Everyone who pays attention in the payments industry knows that Ryan McInerney is a smart guy. I didn’t realise just how smart until I found he had already answered a question I’d only just thought of, before I even got round to asking it!
On Visa’s earnings call from the 30th January 2025, when asked about the UK account-to-account (A2A) product, Visa’s chief executive described it as the case “where we’ve unbundled kind of our brand and acceptance, our rules in terms of charge-backs, disputes, and returns and how things work, as well as kind of our risk management capabilities. We’ve unbundled that from the Visa stack...”
That’s a more candid description of Visa A2A than most of the coverage I’ve read. Since February this year, the UK’s Open Banking story has been recounted as a challenge to the card networks. The Independent reported that the country’s biggest banks were convening to create an alternative to Visa and Mastercard. In June, Finextra wrote up the UK Payments Initiative’s launch as a scheme “to challenge Visa and Mastercard stranglehold”.
In fairness, not everyone read it that way. Simon Taylor made the opposite call in Fintech Brainfood back in June 2025, writing that far from competing with the card networks, “pay by bank is becoming just another way for them to do what they do”. Martin Koderisch of Scalepoint Partners had also already spotted the mechanism in April 2025, observing that by adding dispute resolution, unified messaging, fraud controls and participation rules, Visa A2A was essentially trying to emulate the cohesion of Pix and UPI. Both were right, and both of them got there before I did.
Where I want to build on their thinking is by taking it one step further back, to examine what’s been unbundled and what that’s worth. What the public instant rails have done across the UK, Brazil and India is take the rail out of a bundle that used to have three parts. For decades, a card product was an amalgam of a rail, a rulebook and, in the case of credit cards, a revolving line of credit, all sold together. Today, Visa A2A is the rulebook positioned on its own, on top of a rail Visa doesn’t own.
So the question I wanted to work through isn’t whether Visa is being disintermediated. It’s whether anyone will pay for the rulebook once the rail underneath it is public financial infrastructure. The UK is now the market where three different answers to it are being priced side by side.
What Visa A2A provides, and what Visa is careful not to say
Let’s start with what the product is, because using shorthand and calling it the “card layer on top of open banking” might have the right shape to it, but it gets a couple of details wrong.
Visa announced Visa A2A on the 5th of September 2024 for the UK, built on Open Banking and Pay.UK’s Faster Payment System (FPS), with bills and subscriptions as the first use case and commercial variable recurring payments (cVRP) as the mechanism. Visa called the platform “fully operational” in June 2025. The first live transaction, a Utilita energy bill paid from a Kroo Bank account through Tink, was completed on the 19th November 2025. At the time of writing, Visa hasn’t publicly disclosed transaction volumes or its fee schedule, and has named no live banks beyond Kroo. eCommerce remains something Visa intends to expand into in phases. So this is still an early rollout, and I’d read every article about it, including mine, through that lens.
What Visa does say is worth taking at face value. The FY2025 10-K describes Visa A2A as an open system bringing standards, rules and a dispute management service to bank payments. Visa’s release on the first transaction describes what it supplied as the operating model, the liability and dispute framework and the user experience guidelines, under a Secured by Visa trustmark. Its product page puts it more bluntly, promising “One rulebook. Clear UX guidelines.”, together with a clear liability framework and dispute handling. So in essence, the money moves over FPS, initiated through Open Banking APIs by a payment initiator, and Visa sits above the rail, writing the rules for what happens when a payment goes wrong.
Visa, however, has been silent on two important things. It hasn’t clarified, at least in my reading, whether Visa refunds the consumer or guarantees the payment. The language is that Visa A2A will provide a level of protection similar to what shoppers are used to on cards, which reads as a liability framework allocating the loss between participants rather than a Visa-funded guarantee, and Visa hasn’t said which participant funds a refund. It has also been elusive about what the rulebook costs. The product page offers the words “resilient, risk-aligned pricing”, and Visa didn’t respond to Payments Dive’s questions on fees when it announced the product. Of the two silences, the one on price matters more, and I’ll come back to it.
The part of the card bundle Visa has chosen to keep is the part that decides who loses money when a payment goes wrong.
The public rails wrote rules for fraud but not for commerce
The reason a rulebook is worth selling at all is that the public rails have mostly written a narrow one. Most instant rails that now carry real retail volume have built a mechanism to get money back after push-based fraud or a system failure, but almost none have built one for the ordinary commercial dispute. Think about your typical eCommerce order where the parcel didn’t arrive, or the Netflix subscription that somehow wasn’t cancelled, and you were still billed.
