SBI uses AI to underwrite nearly ₹1 lakh crore in MSME loans in FY26

SBI's AI Just Killed the 90-Day Loan Approval Cycle That Was Bleeding Your Working Capital Dry.

What Actually Happened

State Bank of India has deployed artificial intelligence across its MSME lending vertical and underwritten nearly one lakh crore in loans during FY26. This is not a pilot. This is not a press release trophy. This is 100,000 crore rupees moving through algorithmic decision-making instead of sitting on some loan officer's desk waiting for Monday morning coffee.

The approval timeline has compressed from 90 days to 21 days for loans up to 5 crore rupees. The algorithm reads your GST filings, bank statements, working capital cycles, and credit payment history in 48 hours. A textile unit in Tiruppur that used to pray for loan approval while their supplier payment date approached is now getting funds before the crisis hits. A food processing business in Maharashtra that runs on seasonal working capital can now plan cash flow three weeks ahead instead of three months ahead. This matters because in manufacturing and food processing, a 90-day delay is not an inconvenience. It is a margin killer.

What This Means for India

India's MSME sector has been starved of working capital at the exact speed that it bleeds fastest. We have 6.3 crore registered MSMEs. Only 26 percent get institutional credit. The rest are either borrowing from their cousin at 18 percent or running on vendor credit that edges toward 45-day terms. SBI's move to automate 1 lakh crore in underwriting is not charity. It is SBI realizing that manual underwriting was leaving money on the table because loan officers could only process 400-500 crore per head per year. AI processes 25,000 crore per head per year.

What changes is that the Indian MSME now has a financial option that did not exist last year. You do not need a relationship manager. You do not need to know someone's uncle. You do not need collateral games where the bank values your land at 40 percent of market rate. The algorithm does not care about your caste, your region, or whether you played cricket with the branch manager. It cares about three things: GST history, bank statement patterns, and payment defaults. This is the most democratic credit allocation system India's formal sector has ever seen.

The ripple is massive. Textile exporters can now bid for larger orders without the working capital paralysis that made them walk away from 30 percent margin deals. Engineering job-work units can take on more volume without the cash flow cliff that forced them to operate at 50 percent capacity. Food processing units can buy seasonal inventory when prices are right instead of when loans eventually cleared. For a country trying to build a 5 trillion-dollar economy, removing friction from capital deployment is not optional.

The Deeper Story Nobody is Telling

Here is what SBI is not saying loud enough: this AI system works because India's GST infrastructure created a digital financial footprint for MSMEs for the first time in history. Before GST, bank lending to small business was basically a relationship game. After GST, every single business transaction leaves a data trail. The algorithm is only powerful because the government forced digitization at the point of sale.

But there is a sharp edge. This same AI will reject 40 percent of MSME loan applications that would have been approved under the old relationship-based system. The companies with irregular GST filing, those 90-day payment cycles, those working capital lags, those vendor credit patterns that show instability, will now get algorithmic rejection faster and harder than they used to get human rejection. The AI does not negotiate. The AI does not give the benefit of the doubt. The AI does not have a mother who reminded it to be kind. This is good news if you run a clean operation. This is a wall if you do not.

What MSME Founders Must Do Now

First, audit your GST filings for the last 24 months. Check every return. Check every invoice timestamp. Check payment dates. Late GST filings will kill your AI score faster than a default. The algorithm weights consistency at 35 percent. One irregular filing in a 24-month window reduces your approval probability by 15 percent.

Second, clean up your bank statements. Stop taking cash advances at 5 percent commission. Stop keeping parallel accounts. The AI reads transaction patterns. It understands cash flow seasonality. It understands working capital cycles. But it hates opacity. It hates transactions that smell like black money or cash juggling. Make your money movement boring and predictable.

Third, prioritize vendor payment discipline. If your 30-day terms are actually 60 days in practice, your suppliers have already factored that into pricing. You are not saving money. You are just killing your credit score. The algorithm will see this. The algorithm will penalize you. Fix it now.

Follow BHARAT DECODED on Telegram: t.me/DecodedByRDS — Rajnish Sharma (RDS)