By Rohit Arora, CEO & Co-Founder, Biz2X and Biz2Credit
India’s lending market is walking a tightrope right now. On one side, there is genuine growth — domestic demand is holding up, credit penetration still has room to run, and businesses need capital. On the other hand, there is a landscape increasingly shaped by geopolitical uncertainty, energy-price swings, currency movements, and tighter liquidity conditions.
For NBFCs and digital lenders, this tension cuts deeper than it does for traditional banks. Banks can lean on stable deposit bases. Most NBFCs are pulling funds from bank borrowings, market debt, commercial paper, and external funding lines. When crude spikes or the rupee weakens or liquidity dries up even a little, the impact does not trickle down. It hits fast, showing up in the cost of funds, which then ripples into margins, borrower selection, and how portfolios get built.
The Economic Survey 2025-26 pegs India’s real GDP growth for FY27 somewhere between 6.8% and 7.2%, with domestic demand still doing the heavy lifting. But the same report does not shy away from flagging what could go wrong, like trade disruptions, volatile capital flows, and sudden policy shifts abroad. The opportunity is absolutely there, but the margin for error has got thinner.
Funding costs bite harder
Bank lending to NBFCs jumped 26% in FY26, reaching ₹20.7 lakh crore, significantly faster than the roughly 16% growth in overall bank credit, according to data from the Reserve Bank of India (RBI).
It is a double message: NBFC credit is in demand, yes, but the sector lives and dies by its access to bank funding.
When funding gets expensive, lenders get stuck between bad options. Pass the entire increase to borrowers, and you risk killing demand or worsening repayment behaviour. Absorb it yourself, and margins take a beating. Slow down disbursements to protect asset quality and growth stalls. Pricing alone will not solve this. What is needed is better segmentation, faster risk assessment, and more granular tracking of how portfolios are actually performing.
This is where digital infrastructure stops being just a nice-to-have front end. Lenders need systems that can pull together bank statements, financial records, GST flows, bureau signals, and repayment patterns and read them together in real time. Increasingly, that synthesis is being handled by AI, not as a back-office experiment, but as the core underwriting engine. Machine learning models trained on layered borrower data can detect deteriorating repayment capacity well before it surfaces in a bureau report, giving lenders a window to act rather than react. The institutions getting this right are not just moving faster but also with a precision that human review at scale simply cannot match.
Unsecured lending gets messier
In unsecured and small-ticket retail credit, the picture has gotten more complicated. Consumer appetite hasn’t vanished, but risk distribution across borrower segments has become uneven. Consumer durables, credit cards, and microfinance have shown enough stress to make lenders pull back and refocus on asset quality over pure growth.
Digital lending should not be judged purely on how fast money moves. The real measure is whether it can make credit both faster and safer, like early-warning systems that actually work, cash-flow tracking that updates continuously, document analysis that does not need manual oversight, and collections intelligence that goes beyond automated reminders.
AI is what makes that combination possible at any meaningful scale. A well-trained model does not just flag a missed EMI, but it reads the pattern leading up to it: a dip in account inflows, a spike in minimum-due payments, and a slowdown in GST filings. That kind of longitudinal signal processing is exactly what separates a system that manages risk from one that merely records it. The question is not only if you can onboard a customer quickly but also whether your platform can spot trouble early, tell the difference between a resilient borrower and a fragile one, and help you act before risk calcifies into loss.
Secured lending and MSME finance offer the more resilient opportunity. According to a recent TransUnion CIBIL’s MSME Pulse report, India’s commercial lending portfolio stood at ₹67.5 lakh crore as of December 2025, reporting a 16% year-on-year growth. Asset quality improved too, with delinquencies dropping to a five-year low of 1.87%.
MSMEs are not underserved for lack of demand. The problem is fragmented credit information. A small business might have strong cash flows, but that data is scattered across bank accounts, GST filings, invoices, and bureau records. Traditional underwriting often cannot pull it together fast enough to matter.
This is precisely where AI earns its place in the credit stack. Digital platforms deploying AI-driven analytics can stitch together those fragmented signals—triangulating GST return frequency, invoice settlement cycles, and banking behaviour—to build a dynamic credit picture that no static scorecard can approximate. Whether it is vehicle finance, gold loans, loans against property, or supply-chain finance, the edge will go to lenders who can marry collateral assessment with live cash-flow understanding. AI does not just speed that process up; it makes connections a human analyst, working through hundreds of files a week, structurally cannot.
Regulation isn’t optional at scale
The RBI’s stance on NBFCs has gotten clearer: growth is encouraged, but it has to come with governance, capital discipline, and responsible practices. Some smaller non-deposit-taking entities that do not tap public funds or face customers directly might get calibrated relief on registration requirements. Mid-sized and large lenders will face continued scrutiny, especially where public money and customer protection are involved.
For digital lenders, this should not feel like a pure compliance burden. It is structural and not cosmetic. This applies equally to AI-driven decision systems. Explainability is not optional when regulators and borrowers alike need to understand why credit was extended or withheld. Models need to be auditable, decisions need to be traceable, and portfolio-level risk controls need to be built into the platform from the start—not bolted on later through workarounds that create friction and operational weight. The lenders who treat responsible AI as a governance principle rather than a legal checklist will be the ones regulators trust with scale.
India’s credit story is not going anywhere. Consumption, infrastructure, MSME growth, formalisation and other fundamentals are intact. What is changing is who wins.
The next phase belongs to lenders who can tell good risk from bad risk before the loan goes out the door, not after. AI is the infrastructure that makes that possible, not in the abstract, futuristic sense, but in the very practical sense of building systems that price more accurately, flag stress earlier, protect margins under pressure, and serve borrowers responsibly even when conditions get rough.
What separates the survivors from the casualties will not be the ones that lend fastest but the ones that lend the smartest.