← all companies · 2026-07-07 · moat: narrow · raw .md

Larsen & Toubro: Concrete Forgets, Reputation Doesn't

A bridge in the rain

Picture a highway bridge in the first week of a heavy monsoon. A loaded truck crosses the middle of the span. Under its wheels, invisibly, the bridge bends — perhaps a few millimetres over a thirty-metre span. That tiny sag divides the deck into two worlds. The top surface of the beam is being squeezed, its material pressed together. The bottom surface is being stretched, its material pulled apart. Every beam that has ever carried a load, from a plank across a stream to the deck of a sea-link, lives with this split personality: compression above, tension below.

Now the material problem. Concrete — crushed stone and sand glued by cement — is marvellous in compression. Squeeze it and it shrugs; it is, in effect, artificial rock, and rock has been holding up mountains for rather a long time. But pull on concrete and it is a disgrace: it cracks at roughly a tenth of the stress it can bear in compression, because a crack, once started, has nothing to stop it — the glue simply unzips. Steel is the mirror image: superb in tension, because metal atoms slide and stretch rather than unzip, but expensive, and in long slender pieces it buckles when compressed.

So the nineteenth century's great construction insight was a marriage: pour the cheap rock where the squeezing happens, and bury ribbed steel bars where the stretching happens. Reinforced concrete is not a material so much as a committee — concrete takes the compression, steel takes the tension, and each covers the other's weakness. The marriage works because of a genuine coincidence of nature: steel and concrete expand by almost exactly the same amount when heated. Had their thermal expansions differed much, every summer would crack the bond between bar and stone, and the modern city could not exist. On such small mercies rests everything Larsen & Toubro builds.

One more idea completes the physics. A bridge, a metro viaduct, a refinery, an airport terminal is not a product; it is a site-specific experiment run once, at full scale, with no prototype. A factory that makes soap can perfect the recipe on batch after batch. A builder gets one attempt per bridge, on that river's soil, in that year's monsoon, with that particular mix of labour and steel prices. This is why construction is less like manufacturing and more like surgery: the knowledge that matters is not a formula but a record — thousands of previous operations, survived.

Why the cheapest bid is a trap

Move now from physics to money, because the physics writes the industry's economics directly.

Society plainly needs the work — a country's productive capacity is largely its stock of built things — and governments and corporations pay for it through contracts. The standard contract is brutal in a specific way: the builder quotes a price today for work that will take five to ten years, and most of the costs — steel, cement, diesel, wages, the behaviour of the riverbed — will only be discovered along the way. A fixed-price construction contract is, in effect, a promise about the future made with incomplete information. The bidder who most underestimates the difficulties quotes the lowest price and wins. Economists call this the winner's curse; contractors call it Tuesday.

This is why construction, globally, is a thin-margin trade in enormous numbers, and why the profit pool hides in unglamorous places: in risk pricing (knowing which surprises to charge for), in execution speed (a project finished a year early releases an army of engineers and cranes to earn elsewhere), and in claims (the legal aftermath of who pays for the surprises). The companies that die in this industry rarely die of bad engineering. They die of one underpriced megaproject, or of the slow anaemia of receivables — work done, bills disputed, cash withheld.

And this is why the order book is the industry's true telescope. A builder's revenue this year was decided by contracts signed years ago; the contracts signed this year decide revenue years hence. Think of the order book as a grain silo: wins pour in at the top, execution draws down from the bottom, and the ratio of silo to annual draw-down tells you how many years of work are already assured. But the silo has a catch — the margin on every tonne of grain inside was fixed on the day it was poured in. A fat order book won at desperate prices is not an asset; it is a queue of pre-committed mistakes. The owner of a construction company should therefore always ask two questions, never one: how full is the silo, and at what quality was it filled?

Two Danes, stranded

The company's origin is one of Indian industry's better accidents. In 1938, two Danish engineers, Henning Holck-Larsen and Søren Kristian Toubro, set up in Bombay as agents importing Danish dairy and cement machinery. Then the war came, Denmark fell under occupation, and the supply of machines from Europe simply stopped. Cut off from the products they were meant to sell, the two engineers did the only thing available: they began to build and repair machinery in India itself. A trading agency, stranded, turned into an engineering company — and kept turning, decade by decade, into whatever independent India most lacked: dairy plants, then cement plants, then hydrocarbon refineries, nuclear reactor components, submarines, metros, and the world's tallest statue.

Two features of this history still shape the firm. First, it grew up as an engineer's company in a capital-starved country — its currency was competence, not connections, and it recruited and promoted accordingly. Second, and rarer: the founders went home and left no dynasty. Larsen & Toubro today has no promoter at all — the shareholding register shows domestic institutions at 43.32%, foreign institutions at 18.78%, and the public at 37.64%, with the balance held by government funds. It is one of India's very few large companies that is genuinely ownerless, run by professional engineers who rose through its sites, and defended over the decades — sometimes dramatically — by its institutions and its employee trusts against outsiders who fancied owning it. In a country where the promoter's surname usually explains the company, this firm's surname is a pair of Danes nobody alive has met. The culture is the inheritance.

