Bajaj Finance booked 51.95 million new loans in FY26. That is 5.2 crore individual credit decisions. Their AUM crossed ₹5.09 lakh crore — a 22% jump year-on-year. Gross NPA stayed at 1.27%.
Scale and asset quality, moving in the same direction. That is not luck. That is a built-in advantage from how they underwrite.
Mid-tier NBFCs still run 7–14 day credit appraisal cycles for SME loans. Bajaj Finance's consumer lending desk approves in minutes. L&T Finance's Project Helios cut SME underwriting turnaround by 30%. A consumer NBFC in the southern market went from 48-hour to 6-hour approval for standard cases.
The borrower who waits three weeks for your credit decision has already taken a loan from someone else.
Speed is not just a customer experience issue. Every day of delay is a loan you didn't give. Multiply that across a portfolio and the revenue delta becomes visible in your P&L.
India has somewhere between 16 crore and 19 crore thin-file borrowers — people with no CIBIL score or a score too sparse to underwrite with traditional methods. Most NBFCs decline them by default.
But these borrowers have UPI transaction history. They have NACH mandate performance records. They file GST returns. They have bank statements accessible via Account Aggregator.
The data signals are there. The underwriting models that use them are not standard issue at most NBFCs. Building one internally takes 12–18 months if you have the team. Most lending operations don't.
This is why 90% of personal loans below ₹1 lakh are now originated by NBFC fintechs — not because traditional NBFCs lack capital, but because the underwriting infrastructure favours speed and data.
Bajaj Finance grew AUM 22% while gross NPA moved from 1.18% to 1.27% — a 9 basis point deterioration on a portfolio that grew by ₹92,000 crore. That ratio is nearly impossible without strong underwriting discipline at origination.
The Nomura finding is useful context: NBFCs as a sector are growing at 17% per year versus 12% for banks. The ones outpacing that figure are the ones with better credit decisioning infrastructure, not necessarily more capital or better branch networks.
RBI's 2026 amendment allowing NBFCs to factor Default Loss Guarantees into ECL provisioning when working with fintech partners adds another lever. NBFC-fintech co-lending is getting structurally easier. The provisioning relief is real. But you still need the underwriting to be right at origination — otherwise DLG just delays the loss rather than preventing it.
The CXOs at most mid-size NBFCs understand the data gap. The problem is the credit head. Not because credit heads resist technology — most don't. Because the existing credit process is the credit head.
They built the scorecard. They know which bureau flags to watch for. They've seen the frauds that slipped through. Their judgment is encoded in Word documents and Excel models.
The question is not whether AI underwriting is better in theory. It is whether the specific signals it uses align with what your credit head has spent 15 years learning to identify. That alignment work — translating institutional knowledge into structured inputs — is where most NBFC tech projects fail.
The NBFCs that are actually closing the gap with Bajaj Finance are not the ones that replaced their credit heads. They are the ones that built tools that make their credit heads faster.
What is AI credit underwriting in NBFCs? AI credit underwriting refers to using machine learning models and alternative data sources — UPI transaction patterns, GST filing history, Account Aggregator bank statements — to assess a borrower's creditworthiness faster and with greater accuracy than traditional bureau-based scoring. Indian NBFCs like Bajaj Finance and L&T Finance have deployed these systems at scale.
How did Bajaj Finance achieve 22% AUM growth while keeping NPA below 1.3% in FY26? Bajaj Finance booked 51.95 million new loans in FY26 while consolidated AUM crossed ₹5.09 lakh crore. Their NPA at 1.27% reflects strong origination-level underwriting — aided by AI-powered credit decisioning, bureau integrations, and a large proprietary behavioural dataset built over years of lending.
What is the Default Loss Guarantee (DLG) rule change for NBFCs in 2026? In February 2026, RBI amended IRACP Directions to allow NBFCs to factor DLG provided by fintech partners into ECL provisioning across all stages of loan default. This reverses May 2025 restrictions that had limited NBFC-fintech co-lending partnerships. DLG coverage is capped at 5% of loan portfolio, backed by cash collateral or bank guarantee.
Why do mid-tier NBFCs struggle to match large NBFC underwriting speed? Mid-tier NBFCs typically run manual CAM processes that take 7–14 days for SME credit. The gap is not capital — it is underwriting infrastructure. Large NBFCs have invested in alternative data pipelines, automated bureau pulls, and AI scoring models. Mid-tier NBFCs are often still operating on Excel-based credit templates reviewed by committee.
Abhishek Gupta is Co-Founder at Dekrypt Labs, building BIOS — a Business Intelligence Operating System for Indian businesses. dekryptlabs.com