Pix provides a good case to illustrate the point. Brazil’s central bank, the BCB, limits the scope of its own implementation guide for the Mecanismo Especial de Devolução (MED), the Pix return mechanism, to fraud, including scams, and operational failure, and the guide explicitly excludes purchases where the goods arrived late, weren’t the model ordered or weren’t what the buyer expected. The participant FAQ rules out commercial disputes in as many words, and the BCB’s own recurring-payment product, Pix Automático, launched in June 2025 under the same fraud-only return rules.
India’s UPI is closer to cards than I’d assumed before checking. The National Payments Corporation of India (NPCI) runs its own dispute cycle. Its Unified Dispute and Issue Resolution (UDIR) process dates from a November 2020 circular and, on the account of the payment providers that plug into it, sits above the remitter-bank chargebacks, pre-arbitration and NPCI arbitration. However, the statutory protections set by the Reserve Bank of India (RBI) cover only unauthorised and failed transactions. The 2017 circular limits what a customer can lose on an unauthorised transaction, and the June 2026 revision, which takes effect on 1 January 2027, widens that to fraudulent transactions more generally. Whether “goods not received” exists as a UPI chargeback reason in the way it exists as Visa’s category 13.1 is something I couldn’t confirm. Put simply, UPI has the interbank dispute plumbing, but its guaranteed protections stop at fraud and failure.
The UK’s Faster Payments, the rail Visa A2A actually sits on, carries a mandatory reimbursement regime for authorised push payment fraud, capped at £85,000 per claim and split between the sending and receiving banks. That regime explicitly excludes civil disputes, such as where a consumer has paid a legitimate supplier for goods or services but has not received them. HM Treasury’s National Payments Vision puts the gap in one sentence, stating plainly that card payments currently offer greater consumer protection than payments by bank transfer.
The public rails that have written rules at all have written them for fraud, some of them very good ones, and left the commercial dispute to whoever wants to write rules for it. Cards, in their wisdom, wrote both. Visa’s public rules cover four dispute families (fraud, authorisation, processing error, and consumer dispute), with goods not received, cancelled recurring, and not-as-described each sitting inside the fourth with a 120-day clock, the issuer raising the dispute, and the network arbitrating at the end. That’s the product Visa is now selling on its own, because it was never on the public rails.
The rulebook costs merchants time before it costs them money
It’s only fair to look at the same situation from the merchant’s side, where the gap between the rulebooks turns into a cost you can put a number against.
A merchant offering cards, a public instant rail and a wallet or two is running several different clocks at once, and none of them agrees on what time it is. With cards, the acquirer pays the merchant in days, but disputes can arrive up to four months later, so the acquirer carries the contingent liability in between (Global Payments’ 10-K puts it as “we incur chargeback losses when our merchants refuse or cannot reimburse us”). On Faster Payments and Pix, there is no chargeback clock at all, because there are no chargebacks. A refund on Faster Payments or Open Banking is a new payment instruction the merchant initiates, and a refund on Pix is a linked return the merchant’s institution sends.
You can see what that difference is worth by looking at what one provider does with it. Stripe’s payout schedule for Brazil holds domestic card payments for 30 calendar days before paying the merchant, against two business days for Pix. Stripe’s own support page for Brazilian accounts sets the same terms for every domestic card payment, with each day’s payout picking up card payments from 30 days earlier and Pix payments from two business days earlier. Stripe doesn’t say why, so I won’t claim the whole gap for the rulebook. What I will say is that for the same merchant, with the same provider, in the same country, the rail with a chargeback right pays out four weeks after the rail without one, and somebody is funding that right in the meantime.
So a rulebook isn’t free to the merchant even when nobody invoices for it. It’s paid for in held funds, reserves, dispute operations with staff to train, deadlines to hit, and the acquirer’s pricing of all of that. That’s the mechanism behind the Wero article I wrote in July. The European Payments Initiative (EPI) built a purchase-protection rulebook for Wero, so its account-to-account wallet could compete with cards at checkout. Then, in July, it deferred full Dutch coverage to January 2028, after consulting merchants and payment service providers (PSPs), and held its pricing to the end of 2028 as well. The deferral moved the “when,” but not the “who,” and the “who” is still the merchant side.
The UK is pricing three answers to the same question side by side
Coming back to the question, then. If the rulebook is now sold separately from the rail, who pays for it? The UK market hasn’t settled on one answer. Instead, it’s in the process of testing three, and the regulators have made it clear that they want it to.
