Digital Public Infrastructure: India's Gift to the World — or a Template for Control?

Digital public infrastructure is one of those phrases that sounds dull until you realise how much of daily life it now quietly runs. In India, it means the shared national rails for identity, payments and data — the systems that let a street vendor accept a payment with a QR code, a migrant worker prove who she is without a passport, or a pensioner receive money directly in her account. Think of it the way you think of roads, or the electricity grid: a common foundation that both the state and private companies build on top of. Over the last decade India has built the most ambitious version of this anyone has attempted, and the world has noticed. But the same rails that can include a billion people can also track and steer them, and that tension is what this piece is about.

What DPI actually is

The label “digital public infrastructure” is newer than the thing it describes. At its core, DPI is a set of foundational, reusable digital systems — owned or governed in the public interest — that many different services can plug into. In India these are often grouped under the informal name “India Stack,” and they come in three broad layers.

The first is identity: Aadhaar, the biometric ID number now held by well over a billion residents. The second is payments: the Unified Payments Interface, or UPI, which lets money move instantly between bank accounts through a phone, free at the point of use, across hundreds of banks and apps. The third, and least visible to ordinary users, is data exchange — consent-based systems that let you move your own records (bank statements, tax filings, health data) from one institution to another when you choose to.

The unifying idea is that none of these is a product. They are rails. A bank, a lending startup, a hospital, a government welfare department — each builds its own service on top of the same shared foundation, rather than every player reinventing identity or payments from scratch. That is what makes it infrastructure rather than an app, and what makes it public is that it is meant to be an open, neutral utility rather than a private walled garden.

Why India’s version is genuinely admired

It would be easy, given where this essay is heading, to skip past the achievement. That would be dishonest, because the achievement is real and it is large.

Start with scale and speed. UPI now processes billions of transactions a month — a volume that dwarfs most card networks anywhere — and it went from a standing start in 2016 to near-ubiquity in Indian cities and, increasingly, villages. I have watched tea sellers and auto-rickshaw drivers who would never have qualified for a card machine accept digital payments with nothing but a printed QR code and a cheap phone. That is not a marketing story; it is a genuine shift in who gets to participate in the formal economy.

Then there is the cost of inclusion. Bringing hundreds of millions of people into a banking and payments system usually takes decades and enormous fees. India did a version of it in years, and largely without charging users per transaction. For someone earning a few dollars a day, the difference between “free” and “two percent” is the difference between using the system and staying outside it. Direct transfer of welfare payments into bank accounts, whatever its flaws, has in many cases cut out layers of middlemen who used to skim from the poor.

And there is export. India has actively promoted this model abroad, offering the underlying approach — and sometimes the actual software — to other countries, and using its international platforms to pitch DPI as a development tool for the Global South. Several nations have signed on to explore or adopt versions of it. There is something legitimately admirable in a lower-income country building world-class public technology and then offering it as a shared good rather than hoarding it. When people say India’s DPI is a gift to the world, this is the part they mean, and they are not wrong to.

The same rail that lets a vendor accept a payment can, in the wrong hands, become a record of everywhere she has been and everyone she has paid.

The double edge

Here is where I have to be careful, because the point is not that any of this is secretly evil. The point is subtler and, I think, more important: the very properties that make DPI so good at including people are the same properties that make it powerful over them.

A rail that everyone uses is, by definition, a rail that sees everything. A payments network that reaches every corner of the economy is also a map of the economy in real time. An identity layer that lets a billion people prove who they are is also a way to link a person’s activities across dozens of otherwise separate systems — welfare, telecom, banking, travel — into a single joined-up profile. None of that is a bug. It is simply what happens when you build one shared foundation instead of many separate ones. The convenience and the surveillance potential come from the same design.

This is a pattern I keep returning to in my own writing about technology, and it is worth naming plainly: tools that concentrate capability tend, over time, to get captured by whoever holds the most power near them. I’ve written separately about how technology gets captured — the recurring cycle in which a system built to empower people slowly tilts toward controlling them, not because anyone flips a switch, but because the incentives all point one way. DPI is a textbook case in the making. Ask the three questions I always ask of any powerful tool: who takes the value, who pays the cost, and who can fight back? With DPI the answers are not fixed. They depend entirely on the rules wrapped around the rails.

