty of transactions — a trusted merchant, a small basket, a digital good delivered instantly — the dispute right was never worth what it cost. I am saying it is a risk transfer priced at one percent, and that someone should say so out loud.

From a Mexico City desk

Three observations from the market where this proposition should land hardest.

First, the cross-border saving is not marginal here. It is the entire product. A Mexican buying from a US Shopify merchant with a Mexican card pays a cross-border assessment and a currency conversion spread, and the spread is set by someone else and disclosed to nobody in a form a shopper would recognise. Compressing that is not a rounding error on a forty-dollar basket. It is one of the few genuinely consumer-facing improvements anyone has proposed in cross-border e-commerce in a decade. The pitch that sounds abstract in a country that already uses dollars is concrete here.

Second, the constraint is not the rail. It is whether a Mexican merchant can get onto Shopify Payments at all, and whether a Mexican buyer holds any USDC. The second is the real wall. A payment method nobody holds is not a payment method, and "worldwide" in an announcement and "available in your country" on a signup form are different claims. The 1% is aimed squarely at that wall, which is exactly why it will work while it is funded and stop working when it is not.

Third — and this is what I would raise in any Mexican risk committee — the chargeback is not a nice-to-have here. In a market where consumer protection enforcement is slow and expensive to reach, the card dispute right is frequently the only remedy that actually functions, precisely because it is adjudicated by a scheme abroad rather than by a local process that takes years. Trading it away for one percent is a materially worse deal in México than in Toronto, and nothing in the product knows the difference. If this rail grows here, the complaints will arrive at the banks and at CONDUSEF, and neither of them was party to the design.

My judgment

This is the most serious piece of crypto payments infrastructure built so far, and the reason it is serious is that it stopped arguing.

Every previous attempt I have evaluated — and I have evaluated several — began from the premise that cards are a legacy system with unnecessary complexity, and that removing the complexity was the product. Authorization, capture, refund, dispute, tax: friction to be engineered away. The result was always a rail that worked beautifully for the one transaction type that does not need any of it, and nowhere else.

Coinbase and Shopify built delayed capture. That is an admission with real content: that the card message set encodes thirty years of learning about how commerce actually behaves, and that if you want to serve commerce you will end up building most of it again. They did the honest thing and built it.

Which means the competitive question is now clear, and it is not a technology question. The card networks' cost is not the authorization — that is nearly free. Their cost is the guarantee: the dispute system, the fraud liability framework, the rules, the arbitration, the enforcement against bad merchants, the global acceptance agreement. Those are expensive because they are a legal and operational apparatus, not because they are slow software.

So a stablecoin rail competes by being cheaper, and it is cheaper principally because it does not have that apparatus. That advantage is real today and it is temporary, for the same reason every disintermediation story in payments has been temporary: the moment the volume gets large enough to attract fraud, somebody will have to build the apparatus, and it will cost what it costs.

My prediction, held loosely: within a few years, the on-chain commerce protocols that survive will have dispute resolution, merchant underwriting, and liability allocation. They will be cheaper than cards, because starting from scratch with better technology usually is. They will not be an order of magnitude cheaper, and they will look much more like card schemes than today's announcements suggest. The delayed capture was the first brick. It will not be the last one they lay.

Three questions for the people who own this

1. If you are evaluating a stablecoin acceptance rail, has anyone written down, on one page, which card functions it provides and which it does not — authorization, partial capture, incremental authorization, refund, dispute, representment, fraud liability shift? Who signed that page?

2. Your business case for lower acceptance cost: what share of the saving comes from technology, what share from a better currency conversion rate, and what share from the absence of a dispute mechanism? If you cannot split the three, you have not finished the case.

3. When a launch incentive funds a payment method's adoption, your conversion data is measuring the incentive, not the method. What is your plan to find out which one your customers actually prefer before the subsidy ends?

Reply to this email with a line or two. The sharpest answers will shape the next issue.

Sources

CoinDesk, 12 June 2025 — "Shopify to Enable USDC Payments on Coinbase's Base for Merchants Worldwide." Source for the announcement date; that early access began the same day with broader availability to Shopify Payments merchants expected later in 2025; that merchants can accept USDC on-chain and receive settlement in local currency, avoiding foreign transaction fees; that Shopify plans to return 1% to customers paying with USDC, launching later in the year; that Base was chosen for cost, speed and security; and that the system runs on an open-source smart contract and payments protocol co-developed by Coinbase and Shopify supporting delayed capture, tax calculation and refund processing, and integrating with merchants' existing order fulfillment systems.

The Paypers, 13 June 2025 — "Shopify integrates USDC with Coinbase and Stripe." Source for the settlement detail: that merchants receive local currency by default with no foreign transaction or exchange fees, that they may instead elect to receive USDC directly in their own wallet, that the fiat leg runs through the Stripe partnership, that merchants use their existing payment and order fulfillment flows with no new integrations or gateways, and that USDC is described as backed 1:1 by US dollar reserves.

On what the sources do not say. Neither article discusses chargebacks or consumer dispute rights, lists the countries where the capability is available, states the conversion rate or spread applied when USDC is settled into a local currency, or gives acceptance pricing beyond the absence of foreign transaction fees. CoinDesk does not describe Stripe's role; The Paypers does not describe the protocol. I have not read the protocol's source code or its documentation. Everything I say about the absence of a dispute mechanism is an inference from the architecture of on-chain settlement and from what the announcements chose to describe, not a statement about what Shopify's own buyer protection policies do or do not cover.

A note on what is mine. The reading that delayed capture amounts to rebuilding the card authorization model in a smart contract; that the merchant's default experience is a card-like settlement rather than stablecoin acceptance; that "no foreign transaction fees" is a claim about fees rather than about the conversion rate; that the 1% is customer acquisition spend rather than a structural reward and will therefore end; that a material share of any acceptance-cost saving is the consumer's dispute right rather than technology; and that surviving on-chain commerce protocols will converge on scheme-like rules — those are arguments, not findings. The description of how card authorization and capture work is standard industry practice and comes from my own work, not from a document about this announcement. Nothing here reflects the position, portfolio, clients or vendors of my employer.