The Payments Rail Is Becoming a Credit Rail
India's Unified Payments Interface was built to move money instantly. It is now being repurposed to move credit instantly as well. NPCI's credit-line-on-UPI architecture allows a sanctioned credit line to be linked to a UPI handle, so a borrower can pay a merchant directly from a pre-approved limit rather than from a deposit balance. That sounds incremental. In practice, it is structural: it turns UPI from a settlement utility into a distribution channel for working capital, consumer credit and merchant finance.
The macro significance is hard to overstate. India's banking system is roughly ₹200 lakh crore in size, but the formal credit market still struggles to serve micro and small enterprises that lack collateral, audited statements or long banking histories. UPI's scale — now processing billions of monthly transactions — gives lenders a behavioral data layer that traditional branch banking never had. Every payment, refund, merchant category and cash-flow pattern becomes a signal. The promise is lower acquisition cost and faster underwriting. The danger is that lenders may confuse transaction frequency with repayment capacity.
NPCI's model matters because it is not a private wallet experiment. It is public digital infrastructure, governed by a quasi-utility architecture that can standardize access across banks, fintechs and merchants. That lowers integration costs and reduces dependence on proprietary credit apps. But it also creates a new policy question: if the rail is public, who bears the reputational and systemic risk when credit losses rise? The answer is not NPCI. It is the originating bank, the co-lender, and ultimately the regulator watching unsecured exposure expand.
Account Aggregators and the New Underwriting Stack
The Account Aggregator ecosystem is the second pillar of this transformation. AA is designed to let customers share financial data — bank statements, GST-linked cash flows, mutual fund holdings, insurance records and other consented information — across institutions through a standardized, revocable consent framework. In theory, this solves the classic MSME problem: lenders can see enough of a borrower's real cash flow to price risk without demanding property collateral or years of audited accounts.
For lenders, the attraction is obvious. A small trader with irregular income may look unbankable on paper but highly financeable when payment inflows, invoice cycles and merchant settlement data are visible. Fintechs have built underwriting engines around this premise, using AA data to score borrowers in minutes rather than days. Pine Labs and Razorpay, among others, have positioned themselves as orchestration layers: they sit close to merchant transactions, capture operating data and route eligible borrowers to bank partners. The bank supplies the balance sheet; the fintech supplies the distribution and analytics.
Yet the AA model is only as good as the data it receives and the assumptions embedded in the score. Cash-flow data can be noisy, seasonal and vulnerable to manipulation. A merchant may show healthy UPI inflows while masking supplier arrears, informal borrowings or concentration risk in a single customer. That is why RBI's prudential stance remains conservative. Unsecured loans attract higher risk weights because the absence of collateral makes loss-given-default materially worse. In effect, the regulator is telling lenders that data-rich underwriting is not the same as risk-free lending.
This is where the tension becomes visible. Fintechs argue that traditional risk weights are blunt instruments that penalize innovation and exclude millions of viable borrowers. Banks counter that algorithmic optimism can create a false sense of precision. The truth is somewhere in between: AA can improve selection, but it cannot eliminate macro shocks, fraud, or borrower overleveraging. In a downturn, the same digital rails that accelerate disbursement can accelerate stress.
SBI, HDFC and ICICI: Balance Sheets Meet Distribution Networks
The most important institutional shift is not technological but organizational. Public sector banks such as State Bank of India and private lenders like HDFC Bank and ICICI Bank are increasingly using fintech partnerships to reach MSMEs that would be too expensive to serve through branches alone. The logic is simple: banks retain the regulated balance sheet, while fintechs handle customer acquisition, data capture and sometimes collections. Co-lending structures then split exposure between the bank and non-bank partner according to pre-agreed ratios.
For SBI, the appeal lies in scale and reach. As the country's largest lender, it can absorb lower-yield, high-volume small-ticket credit if origination costs fall enough. For HDFC Bank and ICICI Bank, the attraction is more tactical: preserve asset quality while expanding into merchant and working-capital segments where digital data can sharpen underwriting. The partnerships with Pine Labs and Razorpay are especially telling because both firms are embedded in merchant payment flows. They see when a shop is busy, when it is idle, and how quickly money turns over. That is a powerful proxy for short-term liquidity.
But co-lending also creates governance complexity. Who owns the customer? Who sets the credit policy? Who handles collections when a borrower misses a payment? In a branch-led model, these questions are internal. In a fintech-led model, they are contractual and often fragmented. That fragmentation can become a risk multiplier if incentives are misaligned. A fintech paid on disbursement volume may favor growth; a bank judged on NPA ratios may prefer restraint. The result can be a race between origination speed and portfolio discipline.
There is also a capital efficiency angle. By using digital rails and co-lending, banks can deploy capital more selectively and potentially improve return on assets. But RBI's unsecured loan risk weights reduce the temptation to chase growth blindly. Higher capital charges force lenders to hold more equity against risky exposures, which should, in theory, slow exuberance. In practice, banks may respond by shortening tenors, lowering ticket sizes or tightening eligibility thresholds — all of which can preserve asset quality but limit the very inclusion the system is trying to expand.
The NPA Test: Can Digital Credit Scale Without Breaking?
The central question is whether India can scale unsecured digital credit without repeating the cycle of exuberance, stress and write-offs that has historically plagued retail and MSME lending. The answer depends on whether lenders can build NPA models that are dynamic enough to reflect real-time cash flows but conservative enough to survive a downturn. That is harder than it sounds. Default behavior in small-business lending is often non-linear: borrowers may remain current for months and then suddenly slip when inventory cycles break, input costs rise or a major buyer delays payment.
RBI's framework is designed to force discipline. Higher risk weights on unsecured loans make such assets more expensive to hold, while NPA recognition rules prevent lenders from hiding stress through evergreening. This is particularly important in a digital ecosystem where disbursement can be nearly instantaneous. Speed is not the problem; speed without post-disbursement monitoring is. The best lenders are therefore moving beyond approval scores toward continuous monitoring: transaction velocity, settlement gaps, bounce rates, GST mismatches and merchant concentration are being tracked as early-warning indicators.
Still, there is a structural trade-off that policymakers cannot ignore. If regulation is too tight, the digital credit stack may remain a pilot for urban merchants and formalized MSMEs, never reaching the informal economy that needs it most. If regulation is too loose, lenders may flood the market with small unsecured loans that look diversified but are highly correlated in a downturn. The system's resilience will depend on whether public infrastructure can support private discipline. That means interoperable data, auditable consent, transparent pricing and clear accountability when loans sour.
The deeper story is that India is not merely digitizing banking; it is redesigning the architecture of trust. UPI solved the problem of payments at scale. Account Aggregator is solving the problem of consented data portability. Credit-line-on-UPI is now trying to solve the problem of last-mile lending. If it works, the country could unlock a new phase of MSME formalization and productivity growth. If it fails, the losses will not be confined to fintechs. They will land on bank balance sheets, capital ratios and the credibility of India's digital public infrastructure itself.