For nearly a decade, one of UPI’s most powerful features was almost invisible.

It was not speed. It was not the QR code. It was not even interoperability.

It was the fact that, for most users and merchants, using it felt free.

That helped turn the Unified Payments Interface from a payments experiment launched in 2016 into the backbone of India’s digital economy. In August 2026 alone, UPI processed 24.5 billion transactions worth ₹29.82 lakh crore, according to data from the National Payments Corporation of India. More than 750 banks were live on the network that month.

Now that model is changing.

From October 15, 2026, a Merchant Discount Rate, or MDR, of 0.4% will apply to specified person-to-merchant UPI payments above ₹2,000. Transactions worth ₹75,000 or more will have the fee capped at ₹300. Person-to-person transfers will remain free, as will payments up to ₹2,000 and qualifying payments received by small merchants. Essential sectors such as railways, telecom, insurance and fuel will instead face a flat ₹5 charge on applicable transactions.

The government says approximately 96% of merchant UPI transactions will remain outside the MDR framework. Consumers are also not supposed to be charged the MDR directly.

On the surface, then, this looks like a debate about 0.4%.

It is not.

The larger question is what happens when a technology succeeds so completely that society begins to treat it less like a product and more like infrastructure.

And when that happens, who pays to keep it running?

UPI has crossed the line from product to infrastructure

The scale of UPI makes the distinction important.

The government says UPI handled more than 24,162 crore transactions worth over ₹314 lakh crore during FY2025–26. Its transaction volume has grown roughly 12,000-fold since FY2016–17.

By June 2026, 55.49 crore users had been onboarded onto UPI.

Reuters reported that UPI represented roughly 84% of India’s digital payment transactions by volume in August and approximately 49% of global real-time payments.

These numbers change the nature of the problem.

A payments app can change its pricing model and lose customers.

A piece of infrastructure cannot make the same decision so casually.

Millions of consumers now assume that UPI will work. Merchants build checkout experiences around it. Banks have built systems to support it. Startups incorporate it into products. Entire categories of businesses have been created around digital payments.

When that degree of dependence exists, pricing stops being merely a commercial decision.

It becomes part of the architecture of the economy.

The zero-fee era was never actually free

The phrase “free UPI” can also be misleading.

Removing an MDR did not remove the cost of processing payments.

Someone still had to pay for servers, fraud-monitoring systems, customer support, reconciliation, banking infrastructure, merchant onboarding, security, product development and network expansion.

For several years, part of that cost was deliberately supported by the government.

In FY2021–22, the government paid ₹957 crore in incentives for BHIM-UPI transactions. That rose to ₹1,802 crore in FY2022–23 and ₹3,268 crore in FY2023–24. For FY2024–25, the Cabinet approved another ₹1,500 crore incentive scheme, specifically supporting low-value UPI payments to small merchants.

That policy had a clear objective: accelerate digital payment adoption by removing transaction friction.

And it worked.

But successful subsidies often produce a second question that is more difficult than the first:

When does an adoption incentive become a permanent operating model?

PhonePe CEO Sameer Nigam has argued that the answer cannot simply be “forever”.

Nigam says banks and fintech companies spend approximately ₹10,000–12,000 crore every year supporting UPI at current volumes and that the industry cannot indefinitely rely on government subsidy.

That figure comes from an industry participant rather than an independent audit, but it captures the economic argument now emerging from payment companies.

If UPI must continue expanding, become more resilient and absorb increasingly sophisticated fraud and cybersecurity threats, someone must generate enough revenue to fund those investments.

Then came the other argument: hasn’t UPI already paid for itself?

Former BharatPe co-founder Ashneer Grover has attacked the new levy from almost the opposite direction.

His argument is that UPI already generates significant indirect economic value for banks, the government and the financial system, and that those gains should help finance the infrastructure instead of introducing charges into the payment flow.

It is a provocative argument because digital infrastructure does create benefits that are difficult to capture in a simple transaction fee.

UPI reduces dependence on cash.

It can lower cash-handling costs.

It creates electronic transaction histories.

It brings merchants into formal financial channels.

It facilitates financial products built on digital payment behaviour.

And at a national level, digital payments can improve efficiency across commerce.

The difficulty is that the organisations generating those benefits are not necessarily the organisations paying all the operating costs.

That mismatch is the heart of the debate.

UPI may create enormous economic value collectively while still producing weak economics for individual participants responsible for operating parts of the network.

Both things can be true at the same time.

0.4% looks small. At UPI scale, it isn’t.

The new MDR is far lower than typical card-processing fees. Nigam has pointed out that credit-card MDRs commonly range around 1.7% to 2.5%, making the 0.4% UPI rate comparatively low.

But percentage comparisons do not tell the entire story.

At UPI’s scale, even a narrow levy creates substantial revenue.

Citi analysts estimate that the new framework could eventually generate approximately ₹16,000–17,000 crore in annual revenue, distributed across banks, payment apps and aggregators.

Other estimates cited by Moneycontrol range from ₹13,000 crore to ₹26,000 crore.

