Before UPI, paying someone was genuinely annoying

If you're under thirty and live in an Indian city, there's a decent chance you've never written a cheque. It's worth remembering what came before, because it explains why UPI landed the way it did. Moving money between two bank accounts in India used to mean collecting the other person's account number, IFSC code, and bank branch, entering all of it correctly into a net-banking form that timed out if you looked away too long, and then waiting anywhere from a few hours to a full working day for NEFT to clear. Mobile wallets had shown up by the early 2010s, but Paytm couldn't send money to a MobiKwik user, and neither could talk to your bank account directly. Cash still did the heavy lifting for the vast majority of everyday transactions.

The groundwork had actually been laid earlier than most people realise. The National Payments Corporation of India was set up in 2008 as a not-for-profit utility, jointly backed by the Reserve Bank of India and the Indian Banks' Association, specifically to build shared payment infrastructure rather than leave it to individual banks or private companies. In 2010 it launched IMPS, the Immediate Payment Service, which quietly proved that instant, 24x7, interbank transfers were technically possible in India. UPI, when it arrived, was built as a friendlier layer sitting on top of that same IMPS backbone.

The pilot began in April 2016 with 21 banks. The public rollout followed that August. NPCI's own BHIM app came a few months later, in November 2016 — coincidentally the same month India demonetised its high-value currency notes overnight, which did more for digital payment adoption in a few weeks than any marketing campaign could have. Early users were sceptical, understandably. Sending money by typing a nine-character address instead of a sixteen-digit card number felt unfamiliar. It took a couple of years, and a wave of well-funded apps like PhonePe and Google Pay building genuinely pleasant interfaces on top of the same rail, before UPI crossed a billion transactions in a single month, in late 2019. Everything since has been that same curve, getting steeper.

What's easy to forget now is how uncertain that outcome looked at the time. Banks were nervous about a shared, government-backed rail cutting into the fee income they earned from their own net-banking and card products. Some dragged their feet on integrating cleanly, and for the first year or two, transfers between certain bank pairs were noticeably less reliable than others, which quietly pushed early adopters toward whichever combination of app and bank happened to work smoothly. None of that shows up in the retrospective growth charts, but it's a useful reminder that UPI's current reliability was earned gradually, not switched on at launch.

The trick behind the simplicity

The clever part of UPI was never the interface. It was a small piece of abstraction: the Virtual Payment Address, something like name@bank, that sits in front of your actual account number so you never have to type or share it again. Underneath that address, every transaction still moves through what's called a four-party model — your payment app, your bank, the receiver's bank, and the receiver's payment app — with NPCI acting as the neutral switch in the middle that routes the message and guarantees settlement.

Two design choices made this durable rather than just clever. First, interoperability was mandatory from day one: any UPI app can talk to any participating bank, so switching from one app to another has never meant losing access to your money or your contacts, the way switching wallets used to. Second, and this is easy to overlook, NPCI is not a private, for-profit card network extracting a toll on every swipe. It's a public utility. That single fact is why UPI could offer merchants a zero merchant discount rate on person-to-merchant payments, something Visa or Mastercard were never structurally going to do, and it's the single biggest reason a roadside vendor selling tea has the same access to digital payments as a shopping mall.

A payment rail owned by no one in particular turned out to be exactly what let everyone use it.

There's a third design choice worth naming, because it rarely gets credit: the QR code standard itself is shared and interoperable, not owned by whichever app happens to print it. A single Bharat QR sticker outside a shop can be scanned by any UPI app, from any bank, without the merchant needing to display five different stickers for five different apps. Compare that with how card terminals or closed-loop wallets typically work elsewhere, where accepting one network's cards sometimes means separate agreements, separate hardware, or separate stickers for each competing service. Standardising the QR code the same way UPI standardised the payment address meant a shopkeeper only ever has to deal with one sticker, not a wall of them.

August 2026, in numbers

Scale is easy to state and hard to actually picture, so here's one month, laid out plainly. Divide 24.51 billion transactions across August's 31 days and you get roughly 790 million UPI payments going through every single day in India — more than the entire population of Europe, transacting once each, before breakfast.

Figures from NPCI's published transaction data and the IMF's 2025 review of real-time payment interoperability. The 49% share reflects 2024 data, the most recent year for which a global comparison exists.

