Why:
- A third of the record year's profit has no stated cause. FY2026 gross margin doubled (3.9% to 8.0%) while revenue per kilogram outran the gold price by 3.6% and cost tracked it exactly (T1, derived to within 0.3%). The industry-wide mechanisms cover perhaps half the move; ₹30-40 crore of profit — roughly a third of PAT — remains unattributed, and the best candidate the filing contains is ₹144.6 crore of "non-hallmarked processing revenue" that appeared from nowhere in FY2026 and is explained nowhere. When gold flattened in April-June 2026, three of five listed peers' margins compressed; Deepa offers no post-March numbers at all.
- Volume went backwards while the listed peer set grew. True jewellery grams fell about 7.5% in FY2026 (the headline −13% includes bullion) while Shanti Gold — the closest listed B2B comparable, on a near-identical kilogram base — grew volume 15% (T1/T2). The CNC bangle line, half the stated specialisation, lost roughly a third of its volume unmentioned; growth survives only in vaddanam, a pool nobody sizes, while the biggest buyer group (Malabar) builds in-house capacity at ~2.9x Deepa's output (T3).
- Profit is not turning into cash, and the offer funds the gap. Operating cash flow is negative over three years against ₹170 crore of cumulative profit; FY2026 receivables absorbed more than the entire gross-profit increase, the worst deterioration in the peer set with no industry alibi; an off-books vendor-financing recourse is unsized in any note (T1). The ₹215 crore working-capital object is 2.35x the company's own certified FY2027 need in its first tranche and is sized on a +78% revenue plan with no order book and no disclosed capacity to process it.
Valuation at the band
| Floor ₹168 (T1) | Cap ₹177 (T1) | |
|---|---|---|
| Bid window | 1 Sep to 3 Sep 2026 | |
| Post-issue shares | ~9.69 cr | ~9.61 cr |
| Market capitalisation | ₹1,628 cr | ₹1,701 cr |
| P/E on FY2026 profit (post-issue shares) | 15.5x | 16.2x |
| Total offer | ₹449 cr | ₹460 cr |
| — of which promoter OFS | ₹199 cr | ₹210 cr |
| Promoter holding after | ~73% | ~73% |
The printed peer average is ~24x, so the band looks cheap — against a profit year whose quality is the first bullet above. At FY2025's margin the same company earned ₹40.6 crore, and the multiple on that is ~40x.
The story
You are buying a genuinely lean, family-built jewellery fabricator at what may be the top of its own margin cycle, in the year its volumes went backwards, its receivables blew out, and its owners monetised for the first time — with the one number that decides everything (why the spread doubled) not stated anywhere in 434 pages.
What this business is
Deepa Jewellers, Hyderabad, makes and supplies 22-karat gold jewellery wholesale to South Indian retailers — vaddanam bridal waist belts (42% of FY2026 revenue, its real specialisation) and CNC machine-cut bangles (31%), through 29 karigar (artisan) workshops it doesn't own and almost no fixed assets: 133 employees, operating costs of 0.40% of revenue, one-day payables, 18-day inventory, 84% of stock price-fixed back-to-back against orders. That model produces a real and replicable 52% ROCE. FY2026 revenue ₹1,927 crore, profit ₹104.8 crore.
Easy or difficult business? Fabrication coordination on razor-thin spreads. The lean model is the edge and it is copyable; there are no contracts, no brand with end-consumers, and the largest retail chains increasingly make in-house. What protects Deepa is craft specialisation in heavy bridal work and relationships — real, and unpriceable.
