A Billion Users, and the Rent Goes to Mountain View

Every argument about Big Tech in India tends to collapse into one of two bad positions within about five minutes. Either foreign companies are looting us and must be thrown out, or India is a roaring digital success story and anyone complaining is a crank who has not seen the download numbers. I do not believe either. The numbers are genuinely extraordinary — by most reported measures India is the largest Android market on earth, with something like 95% of mobile usage running on Android and Google handling upwards of 97% of searches. What interests me is not that these are American companies. It is that the layer they sit on is the layer nobody in India controls, and almost everything else we have built sits on top of it.

What we own, and what we rent

Start with an honest inventory, because the honest version is more useful than the angry one.

India has built real things. UPI is world-class public infrastructure by any standard — reported volumes crossed 24 billion transactions in a single month in August 2026, which is a scale no private payment rail anywhere reached by inviting people in and then charging them rent. Aadhaar, for all the arguments I have with how it is used, is an identity layer other countries study. We have homegrown commerce platforms, content platforms, fintech, a services industry that writes an enormous share of the world's software, and increasingly the researchers and engineers whose names appear on the papers that make the frontier models work.

Now the other column. The operating system on nearly every phone in the country: not ours. The store through which almost every app reaches a user: not ours. The search box that decides which Indian business exists and which does not: not ours. The video platform where an entire generation of Indian creators earns a living, and the feed that decides who among them eats this month: not ours. The cloud most Indian startups run on, and the model weights the next decade of Indian software will be built against: overwhelmingly not ours.

So the sentence I keep coming back to is this. India supplies the users, supplies the data, and now supplies a growing share of the engineers — and owns almost none of the layer that sets the terms. That is not a xenophobic observation. I have no objection to an American company earning money in India, any more than I object to an Indian company earning money in America. The objection is structural: when one layer is a chokepoint, whoever owns the chokepoint sets prices, rules, and rankings for everyone above it, and the people who live with those rules have no vote on them.

The problem is not that the rent is paid to a foreign company. The problem is that it is rent — collected at a layer nobody can route around, on terms set where none of the payers live.

The layer that matters

It helps to be precise about which layer. Indian founders build brilliant applications. That is the top of the stack, and competition there is fierce and healthy. Underneath sits distribution — the OS, the store, the search index, the recommendation feed — and that layer behaves nothing like a market. It behaves like a toll road.

A toll road has a specific economic property: you do not have to be better than the alternatives, you only have to be unavoidable. A developer in Kochi does not choose the Play Store the way she chooses a hosting provider, comparing features and switching when a better one appears. She ships there because that is where the phones are. The absence of a real second option is what converts a service fee into a tax, and it is the same mechanism I have written about in how technology gets captured — a technology that begins as an open platform gradually becomes a private tollbooth, not through any dramatic act, but by accumulating small unavoidabilities until nobody can leave.

This is also why the data question and the infrastructure question are the same question. When behaviour from a billion people is refined into products elsewhere and sold back to us as a subscription, that is the pattern I have called data colonialism; the pipes that do the extracting are exactly these distribution layers. And it is why the emerging fight over foundation models worries me more than the app-store fight does. If the AI monopoly settles the way mobile did — a handful of firms owning the substrate, everyone else renting access through an API — then India will have spent a decade building on someone else's road again, just a newer one.

A cut of every rupee

The app-store commission is where the abstraction becomes a number on an Indian founder's P&L. Google Play's headline service fee has been widely reported at 30% on digital purchases, reduced to 15% for a large share of developers under various small-business and subscription programmes; Apple's structure is broadly similar. Following orders from India's competition regulator, developers serving Indian users have been able to offer an alternative billing system alongside Google's, with the service fee reduced by roughly four percentage points when a user picks the alternative — so 15% becomes about 11%. Google also announced a wider billing-choice programme in 2026 with lower fees, though by its own published timelines the early phases cover the US, UK and the European Economic Area, with markets like India following considerably later. All of these figures move; treat them as reported rather than eternal.

Four points off a commission is a concession, not a correction. The underlying fact survives it: a percentage of the revenue of every Indian subscription app, every Indian game, every Indian edtech and dating and music service is set in a room in California, changed when that room decides to change it, and cannot practically be avoided while distribution is this concentrated. An Indian developer can dislike the rate. She cannot decline it and still have a business.