The first test is the industry’s own scheme. UK Payments Initiative Ltd (UKPI), a company formed by 31 firms, formally launched its cVRP scheme in June 2026 for a first wave of lower-risk use cases, including utilities, financial services and government. Neither UKPI nor the regulators have yet published its centralised access fee. Open Banking Tracker’s guide reports it as a fixed sum in pence per transaction, at around two pence plus a scheme fee, but that remains unconfirmed. What is confirmed is what it doesn’t include. UK Finance’s own commercial-model paper states that existing open banking payments offer only limited purchase protection for non-fraudulent transactions compared with card schemes, and this first wave was scoped to what the regulators call lower-risk use cases, which are the ones that can operate without one.
The second test is the industry’s proposal for what comes next. The same UK Finance paper models Wave 2, the eCommerce wave, on an ad valorem fee rather than a pence fee, in two tiers. A “SIP+” tier (a SIP is a single immediate payment, the plain open banking push) with no card-style purchase protection beyond cover for merchant failure is modelled at 21 basis points, and a Debit Card Efficient tier that carries card-like protection is modelled at 11 or 23 basis points depending on the average transaction value the scenario assumes, with the paper careful to say that no fee level has been agreed. The paper also puts a number on what it’s pricing, estimating debit card chargebacks at around 9p per transaction on average. That’s the industry’s own price tag on a dispute rulebook.
The third test is Visa’s. Visa A2A is card-shaped in structure (one rulebook, a liability framework, dispute management run by the network), but its price is the part we don’t know. The regulators have been explicit that they expect schemes to compete rather than be absorbed into one. The December 2025 delivery update from the Payment Systems Regulator (PSR) and the Financial Conduct Authority (FCA) says that new schemes should be able to enter the market if cVRPs are to compete and grow, and the FCA’s launch statement in June says it wants to see competition between commercial open banking schemes. Neither names Visa, and Visa A2A is the only other scheme with a live transaction behind it and a rulebook to sell.
Only one of the three has a number in the market, and even that isn’t published, but in the UK, the question “Will anyone pay for the rulebook on a public rail?” has already shifted to “Which rulebook, and at what price against the bare rail?”
Whoever sets Visa’s A2A price can already see the other two.
Mastercard is running the same strategy from the other end
Mastercard’s UK position looks different at first glance, but I think it’s actually the same move approached from the opposite direction.
Mastercard owns and operates the infrastructure that underpins the rail. Vocalink, which it has owned outright since February 2025, when Santander sold its residual stake (I set out the ownership chain in my article from August), runs the Faster Payments infrastructure under contract to Pay.UK, the scheme’s operator. As far as I can tell, Mastercard doesn’t sell a branded consumer scheme on top of it. The “Pay by Bank” button on Amazon UK is TrueLayer’s, and the older Pay by Bank app that came with Vocalink carried no card-style dispute framework that I could find. So, on the face of it, Mastercard has the plumbing and no rulebook, the mirror image of Visa.
Two moves suggest the mirror is the point. In July 2025 Mastercard announced A2A Protect, a fraud-first overlay on Faster Payments built with Monzo, NatWest and Santander, with disputes over goods and services flagged as a later expansion and no release yet confirming it live. And this July the Financial Times reported early-stage talks over the sale of a majority stake in Vocalink, possibly to the industry-owned DeliveryCo, which Mastercard hasn’t confirmed.
Michael Miebach put the logic plainly on Mastercard’s April call. Asked about the infrastructure business, he said, “the franchise rules are different in that space than they are in a card space”. On the July call he drew the line even more clearly, describing cards as an ecosystem whose common fraud rules come through the network’s own rulebook, and account-to-account as infrastructure, where Mastercard is now applying what it learned running Vocalink to a consortium approach with the UK banks. If you net it out, the two global card networks are converging on a similar approach from opposite sides. Visa is selling a rulebook above plumbing it never owned, and Mastercard is building one above plumbing it’s reportedly weighing whether to keep.
Whichever way you come at it, the asset the networks are choosing to hold is the rules, and the asset Mastercard is reportedly willing to put up for sale is the rail.
