Why renting payment rails is a slow leak, and what neo-banks should build before the margins disappear
Most neo-banks are built on rented payment infrastructure. Here is why that model struggles to survive, how payment stack ownership changes unit economics, and what founders should prioritise in India’s crowded digital banking market.
Generative Ai
A few years ago, launching a neo-bank felt like assembling furniture from a catalogue. You picked a banking-as-a-service partner, plugged in a payments API, designed a clean app, and went live. The screens looked modern, the onboarding was fast, and the cashback offers did their job. Users came in quickly.
Then the bills started arriving. Every transaction routed through someone else’s rails carried a cost. Every new feature depended on a partner’s roadmap. And every customer who loved your app still belonged, quietly, to the company that owned the pipes underneath it.
That is the problem I keep coming back to. Most neo-banks are beautiful storefronts built on borrowed plumbing, and that plumbing decides more of their future than their design ever will.
Renting rails is cheap until it isn’t
Early on, renting is the right call. You do not have months to spare, and building a payment stack from scratch is expensive and slow. Partners let you test ideas, reach customers, and prove demand.
The trouble starts when scale arrives. Per-transaction fees that looked small on a spreadsheet add up fast. Revenue share agreements quietly take a slice of every UPI-linked interaction, every card swipe, and every bill payment. A company growing at 40% a year can find that its gross margin is shrinking at the same pace.
Renting also means your customer experience is only as good as your partner’s uptime. When a settlement delay happens, your users blame your app, not the vendor you never mentioned in the onboarding flow.
Ownership is more than a licence
When founders hear “own your payment stack,” many imagine a giant engineering project. That is not quite right. Ownership is a set of decisions about which layers you control.
At the most basic level, it means understanding routing: which rail a payment takes, why, and what it costs. Beyond that, it means owning reconciliation, so your ledger matches what actually moved. It means controlling retries, failure handling, and the logic that decides whether a transaction is worth approving.
Licensing sits in the background of all this. Under RBI’s payment aggregator framework, the entities handling merchant payments face clear regulatory expectations around fund flow, settlement, and safeguards. Whether you hold the licence yourself, partner with a licensed entity, or structure a hybrid model, the question is the same: do you know exactly where the money sits at every moment?
I think many founders underestimate how much value lives in those unglamorous layers.
Where the real margin hides
Here is what I wish more founders looked at before raising their next round. Payment stack ownership changes three things at once.
Unit economics. When you control routing and negotiate directly with networks and processors, your cost per transaction can fall meaningfully. That margin compounds, especially for products built around high-frequency, low-value actions like UPI transfers and merchant payments.
Data. Owning the payment flow means owning the signal. You see which transactions fail, why, and at what hour. That data improves fraud models, credit decisions, and product decisions in ways a partner dashboard rarely allows.
Control over the customer relationship. If a payment fails, the customer should hear from you, with a clear next step. Owning the stack lets you design that moment instead of apologising for someone else’s error message.
What to prioritise if you are already live
Maybe you have a few hundred thousand users and a partner contract that renews next year. You cannot rebuild everything overnight, and you should not try.
Start by mapping your stack honestly. List every layer that touches a rupee of your customer’s money, and mark who controls it. You will likely find two or three layers where you have no visibility at all. Those are your first targets.
Next, negotiate for data access even if you cannot yet negotiate for ownership. Real-time transaction logs, failure codes, and settlement timelines are worth fighting for. Without them, you are flying blind.
Then build your reconciliation and ledger layer in-house. It is the least glamorous part of the stack and one of the most valuable. Once you can prove, to the paisa, where every rupee went, your options expand considerably.
Finally, plan the migration path before you need it. Partnerships end, regulations shift, and vendors get acquired. A neo-bank that can switch a rail in weeks has leverage. One that cannot is at the mercy of whoever holds the keys.
The banks that survive will look a little boring
I suspect the neo-banks still standing in five years will not be the flashiest ones. They will be the ones that quietly built strong infrastructure, kept their unit economics healthy, and treated compliance as a competitive advantage rather than a cost centre.
Design still matters. A good app wins the first download. But the second and third years belong to the company that controls the money, understands the data, and can keep promises when a partner cannot.
If you are building a neo-bank today, ask yourself a simple question: if your main payment partner disappeared tomorrow, how long would your product survive? The honest answer will tell you where to start.
Why Most Neo-Banks Will Fail Without Owning Their Payment Stack was originally published in Coinmonks on Medium, where people are continuing the conversation by highlighting and responding to this story.