The risks are not hypothetical, and I go into the specifics elsewhere. On the identity and payments side, the debates over exclusion (people wrongly denied benefits when biometric authentication fails), over profiling, and over consent are real and ongoing — I unpack them in a companion piece on Aadhaar and UPI. On the legal side, whether any of this is actually safe comes down to data protection law, and India’s framework — examined in my piece on the DPDP Act — has genuine gaps, especially around how freely the state itself can access citizens’ data. And on the geopolitical side, the export of these systems raises the question I explore under data colonialism: who ultimately controls the rails a country runs on, and what dependency comes bundled with the gift.

A model that travels — to democracies and autocracies alike

The export story deserves its own hard look, because it is where the “gift or template” question gets sharpest.

When India promotes DPI globally, the pitch is inclusion: here is a low-cost way to give your citizens identity, payments and services. That pitch is honest as far as it goes. But infrastructure is neutral about who installs it. The same architecture that a healthy democracy uses to deliver pensions efficiently can, in a state with weaker checks, become the most comprehensive population-monitoring system that country has ever had. A national identity layer plus a total-visibility payments network plus a data-exchange system is, viewed coldly, also the toolkit for knowing where everyone is, what they buy, and whom they associate with.

I want to be fair here: this is not a uniquely Indian danger, and it is not a reason to abandon the technology. Every powerful general-purpose system has this property. Roads move ambulances and tanks. But it does mean that promoting DPI as a ready-made template, to be adopted by any government regardless of its institutions, is not automatically a humanitarian act. A tool that includes can also control, and when you hand the blueprint to a government that lacks independent courts, a free press, or a real data-protection regulator, you may be handing over the second capability along with the first. The technology arrives; the safeguards do not come in the box.

The infrastructure is neutral. Everything that decides whether it is a public good or a control system sits in the governance wrapped around it.

The crux: governance, not gadgetry

If there is one thing I want a reader to take away, it is this. The debate over whether DPI is good or bad is mostly the wrong debate. The rails themselves are neither. What determines which way they tip is everything around them — and that part is not technical at all.

Concretely, the safety of a DPI system depends on a handful of unglamorous things:

• Hard limits on data linkage. Can a person’s records across systems be joined into a single profile at will, or is that deliberately made difficult? The design choice matters as much as the law.
• Genuine consent and purpose limitation. Is data used only for what it was collected for, with the person actually in control — or is “consent” a checkbox that unlocks everything?
• Independent oversight. Is there a regulator with real teeth and real independence from the government it is supposed to check, or one that answers to the same executive that runs the systems?
• Access controls on the state itself. The hardest question. Democracies restrain their own governments’ access to citizen data through courts and law. A DPI without that restraint is an inclusion engine and a surveillance engine wearing the same clothes.
• Recourse when it fails. When the system wrongly locks someone out — a failed fingerprint, a mismatched record — is there a fast, human, accountable way to fix it, or does the citizen simply lose access to food, money or identity?

Notice that not one of these is about the code. You cannot tell, from the architecture diagram alone, whether a given DPI deployment is a public good or a cage. The exact same stack can be either. That is precisely why it is so important not to let the impressive engineering distract from the boring, essential questions of oversight, limits and independent institutions. The gadget is the easy part. The governance is the hard part, and it is the whole game.

What makes it a public good rather than a control system

So where does that leave the honest verdict? Not with a shrug, and not with a scare.

India has built something remarkable, and much of the admiration is earned. For the person who was previously invisible to the formal economy — no bank, no ID, no way to receive a payment — DPI has been closer to liberation than to control. It would be a smug and privileged mistake to lecture that person about surveillance while ignoring the exclusion the system lifted them out of. That is a real benefit accruing to real people who had the least, and it should be defended.

But a benefit is not a guarantee. What keeps DPI a public good is a set of choices that have to be made and re-made continuously: to keep the rails open rather than let them be quietly enclosed by whoever runs them; to build in the limits that make abuse hard rather than merely illegal; to fund and protect the independent institutions that can say no to the government; and to treat citizens as the owners of the system rather than its data sources. Where those choices are made, DPI is one of the most promising things happening in technology anywhere. Where they are skipped, the identical technology becomes the most efficient instrument of control a state has ever possessed.

That is the real answer to the question in the title. India’s digital public infrastructure is not, in itself, either a gift or a template for control. It is a set of extraordinarily powerful rails, and it will become whichever of those two things the governance around it allows. The engineering has largely been solved. Whether it stays a public good is a political question, an institutional question, a question about who is allowed to fight back — and it is being answered, quietly, right now. It deserves far more of our attention than the technology itself.