Visa’s A2A is a toehold today, but the strategy behind it is a defence of debit
As much as I enjoy drinking my own Kool-Aid, I’ve also got to be pragmatic before I send the hype machine into overdrive. The reality is that Visa A2A is a single product in a single market, focused on bills and subscriptions only. It has just one bank confirmed live two years after it was announced, and Visa’s own July earnings call didn’t mention it, Tink or Open Banking at all. Neither network has anything like it on UPI, Pix, PayNow or SEPA Instant, and in the markets where shoppers have actually moved significant volumes of commerce onto public rails, the networks haven’t reported any damage at the group level. Pix ran roughly 80 billion transactions in 2025, a little over four in ten of them from people to businesses. Cards in Brazil still grew 5.4% to 48.1 billion in the same year. If the public rails were a threat that needed a rulebook overlay as the response, you’d expect the response everywhere the threat is, and that hasn’t materialised.
The case that it’s a broader strategy rests on the networks’ own filings. Visa’s 10-K counts real-time payment networks in at least 80 countries, names FedNow, Pix and UPI, and describes them both as a growing alternative to card schemes and as customers for services like risk management. Its stated plan is to reach account-to-account flows through card-enabled products like Visa Pay and through Visa-branded A2A products built on Tink. Mastercard’s 10-K says much the same, listing Pix, FedNow and UPI as alternatives to its own schemes. Most of the services revenue both report is still card-linked, on their own account, but the capability they’re investing in is the one that travels to any rail, and in my view Tink and Finicity were bought to make that possible.
We also have to acknowledge the growing sovereignty clamour around the globe. There’s a case for a British rulebook on British infrastructure, and a European one for the EPI on SEPA Instant. That isn’t something pricing tiers answer, and it’s the case beneath the Bank of England’s “extra resilience” line and most of the “alternative to Visa and Mastercard” coverage in the press.
Cost, resilience and sovereignty all argue for the bare rail. The shopper’s confidence is the one thing that argues for the rulebook.
With all of that said, my view is that this is strategy, and specifically a defence of debit. McInerney’s unbundling from the Visa stack and Miebach’s different franchise rules for infrastructure are the same sentence from two chief executives, and the rulebook is the component neither has ceded. What the “challenge to the card networks” framing misses is why the rulebook matters so much to them, and for me that comes down to trust.
A regulator or a central bank can mandate a rail. Brazil made Pix free for the person paying, India wrote the price of UPI to zero, and all three countries wrote the fraud rules themselves. What none of them can do is make the shopper choose it at a checkout. In a market that grew up on cards, the shopper pays with a card because there’s a logo they recognise and a promise behind it that their parcel arrives or their money comes back. A bank transfer without that promise is just a bank transfer, and Simon Taylor made the same point in the piece I credited earlier, that “the lack of consumer protections is a genuine concern”. A merchant can supply that promise for its own store, which is what Amazon’s A-to-Z guarantee does behind its Pay by Bank button, and a bank can supply it inside its own app. Only a scheme brand carries it across every merchant and every bank, and in a card-dominant market the networks are better placed to supply that than anyone else, because the shopper already knows the brand. That’s the adoption lever, and it’s a much better explanation than a toehold for why Visa wants to sit on top of a Faster Payments rail at all. Visa’s brand is already on a large share of the debit and credit cards issued in the UK, so it was never a question of whether Visa should be in this market. The real question is why a network that already has the shopper’s confidence would want its brand on a rail it doesn’t own, and my answer is that the bare rail doesn’t come with that confidence built in.
That also tells you what’s being defended. Account-to-account moves a balance the shopper already has, so where shoppers move to it, the volume comes out of debit rather than credit. India shows the pattern, with debit card transactions down from 4.09 billion in 2021 to 1.34 billion in 2025 while credit card transactions rose from 2.16 billion to 5.7 billion, and the RBI’s own reading is that UPI, wallets and credit cards took that volume between them. Andrew Dresner has seen the same thing in the US, where the great majority of the bank-funded wallet users he looked at had previously paid with debit rather than credit. The credit line, the third part of the bundle, stays with the issuer, and in both of those markets it’s growing. Visa A2A’s first use case, bills, takes more from Direct Debit than from debit cards, which is why the eCommerce phase is the one to watch, but the direction is set. The rulebook on a public rail is how the networks stay inside the debit transaction as debit migrates, with their brand and their rules attached to a payment that no longer needs their rail, and the networks have chosen to stand where the shopper is.
So this isn’t a network conceding the rail. It’s a network deciding that where the rail has already gone public, the rules are the business, and picking a market where the rules are about to be priced to make sure its rulebook is a contender. Whoever writes the rules the shopper trusts will set how quickly account-to-account takes debit in the card-dominant markets, and at what price, and for me that is a much better question than whether Visa is losing the rail.