That helps explain why financial markets responded quickly.

Shares of several Indian payment companies and banks rose after the MDR announcement, as investors recalculated how a previously zero-MDR ecosystem might translate into earnings.

PhonePe, for example, processes about 11.5 billion UPI transactions worth more than ₹14 lakh crore each month and controls more than 45% of UPI transaction volume, according to Moneycontrol.

Nigam has said the MDR decision could also strengthen the company’s case ahead of a potential public listing.

That creates a politically sensitive tension.

A measure presented as necessary for infrastructure sustainability also creates large new commercial revenue pools.

That does not make the fee inherently wrong.

But it makes transparency around how the money is distributed, invested and governed much more important.

Retailers see another risk: behaviour changes

The economics do not end with banks and fintech firms.

Retailers ultimately sit at the point where the fee is incurred.

Several merchant groups have warned that additional payment costs could encourage some businesses to steer customers toward cash, particularly in sectors with thin margins. The Retailers Association of India and Clothing Manufacturers Association of India have raised concerns, and the framework has also drawn objections from parts of the brokerage industry.

This is where the UPI story becomes a technology-design problem rather than merely a pricing problem.

Infrastructure depends on incentives.

A 0.4% charge may appear small from a systems perspective.

For a merchant operating on a 2% or 3% margin, however, it may not feel small at all.

That does not necessarily mean merchants will abandon UPI. The vast majority of transactions will remain outside the fee regime, and the network effects surrounding UPI are extremely strong.

But economic systems respond to margins.

A merchant could favour cash for certain purchases.

A platform could adjust pricing.

A bank might aggressively pursue merchant acquiring because the economics have improved.

Fintech firms could increase investment because each incremental payment now creates more direct revenue.

The policy challenge is therefore not simply to generate money.

It is to generate enough money to sustain the infrastructure without reintroducing the friction that the infrastructure was designed to remove.

This is what happens when technology becomes a utility

UPI is an unusually visible example of a much broader technology problem.

In the early stage of a technology platform, the objective is often adoption.

Make it inexpensive.

Make it accessible.

Encourage businesses to integrate it.

Allow developers to build on it.

Create network effects.

Scale quickly.

If those efforts succeed, the technology can eventually become foundational.

Then the incentives change.

Reliability matters more.

Cybersecurity becomes more expensive.

Governance becomes harder.

Downtime becomes economically consequential.

Fraud becomes systemic rather than isolated.

And monetisation decisions can affect millions of people who never consciously chose to participate in the platform’s original business model.

We have seen versions of this problem elsewhere.

Cloud computing has become core infrastructure for businesses.

Open-source software powers enormous portions of the internet while maintainers continue to debate sustainable funding.

Digital identity systems create enormous public utility while raising questions about who should finance and control them.

Artificial intelligence may ultimately confront the same tension if foundation models and compute infrastructure become basic layers on which businesses and public services depend.

The central question is consistent:

Who should finance technology once everyone benefits from it?

There are only a few possible answers

At a high level, infrastructure can be funded in three ways.

The state pays.

Government treats the technology as a public good because its wider economic benefits justify taxpayer support.

That preserves low-cost access, but it also creates a continuing fiscal burden and potentially makes infrastructure investment dependent on annual political priorities.

The ecosystem pays.

Banks, platforms and other companies absorb costs because the infrastructure enables profitable businesses elsewhere.

That can preserve zero-cost usage, but only for as long as those indirect economics remain attractive.

Usage pays.

Commercial participants contribute directly based on transactions or consumption.

That creates a clearer economic model and can produce sustainable investment, but it also introduces price signals that may reduce adoption or change behaviour.

UPI’s new MDR framework does not fully choose one of these approaches.

It blends them.

Consumers remain protected.

Small merchants continue receiving preferential treatment.

Government support has not disappeared entirely.

But larger commercial users are beginning to contribute directly to the network economics.

That hybrid structure may ultimately prove more sustainable.

It will also be closely watched.

The real test starts after success

The first decade of UPI was a technology story.

Could India create a real-time payments system that was interoperable, easy to use and capable of operating at extraordinary scale?

The answer is now clear.

Yes.

The second decade is becoming an economics and governance story.

How should the system be funded?

How much should businesses pay?

How should revenues be divided between banks, payment apps and infrastructure providers?

How do you protect small merchants?

How do you keep incentives aligned?

And how do you ensure that monetisation strengthens the infrastructure rather than gradually making it less open?

Those questions are harder because there is no purely technical answer.

The government’s current position is that the MDR is not a tax and that the revenue will be distributed within the payment ecosystem to support UPI’s operation and expansion.

That makes the next few years an important experiment.

Not in whether people want UPI.

That has already been settled.

The experiment is whether India can attach sustainable economics to one of its most successful pieces of digital public infrastructure without damaging the behaviour that made it successful in the first place.

Because eventually every transformative technology encounters the same problem.

Building it is one challenge.

Getting everyone to use it is another.

But once everyone depends on it, the question becomes much harder:

Who pays for the technology nobody wants to live without?