And the rate is only the visible part. The harder terms are the ones about steering — whether an app may even tell its own users that a cheaper way to pay exists — and about self-preferencing, where the platform's own services play by house rules while everyone else pays admission. A commission you can see and price into your model is survivable. A ranking decision you cannot see, cannot appeal, and cannot predict is not.

The regulator's long, real fight

To India's credit, this has not gone unchallenged. The Competition Commission of India opened its investigation into Play Store billing in 2020 and, in 2022, found that Google had abused its dominant position — mandating its own billing system for in-app purchases while its own services were treated differently. In March 2025 the appellate tribunal, NCLAT, upheld a substantial part of that finding while reducing the penalty to a reported ₹216 crore and declining to endorse some of the more forward-looking, rule-setting directions the CCI had issued. Cross-appeals from Google, the CCI and the developer body ADIF have since been admitted by the Supreme Court, and as I write this the matter is live. I am not going to predict how it ends, and anyone who tells you they know is guessing.

What the long arc of that case already shows is the limit of the tool. Case-by-case antitrust is retrospective: it investigates conduct that happened years ago, litigates for years more, and lands a penalty that is real money to a person and a rounding error to a company of that size. Meanwhile the conduct continues, the market consolidates further, and the remedy arrives into a world that has already moved. That is why the draft Digital Competition Bill — which would designate a small set of very large firms as systemically significant and impose obligations in advance on self-preferencing, data combination, anti-steering and interoperability — matters more than any single penalty. As of 2026 it remains in consultation, with the government commissioning further market study rather than legislating quickly. Caution is defensible. Indefinite caution is a decision too.

A fine is a receipt for harm already done. A rule is a change in what is possible tomorrow. India has been issuing receipts.

What would actually change it

Slogans will not. “Self-reliance” as a mood produces press conferences; ownership and access produce outcomes. Four levers seem to me to do real work:

• Enforceable rules on stores and self-preferencing, set in advance. Not merely a lower commission, but hard obligations: sideloading and alternative stores that actually function, no punishing a developer for telling users about a cheaper payment route, no ranking your own service above a rival's on your own surface, and published criteria for how ranking works. Rules that bind before the harm, with penalties large enough to be a business consideration rather than a line item.
• Open public digital infrastructure at more layers. UPI is the proof of concept and the proof of the principle: build the rails in public, make them open to all participants on equal terms, and let private firms compete on the experience rather than on owning the road. That logic did not stop being true above the payments layer — identity, commerce protocols, and eventually compute and model access are all candidates. Public rails are not nationalisation. They are a refusal to let a private chokepoint form.
• Procurement and data policy that build domestic capability. Government is the largest buyer of technology in the country. What it buys, and on what licensing and data-residency terms, determines whether Indian firms get the revenue and the reference customers to climb from application to infrastructure. Anchor tenancy is how national capability has been built everywhere it has been built.
• Interoperability real enough to make leaving possible. Portability that moves your identity, your graph, your purchase history and your reputation — not a ZIP file of old posts. Switching costs are the moat. Lower the moat and you do not need to break anything; the market does the work that regulators otherwise have to do by hand.

Notice that only one of these is about punishment. The argument for breaking up Big Tech is really an argument about structure — separating the firm that owns the road from the firms that drive on it — and structural separation is one instrument among several. Interoperability and public rails get at the same thing from the other direction: they make the road less worth owning.

Rules, not slogans

India has more leverage here than it usually acts like it has. A market of this size is not optional for any global platform; terms offered here get adopted elsewhere. The constraint has never been leverage. It has been the willingness to write rules about ownership and access and then enforce them against companies large enough to litigate for a decade.

So I would retire the language of self-reliance, which flatters us and commits us to nothing, and replace it with a shorter set of questions. Who owns the layer this depends on? What can they change unilaterally, and who can stop them? If they raised the price tomorrow, what would we do? If the answer to the last one is “pay,” then it is rent, whoever collects it — and the way out is not a slogan about swadeshi technology. It is law, public infrastructure, and the unglamorous work of making the road something no one gets to own.