Skip to main content
Back to blog
  • Fintech

Updated 14 min read

Pay by bank vs cards in 2027

Account-to-account (A2A) payments hit 24 billion a month in India. Where they beat cards, where cards hold, and what 2027 changes

In September 2026, India’s UPI handled 24 billion payments. In the whole third quarter, America’s FedNow handled 6.4 million.

Both move money straight from one bank account to another, with no card in the middle. One took over a country. The other is still looking for its job.

Across the 65 banking and payments reports we read, 12 from McKinsey, BCG, Accenture, Deloitte, Mastercard, PwC and Visa carry the numbers below. We cross-checked them with central banks and the scheme operators.

For a bank, the stakes sit at the checkout. McKinsey puts US credit card interchange at about $94 billion in 2025. A card payment that moves to a bank rail takes its share of that fee with it.

The reports mostly agree on the pattern. Pay by bank wins under three conditions, cards hold on three things, and 2027 brings three changes.

The first condition is one America’s rails have not met.

Sign up with email

The short version

  • Where A2A won, someone set the price near zero. India’s UPI ran 24.07 billion payments in September 2026, NPCI data show, on rails that stayed free for most users.
  • Brazil’s Pix now pays shops more than friends. In September 2026, 3.3 billion Pix payments went from people to businesses and 2.8 billion went between people, by our count of central bank data.
  • In card markets, pay by bank is still small. UK cardholders made about 70 card payments for every open banking payment in June 2026.
  • Cards hold on credit, rewards and refunds. In Brazil, credit cards still carry about 60% of higher-income customers’ transactions, McKinsey found.
  • 2027 brings three changes. Recurring pay by bank goes commercial in the UK, Europe’s Wero heads for the till, and India puts a 0.4% fee on large UPI merchant payments.

1. Pay by bank wins where someone sets the price near zero

McKinsey’s payments team puts instant payment flows at nearly $22 trillion in 2024, across the 15 largest economies that run them. That is about 1% of the $2 quadrillion the world moves. They expect it to grow 15% to 18% a year for the next five years.

Most of the story comes from two countries.

India’s UPI handled 24.07 billion payments in September 2026, worth Rs 29.37 lakh crore, according to NPCI. It has 756 banks live. BCG and NPCI put UPI at 84% of India’s digital retail payments.

Brazil’s Pix ran almost 80 billion transactions in 2025, up 25.7% on 2024, the Banco Central do Brasil reports. In December 2025, 148 million people and 12 million companies sent or received a Pix. McKinsey puts instant payments at nearly 30% of all transactions in Brazil.

Look at what people use Pix for now. In September 2026, 3.3 billion payments settled in the central bank’s system went from people to businesses. Another 2.8 billion went between people.

Pix started as a way to pay friends. Today, paying shops is its biggest use.

Figure 1. Pix payments by type, September 2026 (millions)
TypeValue
Person to business3,330
Person to person2,832
Business to person669
Business to business308

Source: Banco Central do Brasil, Pix open data (EstatisticasTransacoesPix), September 2026, our count of payments settled in the central bank’s system.

McKinsey’s team (Grace Klopcic, Gustavo Tayar, Roberto Marchi, Uzayr Jeenah and colleagues) explains why these two worked. Three things came together:

  1. Everyone had to join. Brazil’s central bank required large banks to connect to Pix.
  2. The price was near zero. India set zero fees for merchants and customers, and paid banks subsidies for small payments.
  3. New features kept coming. Pix added installments. UPI added RuPay credit cards.

Mexico shows what happens without them. Its central bank launched CoDi in 2019 and DiMo in 2023. Joining was voluntary, and banks set their own prices. Instant payments are still under 5% of Mexico’s transactions, McKinsey found.

What to do in 2027. In a market like India or Brazil, the payment itself is no longer the product. Pix already makes up about 40% of the sales mix of small and mid-size Brazilian firms, according to McKinsey’s 2025 survey. Banks there moved credit cards toward rewards, premium benefits and installments. The value sits around the payment now: credit, merchant software, data.

2. In card markets, pay by bank is still a side road

The UK has had open banking since 2018. Its numbers grow fast, and they are still small.

