"Beware the Gold Rush." — A response

By Nitin Misra (Co-Founder, indiagold) · 2026-07-21 · 6 min read

TL;DR — Episode 97 of the India Fintech newsletter is titled "Beware the Gold Rush." It's the sharpest cold-water piece on gold lending I've read this cycle, and if you invest in, build in, or compete in this market, read it before you read me.

Most of the points in the article, I have covered on Gold Loan Insider earlier. I shared the article in an earlier post.

Below is a point-by-point — where they're right, what we do differently at indiagold, and where the criticism lands on us too.

1. "The growth is mostly gold price, not gold."

Right. The industry compounded AUM at ~34% over five years while Muthoot's pledged tonnage grew 1–3% a year. When price per gram rises, the same ornament supports a bigger loan. Book grows; business doesn't. I wrote the same thing a week ago. Grams are price-neutral, rupees flatter you.

Any gold lender not reporting tonnage to its own board is grading its own exam ;-)

2. "Entrants are building thousands of branches at the top of the cycle."

Right, and we want no part of it. NBFCs plan ~3,000 new gold loan branches in FY27. Our model is 4–5 small hubs per city where a traditional lender needs 100–150 branches. Our branches store collateral and ensure compliance. They don't sell. Acquisition, assaying oversight, underwriting and servicing happen in the field and at a central control room. Customers don't walk in; we go to them.

3. "If you can't see a path to sub-4% opex, don't build."

This one hits home, so let me be straight about it. Muthoot runs at 2.6% opex-to-AUM after decades of branch productivity. Our hub economics beats that on paper, and our sites carry more AUM than the industry's breakeven threshold. But paper alone isn't proof. At our scale, company-level sub-4% opex is a target we still have to earn, but we know the solution and are on the path to it.

4. "Pristine credit, expensive operations."

The article's cleanest insight. Gold lending has near-zero credit cost and 4–6% operating cost, so the contest is decided in operations, not in underwriting losses. Zero net credit loss in a fully collateralised product is table stakes, not an achievement. Cost-to-serve is the whole game — which is exactly why point 3 is the one that decides who survives.

5. "Read a gold loan twice: once as collateral, once as behaviour."

Best line in the piece. Some borrowers pledge gold to stay current on personal loans and credit cards, and CIBIL now flags it. We've underwritten this way since 2020, out of necessity: balance transfers are our core business, and a BT customer arrives with a live credit history you'd be a fool to ignore. Our risk engine scores 40+ variables — bureau, bank statements, occupation, neighbourhood, family, stability — on every transfer. Most of our customers also carry unsecured loans at far higher rates; telling the working-capital borrower apart from the distressed one is not a feature of our model. It is the model.

6. "You're at a cyclical peak. What happens when gold corrects?"

Fair question to ask the lender offering the highest LTV in the organised market. Incumbents actually operate at 57–67% LTV; we go to 75–85%.

Here's what the question misses: the correction already came. Gold fell about 18% from its late-January peak. An 85% LTV loan written at the top was underwater on paper within eight weeks. What holds the system: RBI's valuation base is 22-karat at the lower of the 30-day average and the previous close, so the collateral never counted the spike in the first place; monthly interest servicing keeps exposure from compounding; and margin-call discipline (amongst other pricing, product, policy, process measures) does the rest. We came through with our record intact.

But I'll concede what the data shows: demand for refinancing slows when prices fall, because top-up room shrinks. But the LTV gap (industry vs us) will continue to fuel our mouth of funnel.

7. "The June 2025 RBI directions cut both ways."

Agreed, and worth being precise, because the LTV rule is misquoted constantly. Three bands on consumption loans — 85% up to ₹2.5 lakh, 80% to ₹5 lakh, 75% above. Income-generating loans carry no LTV cap at all; the limit sits in your board-approved policy, with a continuous maintenance obligation and public disclosure attached. The regulation rewards exactly what is expensive to retrofit: centralised valuation, audit trails, custody discipline. That's the stack we spent six years building.

8. "The interesting money isn't another balance-sheet lender. It's picks-and-shovels and underwriting intelligence."

Half agree, and the half that disagrees is our biggest bet. We built the picks and shovels — assaying tech, custody infrastructure, the BT risk engine — and then chose to become a lender anyway, licence and all. Why: in gold, underwriting learning only compounds when your own money sits behind every decision. Years of lending under first-loss guarantees taught us more than any SaaS dashboard would have.

The article's warning still stands, though. A lender that banks can underprice by 1,200 basis points, with nothing else going for it, is a bad idea. The stack is the business. The book is the proof it works.

9. "Growth moderates to 12–15%. Books can contract if prices fall."

Probably right, and it changes less for us than for most. Our engine is refinance — winning customers other lenders already hold. A share-taking model doesn't need the category to boom. It needs customers to be underserved, and at 57–67% actual LTV, they are.

The gold rush will end the way the article says: by punishing everyone who mistook price for progress. The way to avoid being one of them isn't optimism. It's counting grams instead of rupees, running costs like they're the product, and reading every loan twice.