What actually changed this year

Two features got the stage time at the Global Fintech Fest in 2026, and they're worth separating from each other because they solve different problems.

UPI Tap & Pay brings genuine contactless payment to UPI for the first time. You unlock your phone, hold it near a compatible point-of-sale terminal, and the payment goes through over NFC — no need to open an app, scan a QR code, or fumble through a scan-and-confirm sequence while a queue builds up behind you. It supports both regular UPI payments and RuPay credit cards linked to UPI, and pin-less transactions are capped at ₹5,000 for now. It's a small change in mechanics but a real one in feel: it's the difference between "paying digitally" and paying the way you'd tap a transit card.

My UPI is the more interesting long-term bet. It's an AI-powered layer, built on FiMI, NPCI's own small language model, that's meant to become a single place to manage everything UPI-related instead of jumping between your bank's app, your payment app, and a customer-care number when something goes wrong. Think dispute resolution, fraud reporting, and account-level visibility, handled conversationally, without having to know which of your five apps is actually responsible for the problem.

Alongside those launches, NPCI quietly raised the ceiling on biometric authentication — face or fingerprint unlock in place of typing a PIN — from ₹5,000 to ₹10,000 per transaction, effective the second week of August 2026. That sounds like a minor tweak until you see the adoption curve behind it: biometric-authenticated UPI payments crossed 600 million transactions in June 2026 alone. People clearly prefer not typing a PIN in a crowded shop, and NPCI is responding to that preference rather than fighting it.

None of this happened in isolation. It builds on features that quietly reshaped UPI over the previous couple of years: Credit Line on UPI, which lets a bank's pre-approved credit facility be spent directly through UPI without a physical card, blurring the line between a payment method and a short-term loan; UPI Lite, an on-device wallet for small payments that skips additional authentication and works with patchy connectivity; and UPI 123Pay, which brought UPI to feature phones through IVR calls and missed-call triggers for the tens of millions of Indians who still don't own a smartphone. Together, these are less about adding features for their own sake and more about closing the remaining gaps — speed, credit access, and reach — one at a time.

It's worth sitting with UPI 123Pay for a moment, because it's easy to overlook next to something as visibly futuristic as an AI assistant. A meaningful share of India's population still uses a basic keypad phone, either by choice or because a smartphone and a data plan simply aren't affordable yet. For that group, 123Pay works through four channels — a missed call to a designated number, an IVR menu you navigate with keypad presses, an app that runs on limited-feature phones, and a sound-based proximity channel for offline retail. None of that is glamorous, but it's the difference between "digital payments" being something available to smartphone owners and something available to almost anyone with a bank account, which was always meant to be the point.

The boring rules that matter more than the flashy features

If you only read the press releases, you'd think 2026 was all about AI assistants and tap-to-pay. The changes that will actually matter to how safely and reliably UPI runs came from a set of far less exciting operational rules.

The first is a small interface change with an outsized effect on fraud: apps must now show you the receiver's bank-verified name before you confirm a payment, not just whatever display name they've set. A large share of UPI scams have always relied on a mismatch between what a victim expects and who actually receives the money — this closes that gap at the exact moment it matters, right before you tap confirm.

The second is a restriction on person-to-person Collect requests, the feature that let someone send you a request to pull money from your account rather than push it to you. It was useful for splitting a dinner bill, and it was also, for years, the single most abused mechanism for phishing — a message arrives that looks like a refund or a cashback offer, you approve what you think is money coming in, and it's actually money leaving. Restricting how and when P2P collect requests can be sent removes a lot of that attack surface in one move.

The third is more of a plumbing fix than a user-facing feature: NPCI capped how often apps can hit the backend for things like balance checks and status checks — roughly 50 such calls per app per day per user — after high-frequency polling from some apps was found to be a real contributor to system slowdowns during peak periods like festival sales or tax-payment deadlines. It's not the kind of change anyone notices directly, but if you've ever had a UPI payment hang for what felt like forever on a big sale day, this is the category of fix meant to stop that from recurring.

