Opinion | India's UPI Miracle Has Created A Difficult, Rs 20,000-Crore Question

Parliament has reopened a question that the country thought was settled: should UPI ever charge? On August 6, the Lok Sabha passed the Taxation and Other Laws (Amendment) Bill, 2026, by a voice vote, and on August 10, the Rajya Sabha, since this is a Money Bill, returned it after discussion, completing the process. It now goes to the President for assent. The change reads small on paper and large in effect. It amends Section 10A of the Payment and Settlement Systems Act, 2007, which until now shielded UPI and RuPay debit from any charge because they were prescribed under the Income-Tax Act. That automatic shield is gone. From here, whether a payment mode stays free depends on what the government chooses to notify.

The case for charging begins, reasonably enough, with a bill nobody disputes. Running the world's largest real-time payment network is not free. Every tap that looks instant sets off authorisation, settlement, reconciliation, fraud screening, dispute handling, cloud capacity, and round-the-clock monitoring, and all of it costs money. One estimate puts the yearly cost of running UPI and RuPay debit infrastructure as high as ₹20,000 crore. As volumes keep climbing - July 2026 alone saw 2,366 crore transactions worth ₹29.9 lakh crore - that bill only grows. So the question the new law opens is fair: if security and a coming wave of AI need real investment, must UPI finally charge?

Right Problems, Wrong Solutions

The costs behind all this are not imaginary. The National Payments Corporation of India (NPCI) chief has said AI will define UPI's next chapter, taking it from about 750 million daily transactions towards a billion, onboarding half a billion more users, and hunting the mule accounts that already run into lakhs of cases a year. Voice onboarding in many languages, models trained on financial data, and real-time fraud engines are not cheap, and all of it needs steady funding. The RBI Governor has put the question plainly: who, in the end, pays for the plumbing?

It helps to recall how that plumbing was paid for, because "free" never meant "unfunded". UPI was built and run by NPCI as public infrastructure, not a profit centre, and its scaling and security were financed the way a public good is: from the exchequer. The government reimbursed banks through an incentive scheme, ₹1,389 crore in 2021-22 and a peak of ₹3,631 crore in 2023-24, paying 0.15% on low-value merchant payments. That model carried UPI from a standing start in 2016 to accounting for nearly half of all real-time payments on the planet.

The strain is that this public funding has thinned even as the network has swelled. The Standing Committee on Finance found that incentive payouts of ₹8,730 crore over four years to 2024-25 covered barely 11% of the industry's cost. The 2026-27 budget set aside ₹2,000 crore, less than before, even as daily volumes touch record highs. So banks and processors now carry a rising share of a growing bill, and want durable revenue, not a subsidy that shrinks each year.

Where It Hurts The Most

And yet, the leap from "the costs are real" to "therefore charge each transaction" skips the harder question of design. Most of what UPI spends on is fixed, infrastructural cost, not something that rises neatly with each ₹40 payment to a vegetable seller. A per-transaction fee is a blunt tool for a largely fixed bill. Worse, it lands where it does the most damage: at the point of acceptance, on the smallest sellers, whose whole reason to accept digital money was that it cost nothing.

The sums show why. Transactions above ₹2,000 make up barely 4% of UPI's volume but nearly two-thirds of its value. The vast bulk of UPI is tiny, like the auto fare, the chai, the ₹30 vada pav. Put even a small charge on that layer, and the rickshaw puller and the roadside vendor do the rational thing and ask for cash. A fee at the bottom of the pyramid is a nudge back to notes, and cash is untraceable, informal, and off the books, the very thing UPI was built to shrink.

To her credit, the Finance Minister seems to grasp this, and has said the amendment is only an enabling provision that imposes no charge on users, that no MDR (merchant discount rate) framework has yet been finalised, and that ordinary low-value payments will stay free. She also reiterated that any future MDR would fall only on a limited category of merchant transactions above a prescribed threshold. That is, almost word for word, the tiered design this column would argue for.

But there is a catch. All of that is a promise, not a provision. The Bill itself writes none of it down; it simply moves the decision into executive notification, to be shaped by a steering committee under NPCI. A protection that lives in a minister's assurance can be revised by the next notification, without returning to Parliament. Opposition members made exactly this point, calling the reply reassuring in tone but thin on paper.

Working Smarter

Better funding routes exist, and they deserve open debate rather than a quiet notification. The RBI transferred a record surplus of about ₹2.69 lakh crore to the government in 2024-25; earmarking even a sliver of it would dwarf today's incentive and cost the ordinary user nothing. Credit on UPI, RuPay cards and pay-later products already earns interchange that can help fund the free rails beneath them.

So, do security and AI need more money? Almost certainly. Must UPI charge the ₹30 payment that is its whole miracle? No. The Bill has passed, and the promises made alongside it are the right ones. The task now is to write those promises into the notification as a statutory floor for small merchants and consumers, and to fund the machine without sending the rickshaw puller back to cash.

(Subimal Bhattacharjee is a policy adviser on digital tech issues and the author of 'The Digital Decades: Thirty years of the Internet in India')

Disclaimer: These are the personal opinions of the author