Ten Minutes,
Twenty Years.

Quick commerce is India's fastest-scaling retail format, and as of the December 2025 quarter it is no longer structurally loss-making. Redseer put category GMV at roughly ₹11,000 crore in January 2026 alone, about double the year-ago level, on 7.8 million orders per day. Annualised, that is a ₹1.3 lakh crore run-rate.

The milestone was Blinkit's first positive adjusted EBITDA: ₹4 crore on ₹13,300 crore of net order value. The number is trivial; what it establishes is not. Management disclosed that Delhi NCR ran at roughly a 3.5% adjusted EBITDA margin, Gurgaon and Noida nearer 5%. That is the only public evidence of what a mature Indian dark-store cluster earns, and every valuation in the sector is underwritten against it.

The consequence is uncomfortable. Once a format is proven to earn 4 to 6% at maturity it becomes a strategic retail question, which is how Reliance, Walmart and Amazon think, and all three moved in CY2025. The category's economics are proven; the industry's economics are not yet defended. Within listed exposure we prefer Eternal over Swiggy on funding capacity rather than execution: food delivery at a record 5.4% margin plus ~₹17,800 crore of cash, against Swiggy's ₹1,065 crore quarterly loss.

Published estimates of this market range from US$3.6bn to US$15bn for broadly the same moment, because different houses measure gross order value, net order value or platform revenue. We work in GOV and NOV and always say which.

Roughly US$7 to 8bn of GMV in FY25 after a 110 to 130% CAGR from a standing start in FY20. The category is already two-thirds of e-grocery orders but a tenth of Indian e-retail spend, against a US$1 trillion retail market where modern trade remains under 15%.

1. Frequency. The cheapest growth available. On Blinkit's 23.6 million monthly customers, moving a user from four to six orders adds 50% GMV at no acquisition cost and improves store utilisation, which is the same thing as margin.
2. Basket. Instamart's AOV rose ~40% to ₹746 on non-grocery mix; Blinkit's was flat. Catalogues went from ~2,000 impulse SKUs to 25,000 to 30,000, raising AOV and gross margin but also inventory risk.
3. Geography. The uncertain one. The category sits in the top 15 to 20 cities, and Reliance and Flipkart are attacking tier-2/3 with existing infrastructure. We doubt metro economics travel intact; see §06.

Everything reduces to one question: what does a single dark store earn at maturity, and how long does it take? No platform discloses a store-level P&L. What follows is our reconstruction, calibrated to Blinkit's Q3 FY26 EBITDA of 0.03% of NOV, NCR at ~3.5%, Gurgaon and Noida at ~5%, and Instamart's −2.5% contribution margin. We model a ₹550 net order value against Blinkit's reported ₹547.

The fixed cost is the store, not the rider. Last-mile delivery is genuinely variable and cheap in India; rent, a staffed picking floor and refrigeration run whether the store does 400 orders a day or 1,400. A new store is therefore catastrophic and a mature one excellent, and any platform's aggregate P&L is a weighted average of two different businesses.

That makes the J-curve the whole model. Blinkit added 211 stores, ~10% of its base, and still printed positive EBITDA, meaning the mature cohort carried an unusually large immature one. A platform can improve reported margin simply by slowing down, which is what Zepto did. Never read a margin print without net store adds beside it. Separately, strip the 4.5% advertising line from the mature column and margin falls from ~5.2% to ~0.7%: the equity value sits almost entirely in retail media.

That ~25% premium is the entire explanation for Blinkit at breakeven and Instamart at −2.5% on comparable assortments and rider costs. Density compounds: more orders per store means tighter radii, higher rider utilisation, better inventory turns and lower shrinkage, four margin lines improving off one operating variable.

Through FY25 this was three venture-funded specialists. It is now six platforms with two cost structures: specialists who build dedicated dark stores and must earn a return on them, and incumbents attaching quick commerce to retail estate, logistics and customers that already exist and are already paid for.

Reliance is the most threatening because it is the most capital-efficient: routing orders through stores that already earn their rent means the marginal cost of a JioMart order is close to pure last-mile. The node needs no 1,000-order threshold because it is justified already. The counter is that this is an operating business, not a real-estate one, and JioMart Express launched in 2022 and shut in 2023. Our base case is incumbents taking tier-2/3 while specialists hold the metro cohort where throughput decides.

1. Is the terminal margin 4 to 6%, or 8 to 10%?

Swiggy frames Instamart's ambition as ₹1 lakh crore of NOV at 4 to 5% EBITDA margins, and Blinkit's NCR cluster sits at 3.5%. Company guidance therefore clusters at 4 to 5%. Bulls argue that is pre-advertising. Bears argue it is what you earn in India's best catchments before Reliance arrives.

2. Does incumbent entry cap returns permanently?

The debate we hold with least confidence. If fulfilment from existing retail estate delivers comparable service at a fraction of the incremental fixed cost, the specialists' density advantage is neutralised in every city Reliance already operates in, which is most of them. The evidence against is operational: a late order destroys the habit the model depends on. We would revise sharply if Reliance began disclosing delivery-time distributions.

3. Do metro economics travel to tier-2?

The market is too casual here. The runway's quality declines with each city added: smaller baskets, lower frequency, and store rent that does not fall proportionally because square footage tracks assortment, not ticket size. Our construct breaks at roughly a ₹400 NOV. That threshold, not the population, is the real boundary of the addressable market.

