Eternal: The Price of Hunger
11:47 p.m., and the kitchen is closed
A student in a Pune hostel, deep in an assignment, realises she has not eaten since lunch. It is 11:47 at night. Twenty years ago this story ends with biscuits. Tonight she opens an app, and consider — really consider — what happens in the next half hour. Software shows her ninety kitchens still cooking within three kilometres. She picks a biryani. An algorithm wakes the restaurant's order screen, simultaneously scans the positions of every delivery rider in the neighbourhood, and offers the job to a man on a motorcycle finishing a drop two streets from the restaurant. He accepts with his thumb. The kitchen's cooking time and his riding time are predicted and interleaved so that he arrives roughly when the rice does. At 12:19 a.m. her doorbell rings. The fee for orchestrating all of this — the matchmaking, the prediction, the man and the motorcycle through the night streets — was less than the price of the raita.
The company behind that half hour began in 2010 as Zomato, a website of restaurant menus, and now calls itself Eternal. The stock market values it at ₹2,73,539 crore. The profit it earned last year was ₹366 crore. Hold both numbers in your head — the wonder of the half hour and the ratio of those two figures, which is 747 — because this chapter is about the uncomfortable space between them.
The mathematics of a crowd
Why does one app show ninety kitchens while a rival shows forty? Not because of better software — software is the most copyable artefact humanity has ever produced. The answer is a piece of mathematics worth learning properly, because it governs a growing share of the modern economy.
Most businesses get linearly better as they grow: twice the shops, twice the convenience. A network gets better faster than linearly, because what it sells is connections, and connections multiply. A platform with N customers and M restaurants offers N × M possible pairings; double both sides and the pairings quadruple. More eaters make the platform irresistible to restaurants, whose joining makes it more useful to eaters, who attract more restaurants — a loop that feeds itself. Economists call this a two-sided network effect. Its practical consequence is winner-takes- most: once one platform is clearly larger, a rational restaurant lists there first and a rational diner looks there first, and the gap widens on its own.
But food delivery adds a third side to the network, and the third side is where the mathematics gets teeth: the riders. Here the crucial variable is density — orders per square kilometre per hour. A rider is paid, one way or another, for his time. If the neighbourhood generates six orders an hour along his routes, he delivers almost continuously and the labour cost of each drop is modest. If it generates one, you are paying a man to wait — and since he will not wait for nothing, either the platform subsidises him or he leaves, and delivery times lengthen, and customers leave too, and density falls further. Run the loop forward: density is destiny. It is why the same app that feels magical in Pune barely functions in a small town, why the second-place platform in any city loses money faster than the first, and why this industry everywhere on Earth has collapsed into duopoly. India's did: two platforms now carry the country's restaurant orders between them, this company being one.
The quick-commerce arm — Blinkit, delivering groceries in minutes — is the same theorem with a warehouse attached. The ten-minute promise is impossible from a distant depot; it requires stocking two or three thousand everyday items in small "dark stores" planted every few kilometres of dense city, each an inventory bet that the surrounding neighbourhood will order enough, soon enough. Density again — now with rent and stock riding on it. It is the boldest and hungriest part of the business: consolidated sales, ₹20,243 crore in Mar 2025, leapt to ₹54,364 crore in Mar 2026 — trailing growth of 169% — substantially because the company began buying and reselling the groceries itself rather than merely brokering them, which multiplies reported revenue and thins reported margin in the same stroke.
Put rough numbers on the rider loop, because the whole industry turns on them. Suppose a rider must take home about ₹150 an hour to stay on the road, all costs included. At three deliveries an hour — feasible only in a dense zone with clever batching — the labour cost per order is ₹50, which a delivery fee plus a restaurant commission can absorb. At one delivery an hour, it is ₹150 on an order whose food may cost ₹250, and no fee any customer will tolerate covers it; the gap comes out of the platform's pocket, order after order. Every rupee this company lost in its first decade — and we are about to count them — is, at bottom, some version of that ₹150-versus-₹50 gap, paid deliberately, in the belief that density would one day close it.
