Nikhil KamathIf UPI Is Free, Who Pays for It? | PhonePe, Razorpay & NPCI | WTF is Payments?
CHAPTERS
- 0:06 – 2:33
Money gets “lighter”: from grain to gold to paper to digital sound
Nikhil frames the episode as a long arc in which money keeps becoming less physical—culminating, in India’s case, in digital payments that feel like “sound.” He sets up the central curiosity about what comes next, and introduces the three guests as builders of today’s digital rails.
- •Nikhil traces money’s progression: physical commodities → gold → paper promises → plastic → digital
- •He floats the idea that the next shift could relate to energy, while admitting uncertainty
- •He positions the guests as people who helped turn money into a digital, near-instant experience
- •The discussion is set up to cover personal journeys, tradeoffs, and what’s next in payments
- 2:33 – 9:45
Sohini Rajola’s path to NPCI and what NPCI is (and isn’t)
Sohini shares her background across banking and cross-border money transfer before joining NPCI to lead growth. Nikhil presses on NPCI’s structure, funding, and whether it functions like a regulator; Sohini explains NPCI is a not-for-profit with governance via committees and close work with government and RBI—but it is not the regulator.
- •Sohini describes her career: Axis Bank (payments/digital initiatives) → Western Union → NPCI growth leadership
- •NPCI is described as a Section 8 not-for-profit set up by RBI and IBA, initially funded by 10 banks
- •She explains surplus is reinvested rather than returned as dividends to shareholders
- •Sohini emphasizes RBI is the regulator; NPCI operates through steering committees with multiple constituents
- 9:45 – 14:13
Banks vs fintechs: regulation, moats, and why payment banks struggled
The conversation shifts to whether banks are becoming irrelevant and how regulation creates uneven playing fields. Sohini and Harshil argue banks face heavier constraints because they take deposits, and discuss why payment banks and small finance banks evolved differently.
- •Nikhil argues banks can be over-regulated relative to fintechs, which can create a fintech “moat”
- •Sohini responds that deposit-taking necessitates stricter regulation; banks still have moats like access to deposits/cost of capital
- •They discuss payment banks’ limitations (e.g., constraints around deposits/lending) as a reason the model struggled
- •They contrast this with small finance banks, noting lending ability as a key differentiator
- 14:13 – 20:42
Harshil Mathur’s origin story: early building, fundraising friction, and starting a payment gateway
Harshil recounts his early interest in computers, creating a campus software group, and learning how hard it is to fund projects through government systems. He describes the naive early attempt to build a payment gateway and how banks initially dismissed the idea.
- •Harshil shares background: Jaipur upbringing, IIT Roorkee, and strong interest in coding despite studying mechanical
- •He describes building a campus software group and how difficult government-style fundraising and spending can be
- •He recounts walking into a bank branch asking how to build a payment gateway and being turned away
- •The segment highlights early barriers for first-time founders entering finance infrastructure
- 20:42 – 24:05
Sameer Nigam’s path: consumer internet, entrepreneurship, and Flipkart years
Sameer describes moving across countries, working in a consumer internet company, then pursuing an MBA and founding a digital music venture that was later sold to Flipkart. His Flipkart experience becomes a bridge into understanding payments at scale.
- •Sameer talks about early life moves (Germany, Delhi, Bombay) and later time in the US
- •He describes working in a startup and his interest in consumer internet products
- •He founded a digital music venture, learned hard lessons about collections, and sold the company to Flipkart
- •He sets up how operating inside Flipkart exposed payments and checkout challenges at scale
- 24:05 – 30:28
Why young Indians want to leave: trust, governance, and “fear” as a growth constraint
Nikhil asks why the earlier wave of returnees came back to India, but many young people now want to leave. Harshil argues there’s societal frustration with slow systems and excessive interference, and the group debates trust, enforcement, and the cost of a weak legal-judicial process.
- •Sameer points to weaker “macros” and shifting sentiment among younger people
- •Harshil argues many young people want progress and are frustrated by constraints and interference
- •The discussion links low trust to weak enforcement and slow judicial outcomes; a speaker notes “the process is the punishment”
- •Nikhil and others emphasize reducing fear, avoiding overregulation, and enforcing laws when rules are broken
- 30:28 – 43:20
UPI’s early builders and why regulation didn’t stop the first sprint
Harshil names a small group he says was central to building UPI and highlights key design choices like mobile-first interoperability. They discuss why the regulator and banks aligned early, partly from a desire to avoid dependence on global card networks and because digital payments were small at the time.
- •Harshil says UPI was built by a relatively small group and lists several early contributors by name
- •He highlights tech principles: mobile-first design, interoperability, and API-driven infrastructure
- •He argues early momentum benefited from regulatory buy-in around building local rails
- •Sohini recalls banks being split—some excited, others skeptical about liability, risk, and settlement in a multi-party model
- 43:20 – 52:03
Flipkart’s Big Billion Day payment failures and the lesson that birthed a payments company mindset
Sameer describes how Flipkart’s sale traffic overwhelmed bank systems, causing massive payment failures and even taking down core banking systems. The team learned edge cases like “failed” payments becoming successful later, forcing emergency fulfillment and revealing how fragile online payments were at the time.
- •Flipkart asked banks to scale capacity dramatically ahead of the Big Billion Day sale, but systems still buckled
- •He describes millions of payment attempts with the majority failing, and bank systems going down under load
- •He notes confusing outcomes like failed responses later turning into successful debits, creating reconciliation chaos
- •The episode frames this period as formative for understanding India’s payments infrastructure gaps
- 52:03 – 59:06
How a UPI transfer actually works: actors, routing, and the 10-API-call reality
The group builds a step-by-step model of a UPI P2P transfer, including the app, sponsor/PSP bank, NPCI, sender and receiver banks, and identity resolution. They explain why sponsor banks exist (risk and regulatory exposure), and how VPAs/phone numbers are resolved before authorization and debit/credit occur.
