Zepto's DRHP and the Inevitable Cost at the Heart of Quick Commerce

ELI5/TLDR

Zepto is going public, and its DRHP makes one thing plain: the whole business rests on dark stores — small neighbourhood warehouses that fulfil 10-minute deliveries. The trouble is that a dark store is a fixed cost. Rent, staff, and software get paid in full from day one, whether the orders come or not. The industry’s pitch is that scale will eventually spread those fixed costs thin enough to turn a profit, but the only profitable player, Blinkit, makes a 0.3% margin, and Zepto’s losses are still growing. The bet is that revenue arrives before the fixed costs bury the company. It’s a bet, not a promise.

The Full Story

The thing under the app

Quick commerce looks like a software business. It is mostly a real estate business wearing a software costume. The episode opens outside a Blinkit dark store in Indiranagar — riders idling on bikes, brown paper packages going out every few minutes — and uses it to make the point that these unglamorous 600-to-2,000 square foot warehouses are the actual machine.

Each store holds 2,000 to 3,000 SKUs and serves a 2-to-5 km radius. The choreography is tuned for speed: an order hits the system, a packer gets item locations on a handheld, walks the aisles, scans, bags it, drops it at a numbered pickup slot. A rider scans the barcode and leaves without ever entering. Ten minutes, give or take.

This hyper-efficient model is what has sprouted this underbelly of a real estate business under the tech layer.

The fixed-cost problem, stated plainly

Here is the uncomfortable bit, and it’s the spine of the whole piece. A dark store is a fixed cost.

No matter what happens, whether demands are slow or high or workers are striking, rents have to be paid, warehouse workers have to be salaried, and shelf space has to be filled.

More than 1,100 stores means more than 1,100 lease agreements, plus managers, multi-shift packers, riders, warehouse software, inventory systems, cold storage, security. All of it bills from day one. Order density — the thing that justifies the cost — takes months to build. So the meter runs before the revenue does.

Zepto’s own number: stores now reach contribution margin positivity in about six months, down from a reported 15-to-18 months two years ago. The episode does the reader a favour by not letting that headline slide past. Contribution margin positive means a store’s orders cover the direct cost of fulfilling them, with a little left over. It does not mean profit. The overheads — and head-office cost, and the cost of the stores still ramping — keep running on top.

The scale argument, and the holes in it

The bull case is the classic one: pile up enough orders per store and the fixed costs spread thin. Shorter delivery radii (more stores, closer to customers) also cut last-mile cost per order.

Zepto is, to its credit, chasing density rather than just planting flags in new cities — clusters of stores inside markets it already owns, plus bundling several neighbourhood orders onto one rider. The numbers move in the right direction: orders per day per store rose ~50%, from roughly 1,500 to over 2,000 between March 2025 and March 2026. Supply-chain variable cost per order edged down from 63 to 61 rupees over two years. Small, but the right sign.

Then come the catches, and there are several:

The model only really sings in the densest, richest pockets — Juhu, Indiranagar, GK — where stores clear 2,000+ orders a day and baskets include high-margin items (artisanal sourdough, cheeses, coffee beans, kimchi) rather than just milk-bread-eggs.
India is price-sensitive. If discounts fade and delivery fees rise, customers thin out. Zepto’s inventory-led model also means it eats the loss on high-margin stock that doesn’t sell.
There aren’t infinitely many affluent urban customers, or infinite urban real estate. Expanding to tier-two cities means weaker demand, worse logistics, more price sensitivity — and rising customer-acquisition spend.

That last point already shows. Despite over 1,300 crore rupees on advertising and promotion in FY26, Zepto’s user base actually fell by more than a million users between December 2025 and March 2026.

The cost side is moving the wrong way too

Real estate is squeezing from the other end. The Print reported quick commerce firms are bidding above asking price to lock locations; with vacancy in Mumbai and Bangalore at 3–5%, prices were already steep. UBS pegs an average dark store at 45–60 lakh rupees of capex, and rising. Flipkart and Amazon are entering harder. FMCG price hikes of 3–10% mean platforms either absorb the cost (eroding margin) or pass it on (chasing away price-sensitive buyers).

The number that hangs over everything

Blinkit is the sector’s only profitable player, profitable since 2024. And yet:

Despite operating over 2,000 dark stores with 17 million square feet of space and deploying over 4 lakh delivery partners, the business reported an adjusted EBITDA of just 37 crore rupees.

That’s a 0.3% margin on net order value. Meanwhile Zepto’s revenue roughly doubled to ~23,000 crore in FY26, but net loss widened to almost 6,000 crore from about 5,000 crore — and the DRHP itself warns losses will continue near-term.

The cautionary tale is Dunzo, once a verb, which ran the same distributed-inventory playbook at 120 dark stores across 15 cities and couldn’t afford it. The giants run the identical model; the only difference is deeper pockets.

Key Takeaways

A dark store is a fixed cost — rent, staff, software, cold storage all bill from day one, before order density builds. This is the structural problem the whole sector is betting it can outgrow.
Contribution margin positive ≠ profitable. Zepto hits the former in ~6 months per store (down from 15–18 months), but overheads run on top of that.
Zepto plans ~1,900 new dark stores (1,600+ crore earmarked from the IPO) on top of 1,139 existing, targeting ~3,039 — neck-and-neck with Blinkit’s plan to reach 3,000 by March 2027.
Blinkit, the only profitable player, earns just 0.3% adjusted EBITDA margin (37 crore rupees) on 2,000+ stores. That’s how thin “success” looks in this sector.
Zepto FY26: revenue ~23,000 crore (up from ~11,000 in FY25), net loss ~6,000 crore (up from ~5,000). The DRHP openly forecasts continued near-term losses.
Despite 1,300+ crore in FY26 ad spend, Zepto’s user base shrank by 1M+ between Dec 2025 and Mar 2026.
The unit-economics gains are real but tiny: orders/day/store up ~50% (≈1,500 → 2,000+); variable cost per order down 63 → 61 rupees over two years.
A dark store costs 45–60 lakh rupees of capex (UBS), and rents are rising because operators bid above asking price for scarce locations.
The model depends on dense, affluent pockets buying high-margin items; India’s price sensitivity and limited urban real estate both cap the scale story.
Dunzo is the warning: same distributed-inventory model, 120 stores, couldn’t fund it. The survivors differ only in pocket depth.

Claude’s Take

This is a tight, honest piece of business explanation — the kind that resists the easy “Zepto is doomed” or “Zepto is the future” framing and instead names the actual mechanism: a fixed-cost asset base betting that revenue arrives before the costs compound. The single best move is refusing to let “contribution margin positive in six months” pass as good news without unpacking it, because that exact phrase is how the sector launders its losses in headlines.

Where it earns the 7 rather than higher: it’s a synthesis episode, not original reporting. Nearly every fact is sourced from elsewhere — Outlook Business, Inc42, The Print, UBS, Entracker, the DRHP. That’s fine for a daily podcast, but it means the analysis doesn’t go past the obvious bear case. It never seriously engages the strongest bull rebuttal — that a 0.3% margin on a hyper-growth base, with density and bundling still early, could expand materially as cohorts mature and ad subsidy normalises. The Blinkit 0.3% number is presented as damning, but a margin can be thin and inflecting; the piece treats it only as ceiling, never as floor. It also leans on the user-base dip (one quarter, possibly seasonal or a cohort-cleanup artefact) harder than one data point deserves.

Still, the core insight is correct and well-delivered: this is a real-estate roll-up dressed as a tech IPO, and “we’ll be profitable at scale” is doing a great deal of unexamined work. For a 17-minute commute listen, that’s a fair trade.