Open Banking Limited counted 40.16 million open banking payments in June 2026. Monthly payments passed 40 million for the first time in May. Payment volume grew 45% in the year to July 2026, OBL data scientist Anton Joachim writes.

In the same month, UK cardholders made 2,336 million debit card payments and 439 million credit card payments, UK Finance reports. That is roughly 70 card payments for every open banking payment.

McKinsey calls this the complement model. In the UK and the rest of Europe, instant payments are about 10% of retail payments. They improve transfers between accounts. Cards stay the default at the till.

The US is further behind. FedNow settled 6.38 million payments in the third quarter of 2026, up 27.6% on the quarter, the Federal Reserve reports. More than 1,800 banks and credit unions are live on it. The Clearing House’s RTP network is bigger, at about 128 million payments in the first quarter of 2026, the Richmond Fed notes.

Put that next to the whole market. McKinsey counts about 345 billion US payments a year. Zelle alone ran 4.2 billion in 2025, about nine times RTP and FedNow together. And of some 9,000 US banks and credit unions, many of those on the instant rails can only receive money, not send it.

McKinsey’s explanation is short. In the US, neither rail came with mandated participation or near-zero pricing. Banks chose their own pace, and users had no money reason to switch. Instant payments are also hard to monetize. Credit cards are not.

Singapore sits in between. PwC counts FAST, the A2A rail behind PayNow, growing 35.8% a year by volume from 2020 to 2024, against 17.2% for cards. Cards still lead at the till.

Figure 2. A2A and card payments per month, latest data (millions)
RailValue
India UPI, September 202624,068
Brazil Pix in the central bank system, September 20267,170
UK card payments by UK cardholders, June 20262,775
UK open banking payments, June 202640
US FedNow, Q3 2026 monthly average2.1

Source: NPCI; Banco Central do Brasil; UK Finance; Open Banking Limited; Federal Reserve Financial Services. Our rounding and monthly average.

What to do in 2027. In a card market, don’t sell pay by bank as a cheaper card. Sell it where cards are weak: wages, payouts, wallet top-ups, bills and transfers between accounts. That is where McKinsey sees US instant payments working today.

3. Cards hold on three things: credit, rewards and the right to a refund

Credit. In Brazil, Pix took a lot of debit card spending. Credit cards held on. Among higher-income customers, they still carry about 60% of transactions, McKinsey found, because of their benefits and rewards.

Rewards. In the US, McKinsey puts gross credit card interchange at about $94 billion in 2025. About $83 billion of it went back to cardholders as rewards, which leaves a net pool of $11 billion. A bank rail has no interchange to pay for points. So a customer who chases rewards has little reason to move.

Refunds and fraud. A card payment comes with a chargeback. A push payment from your own account usually does not. Deloitte expects US losses from authorized push payment fraud to grow from $8.3 billion in 2024 to $14.9 billion in 2028, in its baseline case. Visa’s team notes that as instant payments scale in Europe, the window to catch a fraud shrinks to milliseconds.

The same logic reaches AI agents. McKinsey expects agents in card-heavy markets like the US and Western Europe to keep routing through card rails. Cards are still the standard for consumer protection and liability.

What to do in 2027. If you issue cards, your defense is the three things above, so make them easy to see. If you build pay by bank, close the gaps one at a time: a credit option at the moment of payment, a refund promise the customer can see, and fraud checks that run before the money leaves. Brazil’s banks did the first with Pix installments. India did it by putting RuPay credit cards on UPI.

4. Three changes that land in 2027

Recurring pay by bank goes commercial in the UK

On 2 June 2026, UK Payments Initiative (UKPI) launched the first new UK payment scheme since Faster Payments in 2008. Its members include HSBC, Barclays, Lloyds, NatWest, Nationwide, Santander, Monzo, Revolut and Starling. A customer approves a business once in their banking app. After that, the business can collect varying amounts within the limits the customer set.

The first uses are payments to government, utilities, charities and financial services. Subscriptions and wider ecommerce come later. The FCA expects HM Treasury to bring legislation in 2026 that gives it power to set open banking rules. It plans to assess commercial VRP adoption at the end of 2026.