Key numbers
| ₹ cr | FY2024 | FY2025 | FY2026 |
|---|---|---|---|
| Revenue | 1,025 | 1,397 | 1,927 |
| Revenue growth % | n/a | 36.3 | 37.9 |
| Gross margin % | 3.9 | 4.4 | 8.0 |
| PAT | 24.3 | 40.6 | 104.8 |
| Operating cash flow | +4.9 | -9.9 | -14.7 |
All of FY2026's revenue growth is the gold price (+52.9%); grams sold fell. FY2025's margin is depressed by ₹86.5 crore of zero-margin bullion trading, since exited.
| Product line, ₹ cr (DERIVED) | FY2026 | Share | Note |
|---|---|---|---|
| Vaddanam (bridal waist belts) | ~806 | 41.9% | the only line with growing volume (+5-9% grams) |
| CNC machine-cut bangles | ~595 | 30.9% | revenue +1.8% against +53% gold — volume down ~a third |
| Other jewellery and processing | ~486 | 25.2% | includes the unexplained ₹145 cr non-hallmarked processing line |
| Job work | 17 | 0.9% | conversion price per gram down ~63% in two years |
| Bullion trading | ~2 | 0.1% | ₹86 cr in FY2025, exited |
Scorecard
| Block | Rating | Why |
|---|---|---|
| Right to win | MARGINAL | The lean model is real, fully explained and replicable — which is the problem: nothing defends it. Volumes underperformed the listed tier, the bangle line collapsed, job-work pricing fell 63% per gram, and the one growing line sits in an unsized market whose biggest buyers are integrating in-house. Craft depth in heavy bridal work and 373 named, growing customers keep it off FAIL. |
| Industry and TAM | MARGINAL | The commissioned report's own numbers put South India B2B at 3-5% nominal — falling volumes at current gold prices — and CRISIL separately forecasts FY2027 organised volumes down 13-15%. Share headroom is genuine (below 1% of the bangle pool), but the company is exiting the market that can be sized and growing in the one that cannot. |
| Financial momentum | MARGINAL | Real profits, clean restatement, honest tax line — and a record year built on an unattributed margin step, negative cash conversion three years running, the worst receivables deterioration in the peer set, and a repeatability test peers have already failed once since March. |
| Risks, governance, RPTs | MARGINAL | Zero litigation and a genuinely clean outside record — but the filing's compliance self-description is provably wrong in two places: zero provident-fund contributions in FY2024-25 against statutory-dues tables certified NIL, and a vendor-financing recourse obligation that appears in no note at all. Parallel family firms sit outside the related-party list. |
| Promoter and cap table | PASS | Ten years of retained earnings, no dividend, no outside round ever; personal guarantees and a mortgaged family building behind the company's bank lines; the ₹0.50 "cost" is a split/bonus artefact; sellers take only ~15% of their own holdings and keep ~73%. Signals noted: the loan account churns, and family pay and rent reset upward in the IPO year. |
| Offer structure | MARGINAL | No debt object, monitored, fresh-issue-first waterfall — but the working-capital ask is 2.35x the certified first-year need, sized on a +78% plan with no disclosed capacity, the projected receivable days rest on a basis switch, and up to ₹62.5 crore can legally drift to general purposes if the plan misses. |
Watch out for
- The margin question answers itself by December. H1 FY2027 gross profit per gram, against peers that already compressed when gold flattened, settles whether FY2026 was a level or a peak.
- The ₹145 crore processing line. What it is, who pays it, and whether it recurs is the single highest-value question to put to management.
- Gross or net? Two peers book customer-exchanged metal net of gold value; Deepa sources 57% of its metal that way and appears to book gross — the answer changes the revenue line and every comparison.
- The anchor book lands 31 August, before the 1 September open.
The offer
- Raising ₹449 to ₹460 crore: ₹250 crore fresh, the rest a promoter offer for sale (T1).
- For ₹215 crore of working capital and up to ₹35 crore of general corporate purposes; no debt repayment, no capex (T1).
- Implied valuation ₹1,628 to ₹1,701 crore, from the table above.
- Promoters hold 100% before and about 73% after; two of three sell ~14.8% of their own holdings (T1).
Assessed from the RHP with outside checks on peer results, industry volumes and the people. Numbers carry source tiers: (T1) the filing's audited/certified sections and exchange records, (T2) listed-company filings, (T3) trade press and secondary sources, DERIVED where computed from cited inputs. Not a recommendation. The valuation section states what the announced band implies and is not a view on whether that price is right.