Finally, the Reserve Bank of India folded its various circulars on recurring payments into a single Digital Payments E-mandate Framework in April 2026, covering UPI AutoPay alongside card and prepaid-instrument mandates. Once you've registered a mandate, recurring debits up to ₹15,000 can go through without a fresh authentication step each time, while FASTag and transit-card top-ups are exempted from the usual 24-hour advance notice rule, since nobody wants their toll payment held up by a notification period.

Taken as a set, these four changes share a pattern that's easy to miss if you look at them one at a time: none of them add a new thing you can do with UPI. Every one of them takes something that already worked and makes it slightly harder to abuse or slightly less likely to buckle under load. That's a very different kind of progress from a new AI assistant or a tap-to-pay demo on a conference stage, and it's arguably the harder kind to get right, because it means finding problems before they become headlines rather than reacting to ones that already have.

Ten years, compressed

Tap into any one of these to see what actually happened that year.

The hockey-stick, in bars

Reading the milestones above is one way to see the growth. Looking at the raw monthly numbers, year over year, is another — and it makes the last five years look almost vertical. These are month-specific snapshots rather than annual totals, so treat the shape of the climb as the real story, not the last decimal place.

Monthly transaction volume in billions, at each year's most recently published figure — December for most years, August for 2023 and 2026. Source: NPCI transaction data.

The quiet redesign of the Indian economy

It's tempting to talk about UPI purely as a convenience story, but the more consequential effects have been structural. The Pradhan Mantri Jan Dhan Yojana had already pushed bank account ownership into places it had never reached, opening hundreds of millions of no-frills accounts through the 2010s. UPI is what made those accounts actually useful day to day, rather than dormant lines on a government report. A bank account you can only access by visiting a branch is a very different thing from one you can use to buy vegetables from a cart.

The kirana store on the corner, the auto-rickshaw driver, the vendor selling snacks outside a railway station — these are the users who benefited most, not because UPI was designed for them specifically, but because a QR code sticker costs nothing to print and a bank-owned rail costs nothing to accept payment on. Millions of small merchants who previously operated entirely in cash now have a digital trail of their income, which sounds bureaucratic until you realise what it enables: a lending history. Fintech lenders and government-backed credit schemes increasingly use UPI transaction data, rather than traditional collateral, to underwrite small loans to exactly the kind of micro-business that used to be invisible to formal credit.

There's a less measurable but real shift too, in who gets to transact independently. Women who previously routed money through a husband's or father's bank relationship now often hold and use their own UPI-linked account. None of this shows up cleanly in a single statistic, but taken together, it's a genuine, if gradual, formalisation of an economy that ran largely on cash for decades.

The gig economy has probably felt this shift more directly than any other segment. Delivery riders, cab drivers, and home-service workers on platforms like Swiggy, Zomato, Uber, and Urban Company get paid out through UPI-linked accounts on a near-daily basis, which means their income, however irregular, is now visible and verifiable in a way it never was when it arrived as cash tips or informal settlements. That visibility is exactly what a lender needs to assess risk, and it's exactly what platforms and insurers have started building products around — pay-as-you-go insurance tied to a rider's actual UPI-verified earning pattern, for instance, rather than a flat premium that either overcharges or under-protects. None of this existed as a real business category a decade ago, and almost all of it depends on UPI's transaction data being clean, timestamped, and consistent across providers.

Exporting the idea, not just the app

Somewhere around the mid-2020s, the conversation around UPI shifted from "impressive domestic payment system" to "template other countries might want to copy." That shift is worth taking seriously.

As of the middle of 2026, UPI is accepted for payments in eleven countries beyond India: Bhutan, Singapore, France, Mauritius, Nepal, Sri Lanka, the UAE, Qatar, Cambodia, Greece, and the Maldives, with NPCI International Payments Limited, NPCI's dedicated international arm, striking the partnerships behind each rollout. Cross-border UPI transactions grew roughly twentyfold in a single year, from about 37,000 in one financial year to over 750,000 the next, and more than 1.5 million merchants abroad now accept it. India's government has publicly said it wants to push further into East Asia next.