The demand is already visible. Sweeping VRPs, where people move money between their own accounts, reached 7.73 million in June 2026. That is about 19% of all UK open banking payments. Brazil is on the same path. Pix Automático, its recurring version, ran nearly 600,000 payments in December 2025.

Direct debit and card-on-file are the targets here. For a bank, the 2027 question is simple. When a customer approves a recurring payment, does it happen in your app, with your controls?

Europe’s own wallet heads for the till

Wero, the A2A wallet from the European Payments Initiative, has more than 57 million users, EPI said in July 2026. It runs on SEPA instant transfer rails. Online payments went live in Germany at the end of 2025 and in Belgium in March 2026. EPI plans in-store payments in 2026 and the move of the Dutch iDEAL scheme to Wero by 2027. Lidl is among the first big merchants.

Regulation pushes the same way. In BCG’s survey of 12 European banks, led by Mireia Granzer and colleagues, instant payments legislation ranked as the hardest regulatory challenge, with an average score of 2.7 on a scale where 1 is the hardest. 90% of respondents say they urgently need to modernize their payments systems.

A note of caution. Registered users are not active payers. In January 2026, S-Payment, the Sparkassen payment company, said about 3.5 million people at participating German banks had activated Wero.

Free rails start to charge, and agents start to choose

In September 2026, India set a fee on large UPI merchant payments. It is 0.4% on person-to-merchant payments above Rs 2,000, capped at Rs 300. Payments up to Rs 2,000 stay free, and they make up more than 95% of merchant volume, India’s finance ministry says. The same ministry puts credit card fees at 1.5% to 2.5%. The start was set for October 15. NPCI then pushed it back to after Diwali, with no new date yet.

The Indian Express reports that payments over Rs 2,000 are only 4% of merchant UPI payments by number, but about two-thirds by value. So the fee lands where most of the money is. I read this as a sign that a zero price gets a rail to scale, and a rail at scale still needs money to run.

Meanwhile, AI agents may start to choose the rail. Accenture estimates that if agents move half of card transactions to cheaper methods, payments players lose about $7.2 billion in revenue. McKinsey’s base case has agents disrupting about 30% of US net credit card interchange by 2030, and up to 75% in a rewards war.

Providers are getting ready for that choice. Mastercard surveyed 300 executives and 8,000 consumers in 11 countries. Of the firms, 71% already offer A2A payments, and 21% more plan to within two years.

Where the reports disagree

How big UPI is. McKinsey writes “more than 19 billion” a month. Accenture writes more than 15 billion, and BCG over 20 billion, then 23.2 billion for May 2026. Each is right for its date. NPCI’s own figure for September 2026 is 24.07 billion, and that is the one I use.

Whether agents favor cards. McKinsey expects agents in the US and Europe to stay on card rails for protection. Accenture expects agents to pick the cheapest rail. They may both be right: card rails for the protected purchase, bank rails for the bill and the payout.

How many US banks are on FedNow. PwC writes 1,400. The Richmond Fed counted 1,725 in the first quarter of 2026. McKinsey and the Federal Reserve say about 1,800. The count is rising fast, so each source catches a different moment.

What I’m watching in 2027

  • What UPI merchant volume does once the fee starts after Diwali.
  • Whether UK legislation arrives in time for commercial VRP to reach ecommerce.
  • Wero’s first in-store payments, and how many of its 57 million users actually pay with it.
  • McKinsey’s own test for the US: whether A2A volumes start to approach Brazil and India.

The more of these reports I read, the less the rail itself seemed to decide. What sits around it does: credit, refunds and the moment a customer says yes. The teams that design those well will keep the customer, whichever rail the money takes.

Sources

Let's talk

Have questions? Ask AI
Opens a new chat with context about us pre-loaded — ask anything

We’ll reply within 24 hours with case studies, a timeline, and an estimate.

Prefer email? Write to hi@humbleteam.com
All set – our team’s on it. Expect a reply soon.
Send another one
Oops! Something went wrong while submitting the form.
Have questions? Ask AI
Opens a new chat with context about us pre-loaded — ask anything
Europe
Národní 135/14, Prague
Middle East
UAE, Dubai, Internet City Offices
We use cookies to enhance your browsing experience,
serve personalised ads or content, and analyse our traffic.
By clicking "Accept All", you consent to our use of cookies.
Privacy policy