What a decade of hunger cost
The industry's history is short and written mostly in red ink. This company's own arc rehearses all of it. It began in 2010 doing the gentlest possible version of the business: photographing paper menus and putting them online, then reviews, then advertising — a media company about food, employing servers rather than riders, and pleasantly close to profitable. The trouble was strategic, not financial: a listings site owns the decision but not the transaction, and whoever owns the transaction can eventually buy the decision. When venture-funded rivals began delivering the food itself in the mid-2010s, the company had to follow — from asset-light media into the heaviest of logistics — or watch its audience become someone else's customers. Restaurant discovery made no real money anywhere on Earth; delivery was the monetisable act, and platforms worldwide discovered that customers had to be taught the habit with discounts, and riders paid to exist before density could pay them. Teaching a subcontinent cost what you would expect. This company's accounts, which begin in Mar 2018 in our data, read like a war diary: losses of ₹107 crore, ₹1,010 crore, ₹2,386 crore, ₹816 crore, ₹1,222 crore, ₹971 crore — roughly ₹6,500 crore consumed across six years before the first annual profit of ₹351 crore appeared in Mar 2024. Operating margin touched minus 171% in Mar 2019: for a time, every rupee of revenue was accompanied by nearly two rupees of expense.
The quick-commerce chapter of the history is younger still and stranger. Grocery delivery in thirty minutes was tried and buried in the American dot-com crash a generation ago — the vans were too big, the cities too sparse, the habit unborn. What resurrected it was a peculiarly Indian convergence: megacity density, cheap motorcycle labour, smartphones in every pocket, and a pandemic that taught a hundred million households to have everything brought to the door. The company bought its way in by acquiring Blinkit in 2022 — paying in shares, to widespread scepticism — and the sceptics have so far been wrong about the demand: quick commerce is the engine of that leap to ₹54,364 crore of revenue. Whether they were wrong about the economics is the still-open question this chapter keeps circling.
Who paid for all this schooling? Not lenders — borrowings are a modest ₹4,592 crore. The funding came from selling pieces of the company, over and over: venture investors, then a celebrated 2021 public listing, then more. You can watch it in the equity line, which swelled to ₹919 crore of face value against reserves of ₹30,061 crore — reserves that represent, overwhelmingly, cash raised from shareholders rather than profits retained from customers. This is the inverted physiology of a platform business and the honest way to describe this company's capital history: a decade of capital consumption in exchange for market position. The barrier to entry it bought is real — any challenger must now burn comparable billions against an incumbent with density — but a barrier bought with shareholders' money is only worth what stands behind it can eventually earn.
Munger reads the annual report
Now the register this company's shareholders most need and least enjoy. Munger's habit was to admire a business freely and then refuse to let the admiration touch the arithmetic. Let us do exactly that.
The admiration is genuine. Two-sided density economics is among the strongest competitive structures known; the duopoly is stable-looking; the brand is a verb in Indian cities; and management has shown it can open a second front (quick commerce) and take it to national scale in a few years — optionality of real value.
Now the arithmetic, performed slowly. Last year's net profit: ₹366 crore. Sitting inside the earnings, as the screener's own caution notes, is other income of ₹1,396 crore — largely the yield on the mountain of raised cash. A businesslike reading is therefore that the operations — the apps, the riders, the dark stores, all of it — earned approximately nothing, or less, and the reported profit is substantially the interest on money shareholders handed over. Return on equity: 1.19%. A fixed deposit is not a fair comparison for a business with this trajectory, but it is a humbling one, and Munger would make it anyway. Operating cash flow has at least turned positive — ₹632 crore in Mar 2026 after years like Mar 2020's minus ₹2,144 crore — so the fire is out. But the machine has not yet demonstrated that it can earn owner-sized profits, and the market has priced it at ₹2,73,539 crore, 747 times those thin earnings and 8.75 times book value, as if the demonstration were a formality.