- •They map the parties involved: app (TPAP/PPAP), sponsor/PSP bank, NPCI, issuer bank, beneficiary bank, and identity resolution
- •Nikhil questions why apps can’t connect directly to NPCI; the group explains it’s not purely technical but about regulated exposure and risk
- •They explain VPA/phone-number “namespace resolution” before the payment request is executed
- •Sohini states a UPI transaction can involve roughly 10 API calls executed in milliseconds
- 59:06 – 1:06:02
If UPI is free, who pays? Costs, fraud controls, and where money moves (and doesn’t)
Nikhil pushes on the true economics of “free” UPI: who bears the cost per transaction, and why. The guests discuss ecosystem costs (infrastructure, SMS, fraud/risk models), inter-bank fee flows, and the absence of float for intermediaries in real-time bank-to-bank payments.
- •They estimate an ecosystem-level cost for a P2P transaction and discuss how it differs by context
- •Harshil describes fraud and risk checks done at the app/edge level; Sohini adds ecosystem-wide fraud pattern detection at NPCI
- •They explain issuing banks pay the PSP bank a per-transaction fee (slab-based, capped), justified as a digital service cost vs branch/ATM costs
- •They contrast UPI with wallet-based systems, noting intermediaries don’t monetize float; money sits in bank accounts, with settlement handled separately from real-time posting
- 1:06:02 – 1:41:27
Instant posting vs bank settlement: netting cycles and why the architecture scales
The conversation distinguishes between real-time debit/credit entries and periodic interbank settlement. They describe frequent settlement cycles, netting, and why this design allows massive transaction volumes without needing one-to-one money movement each time.
- •Harshil explains real-time payments are accounting entries, while settlement happens out-of-band
- •Sohini states settlements occur multiple times a day, with netting across banks
- •They note peak transaction bursts can net down into far fewer settlement transfers
- •Nikhil compares the model to market settlement concepts where transfers are netted rather than individually moved
- 1:41:27 – 2:02:36
Cards and merchant payments: pull vs push, OTP flows, and why fees exist
The discussion moves from P2P to merchant payments (P2M), using card payments to explain the “pull” model. They walk through payment gateways, acquiring banks, networks/switches, issuer banks, two-step authentication/authorization, and how fees are split—highlighting that much of the cost funds credit and merchant-side risk.
- •Harshil defines push (P2P) versus pull (merchant-initiated collection) and applies it to cards
- •They outline the chain: merchant → payment gateway → acquiring bank → network switch (e.g., RuPay/NPCI) → issuer bank → OTP/auth → authorization
- •A speaker explains gateway pricing to merchants and that most of the fee flows downstream, largely to issuers due to credit costs and rewards
- •They explain merchant-side fraud/chargeback liability as a reason percentage-based pricing exists for payment acceptance
- 2:02:36 – 2:15:35
Where to build next: B2B workflow, cross-border medical tourism, and the unsolved tipping gap
The guests argue the largest remaining whitespace is B2B payments where the payment itself is only one part of an invoicing, reconciliation, tax, and approval workflow. Sohini describes an idea focused on inbound medical tourism and cross-border payment friction. Harshil calls tipping in the UPI era a surprisingly hard UX and architecture problem.
- •They say B2B payments need tooling for invoices, reconciliation, GST/TDS considerations, and multi-person approval workflows—not just a transfer rail
- •Sohini points to UPI features like multi-signatory as building blocks for “UPI for business”-style products
- •Sohini describes a potential cross-border payments product tailored to inbound medical tourism use cases and risk handling
- •Harshil highlights tipping as unsolved due to how UPI requests work, suggesting the need for a better service-sector mechanism
- 2:15:35 – 2:44:31
Payments data as an economic lens: tier-2+ strength, online spending concentration, and tax sensitivity
Nikhil asks what payments data reveals about India’s consumers and the economy. Harshil argues tier-2 and beyond drives a large share of consumer digital payments, while online discretionary commerce remains concentrated. The discussion also touches on observed correlations between GST changes and consumption behavior, plus the limits of household survey extrapolations.
- •Harshil says tier-2 and beyond contributes a large share of consumer digital payments and that money circulates within and across non-metro regions
- •He claims online e-commerce spend by value is concentrated among a relatively small user base compared to overall population
- •He describes an analysis suggesting consumption responds strongly to GST changes by category, with “control” categories (unchanged GST) showing little change
- •They question household consumption survey extrapolation and note NPCI publishes district/category-level data regularly
- 2:44:31 – 4:00:05
Stablecoins, on/off-ramps, and capital controls: efficiency vs currency control
The episode closes on crypto-related rails as an analogy to real-time settlement—describing stablecoin transfers as ledger movements that bypass correspondent banking. The guests debate whether such rails can coexist with capital controls, centering the argument on regulating entry/exit ramps and preserving the central bank’s ability to manage currency flows.
- •They describe stablecoin transfers as ledger entries that remove multiple correspondent-bank steps
- •The conversation stresses a difference between technology efficiency and policy control over currency movement
- •A speaker argues tight regulation of on-ramps/off-ramps could preserve controls, similar to regulated forex channels
- •They debate whether India can afford full capital account convertibility today and suggest the answer is likely no at present