4. Who is quick commerce taking share from?

CAIT alleges over ₹54,000 crore of FDI funding below-cost pricing; AICPDF claims ~200,000 kirana closures. These are advocacy figures, not audited ones. The defensible read is three things at once: cannibalising the kirana top-up trip, its most profitable business; denting the modern-trade stock-up only at the margin; and creating new consumption in impulse categories. The mix drives §08.

5. What happens when growth normalises?

Blinkit ties its stretch store plan to keeping growth above 100%. Current margins are being reported by companies actively suppressing them, and nobody knows what the sector looks like at 25 to 30% growth because it has never been there. A platform growing 100% at 0% margin is priced on terminal margin; the same platform growing 25% at 5% is priced on earnings. That transition is rarely smooth, and it is where the equity risk lives.

CY2025 gave the category a regulatory perimeter. Five workstreams matured at once, raising the structural cost base at exactly the moment the sector reached breakeven.

India's ~30 million kirana stores are an organised political constituency; against that, quick commerce is a large formal employer of urban gig labour. Our base case is regulation that raises cost and constrains discounting rather than restricting the format, which counter-intuitively favours the leader: rules that make buying share expensive protect whoever already has it.

At a ₹1.3 lakh crore run-rate the category is large enough to move adjacent P&Ls, and in places the read-across is more investable than the sector itself.

Conventional multiples produce nonsense here. Eternal traded on triple-digit trailing P/E because earnings are the residual of deliberate reinvestment, Swiggy has none, and the 1P shift made EV/Sales meaningless by inflating the denominator eightfold with no economic change.

At the mature column and 1,400 orders a day, one store generates roughly ₹1.6 crore of annual adjusted EBITDA on ₹40 to 60 lakh of fit-out plus working capital. An exceptional unit return, which is why four of India's largest companies are building them. The question is never whether the store is good. It is how many reach maturity, how long each takes, and what competition does to the margin en route.

Two variables: how fast GMV compounds off the January-2026 run-rate, and the margin the industry is permitted to keep. They are inversely related, since the fastest-growth scenarios are the most competitive.

Bear: the utility outcome

Incumbents accept structural losses indefinitely, discounting never normalises, contributions land on gross turnover and tier-2 proves worse. An excellent consumer utility and a poor business, roughly what Western grocery delivery produced.

Base: a four-player oligopoly

Blinkit consolidates at 40 to 45%, Instamart reaches sustained profitability around FY28, Zepto lists and survives as number three, and one incumbent holds geographies the specialists cannot serve economically. Margins settle at 3 to 3.5%: below guidance because six-player competition never clears, but decisively positive.

Bull: three players plus advertising

Retail media matures faster than expected, one or two incumbents scale back, and regulation lands on net revenue. The sector earns 5.5 to 6.5% on a US$65bn base, making it one of the most attractive retail profit pools in Asia, and today's valuations look modest.

• Competitive escalation. If Amazon's zero-fee model forces Blinkit to start discounting, Q3 FY26 is a high-water mark rather than an inflection.
• Contribution base. Social security defined on gross turnover rather than net revenue hits 1P platforms disproportionately.
• Tier-2 economics. Our ₹400 NOV breakeven is an estimate. If real baskets fall below it, the addressable market is roughly half what consensus assumes.
• Leadership transition. Goyal's step-down on 1 February 2026, at a company valued on execution.
• 1P inventory risk. Shrinkage, obsolescence and fresh wastage are now a P&L event, not a seller's problem.
• Zepto's listing. A poor reception compresses the sector's valuation reference set regardless of operating trajectory.
• Food safety tail event. Platforms are directly liable; a serious incident carries licence-suspension risk no model prices.
• Our own construct. If central costs scale worse than §03 assumes, the mature margin conclusion changes and much of this report follows it.

Glossary

Method

• Data cut-off 31 January 2026. Latest reported period is Q3 FY26: Eternal on 21 January, Swiggy on 29 January.
• Reported versus estimated. Disclosures appear unqualified; every analyst-derived figure is labelled, including all of §03 and the FY30 scenarios.
• Private company figures come from statements, filings and financial media, with wider error bars than listed disclosures.
• Third-party sizing is directional only; estimates vary by more than 4× on definition alone.
• Conventions. ₹88/US$. Indian financial years; FY26 ends 31 March 2026.

Principal sources

Eternal and Swiggy Q3 FY26 shareholder letters · Reliance Q3 FY26 investor presentation · Redseer, January 2026 · IBEF · Ministry of Labour and Employment: labour codes notified November 2025, draft Social Security (Central) Rules 2026 · CCI cost-of-production regulations, May 2025 · CAIT and AICPDF submissions · SEBI disclosures on Zepto's pre-filing · Indian financial press.

On the construct in §03: reverse-engineered from four disclosed anchors: Blinkit's ₹4 Cr adjusted EBITDA on ₹13,300 Cr NOV; NCR at ~3.5% and Gurgaon/Noida at ~5%; Instamart's −2.5% contribution margin; and Eternal's quantification of the 1P shift at roughly one point of margin. The split between delivery, store opex and central cost is our estimate and could be wrong in composition while right in aggregate. We publish it because a framework you can disagree with is more useful than a conclusion you cannot check.