He would probe one more soft spot: switching costs, which in this industry are close to zero. Both apps sit on the same phone; the rider often wears one company's jacket over the other's T-shirt; the restaurant lists on both. Economists call this multihoming, and it is the termite in every delivery platform's woodwork — network effects deter a new entrant magnificently, but they discipline an existing duopolist hardly at all, because every customer is one tap from defection. The duopoly's peace therefore rests not on lock-in but on mutual exhaustion: both sides have burned enough to prefer margin to war. Peaces of that kind hold until one signatory smells weakness.
Here Munger's old warning applies with full force: a great business can be a poor investment at the wrong price, and no growth rate exempts a buyer from that law. Pay 747 times earnings and you have prepaid decades of flawless execution — margins must expand enormously, the duopoly must never turn on itself, quick commerce must beat back every deep-pocketed conglomerate now entering it (and India's largest business houses are all entering it), and regulation must stay friendly to a model built on a million gig riders whose employment status is a live political question. Any stumble is not priced. Perfection is the base case. One may believe in the business, as this chapter broadly does, and still recognise that belief at this price offers no margin for the ordinary misfortunes of corporate life.
What would a decade well spent look like? Prove the core theorem: take delivery fees and advertising income up gently, year after year, and show that density, once owned, actually converts to margin the way the mathematics promises. Make Blinkit's dark stores individually accountable — each one either earns its rent and stock or closes. Treat the raised billions as a trust, not a war chest for glamour acquisitions. Keep the rider proposition decent ahead of regulation rather than behind it. And begin, someday, the habit that separates institutions from stories: returning cash. Dividend paid in the company's entire life so far: zero.
An owner's ledger, briefly
The Buffett movement of this chapter is short, because the owner's questions mostly return the answer "not yet." Predictability: low — revenue is exploding but its mix is being remade annually. Pricing power: emerging — take rates and platform fees have inched up without customer revolt, the most encouraging fact in the file. Return on incremental capital: unproven — enormous capital has gone in; the returns are a promise. Resilience: the service is a habit millions will not unwind, which is worth much. Ownership: there is no promoter at all — the founders hold a sliver, and the register is institutions all the way down, foreign holders at 32.61% and domestic funds at 35.88% as of Mar 2026, with 23 lakh shareholders along for the ride. A company owned by everyone is disciplined by no one in particular.
The moat verdict is narrow — real, mathematical, defended by density and a duopoly, but young, unproven at converting position into profit, and besieged on the quick-commerce flank by rivals with older, deeper pockets. A narrow moat around a castle priced as a wide one.
Ways this ends badly
Three structural hazards, none of them quarterly. First, the gig-labour question: the entire cost model assumes flexible, per-drop labour; a legal reclassification of riders as employees would rewrite the density arithmetic overnight, and the political wind blows only one direction on this. Second, competitive re-entry: food delivery's duopoly looks settled, but quick commerce is a land war against India's largest conglomerates, who can subsidise groceries from oil, telecom, or retail profits far longer than this company can from ₹366 crore. Third, the commodity trap: if delivery speed and selection converge across platforms, the consumer's loyalty reverts to whoever discounts today — and network effects, powerful while quality gaps exist, offer little defence against a rival willing to lose money indefinitely.
The impatient decades
Structurally, the demand story needs no imagination. Urban India is young, crowded, time-poor and increasingly wealthy; the kitchens of a hundred million households are shrinking as apartments do; and every year the generation that thinks of dinner as a thirty-minute promise grows larger than the one that thinks of it as a stove. Delivery of everything — meals, groceries, medicine, whatever fits on a motorcycle — will be a vastly bigger industry in 2050. Automation may eventually thin the labour question itself.
The open question was never the size of the river; it is how much of the water the boatman gets to keep. Every fee sits in a triangle of pressure — restaurants resent the commission, customers resent the delivery charge, riders need a living — and the platform's profit is whatever the triangle grudgingly permits. This company has built something genuinely difficult and genuinely valuable: the machinery of urban India's impatience. Whether the machinery can ever earn what its price already assumes — that is not cynicism; that is division, and the divisor today is 747.
An Omaha Investments chapter. Educational material, not investment advice. Figures from Screener.in and NSE data via Angel One as of the date above.