Each of these rollouts looks slightly different depending on the partner. In France, acceptance started narrowly, with UPI enabled at a small number of high-footfall tourist sites frequented by Indian visitors, before merchant coverage broadened. In Singapore, the more ambitious move was linking UPI directly with PayNow, the domestic Singaporean instant-payment system, so that money can move both ways between the two countries without either side's users needing to touch a card network or a traditional wire transfer at all — arguably the clearest working example yet of the "network of networks" idea in practice, rather than just India's app being accepted on a foreign terminal. Gulf countries like the UAE and Qatar, with their large resident Indian populations, adopted UPI partly as a remittance and daily-spending tool for that diaspora rather than purely a tourist convenience.

The more interesting part is why other countries are interested at all. India's own G20 presidency in 2023 leaned heavily on the idea of Digital Public Infrastructure — payment rails, identity systems, and data-sharing frameworks built as public goods rather than owned by any single company — with UPI as the flagship example. That framing matters geopolitically. China's Alipay and WeChat Pay expanded internationally earlier and more aggressively, but they carry the baggage of being closed, single-company ecosystems with real data-sovereignty concerns attached. UPI's pitch is structurally different: an open protocol that a country's own central bank and banks can run locally, connecting to India's system through standard interfaces rather than routing everything through a foreign company's servers. Whether that pitch fully holds up in practice is still being tested, but it's a genuinely distinct model, and it's part of why smaller countries in Africa and Southeast Asia have been in conversation with NPCI about building similar rails rather than just plugging into someone else's.

The long-term vision on the table isn't one global payment app. It's a network of national payment systems that talk to each other, the way separate national postal services always could.

UPI against the alternatives

It helps to see UPI next to the payment methods it's been quietly replacing. None of these methods are strictly obsolete — a large one-off property payment still often makes more sense over a wire transfer, and cash still has its place — but the table below is a reasonably honest picture of what each one is actually good for in 2026, rather than what it was built for a generation ago.

Where it still creaks

None of this is as tidy as the growth charts make it look. The most persistent structural argument is economic: zero merchant discount rate means banks and payment apps carry the full infrastructure cost of processing merchant transactions with no direct fee to offset it. Government compensation schemes have plugged that gap so far, but they're negotiated and renewed rather than permanent, and every renewal cycle brings fresh industry pressure to introduce some fee on larger merchants. Doing that without undermining the "free to accept" promise that drove adoption in the first place is a genuinely hard balance to strike.

There's also a quieter concentration risk worth naming. Two apps, PhonePe and Google Pay, still handle the overwhelming majority of UPI's volume between them, which means a serious outage or policy misstep at either company has an effect on India's payment system disproportionate to what "just one app having a bad day" should mean. NPCI has floated market-share caps for individual apps before, partly for this reason, though enforcing a hard cap without disrupting users mid-flow has proven politically and technically difficult, and the deadline for it has been pushed back more than once.

What's likely next

A few directions look reasonably safe to bet on. My UPI is likely to expand well beyond customer support into proactive fraud detection, flagging suspicious patterns before a payment even completes rather than after a complaint is filed, and doing it in more of India's languages than English and Hindi alone. Biometric authentication limits will probably keep climbing as trust in the mechanism grows, following the same path the PIN itself took over the previous decade.

There's also a credible case for UPI eventually interlinking with the e-Rupee, the Reserve Bank's central bank digital currency, which could enable genuinely offline, programmable payments — money that can carry conditions on how or when it's spent, useful for anything from targeted subsidies to corporate expense controls. Credit Line on UPI is likely to deepen further too, continuing to blur the boundary between "payment method" and "short-term loan" in ways regulators will need to watch closely.

Internationally, the stated ambition is East Asia next, alongside continued interest from African and Latin American central banks exploring similar public-infrastructure models rather than building payment rails from scratch or leaning on a single private network. Whether UPI becomes one node in a genuinely interoperable global mesh of national payment systems, or stays mainly an India-plus-a-few-partner-countries story, is probably the single most interesting open question in this space for the rest of the decade.

It's also worth watching what happens to the zero-MDR debate over the next couple of budget cycles. The compensation scheme that currently reimburses banks for processing free merchant payments gets revisited annually, and every year brings renewed lobbying from banks and larger payment apps for some form of merchant fee, at least on high-value transactions from big-box retailers who can plainly afford it. Whichever way that goes will shape whether UPI's next decade looks like its first — free, ubiquitous, subsidised by the state — or whether some segment of it starts to look more like the fee-based card networks it originally displaced.

Questions people actually ask