What it takes to be trusted with a refinery

Could a well-funded newcomer replicate this business? Consider what the newcomer must assemble. Not mainly machines — cranes and tunnel-borers can be bought or leased by anyone. The scarce assets are:

A record. A government awarding a metro, or an oil company awarding a refinery, is buying a decade of its own future. It will not hand that to a firm without a list of finished projects, and the list can only be built by finishing projects — a loop that takes a generation to enter. Reputation here is not marketing; it is collateral posted against every bid.
An engineering bench. Tens of thousands of engineers who have actually poured piles into a tidal riverbed or hydro-tested a pressure vessel. This bench cannot be hired in bulk; it is grown, project by project, and it walks out the gate every evening.
A balance sheet that can bleed. Megaprojects consume cash before they return it — mobilisation, materials, milestones disputed. The company's own cash-flow line shows what this means: cash from operations was negative ₹1,365 crore in Mar 2015, negative ₹3,240 crore in Mar 2016, and negative ₹10,031 crore in Mar 2018, before swinging to +₹23,074 crore in Mar 2021 and +₹16,741 crore in Mar 2026. A builder's cash flow breathes in multi-year tides, and a newcomer without deep lungs drowns in the first exhale.

The global field confirms the difficulty: most countries have exactly a handful of contractors trusted with the largest work, and the club changes membership about once a generation. Scale helps, but the wall is really trust with irreversibility — the licence society grants to firms allowed to do things that cannot be undone cheaply.

The seminar: renting out a reputation

Now reason about this company the way a multidisciplinary investor would.

Start with incentives, because in contracting they are everything. The firm's revenue people are paid to win orders; its project people inherit the wins. Every construction company on Earth fights this internal war, and the losers are recognisable years later by fat order books and shrivelled margins. The evidence here deserves an honest reading: operating margin has slid from 17% in Mar 2015 to 12% in Mar 2026 while sales tripled from ₹91,929 crore to ₹2,85,874 crore. Some of that slide is mix — more construction, less of the higher-margin manufacturing and services in the blend — but an owner should hold the number in view. In this trade, growth is easy to buy and margin is easy to sell; the skill is refusing both temptations at once.

Second, opportunity cost and capital allocation. An engineering conglomerate is permanently tempted to own what it builds — roads, metros, power assets — because ownership flatters the order book twice. But a builder's return comes from turning competence quickly; an infrastructure owner's return comes from sitting patiently on tolls. The two businesses want opposite balance sheets. The company's borrowings of ₹1,25,497 crore against reserves of ₹1,09,015 crore look alarming for a contractor until one recalls that a large financial-services and development portfolio sits inside the group — but that is precisely the point: the owner must keep asking whether capital is compounding in the firm's circle of competence (building) or merely accumulating in adjacent squares. Management has shown it can answer well — the equity line quietly fell from 281 to 275 crore of face value after Mar 2023, a buyback, the rarest of conglomerate behaviours.

Where, then, is the moat? Be precise. There is no network effect, no switching cost — every project is re-bid from zero, and yesterday's client owes the firm nothing tomorrow. Concrete forgets. What endures is the reputation-and-bench combination: the newcomer's twenty-year entry loop described above. That is a real barrier, but a narrow one, because it protects admission to the auction, not the price achieved in it. The firm competes for most work against other qualified builders, and the auction, not the moat, sets the margin.

The decade's to-do list, stated plainly:

Keep shedding owned infrastructure and non-core assets; return the proceeds or deploy them only where the bench is the advantage.
Defend bid discipline in the coming infrastructure boom — the silo must be filled at quality, not just volume; the 17%-to-12% margin slide should bottom and turn.
Convert engineering reputation into the era's new work — grid-scale renewables, green hydrogen plants, data centres, semiconductor fabs, defence platforms — before those niches crown their own specialists.
Guard the succession machine. An ownerless company's culture is its only will and testament; one bad professional generation can spend a century of trust.

The owner's site inspection

Now the Buffett questions, with the company's own numbers.

The whole enterprise is priced at ₹5,55,760 crore, about 33.9 times last year's earnings and 5.1 times book value of ₹794 per share. Earnings have compounded at 15% a year over the decade — ₹4,966 crore in Mar 2015 to ₹18,954 crore in Mar 2026 — on sales growth of 11%, with return on equity steady in the mid-teens (15% over ten years, 16% last year). The dividend payout has held near a third of profits, 32% last year.

Read those numbers as an owner. The good: this is a business whose earnings have grown through every kind of weather — a commodity crash, a pandemic, an interest-rate cycle — without a single loss year in the record, which for a contractor is quietly remarkable; most of the world's famous construction names have at least one crater in any given decade. The engineering bench, the ownerless culture and the builder of last resort status in a country that has decades of building left — these are durable.

The caution: a mid-teens return on equity is a good result for a narrow-moat business, not a wide-moat one, and the price of 5.1 times book asks the mid-teens to persist for a very long time. Predictability is middling — the order silo gives visibility on revenue but not on margin, and the cash flow line will always breathe in tides. Pricing power is close to nil at the moment of bidding; what the firm possesses is selection power — the right to walk away from bad auctions — which is worth a great deal only if exercised. And note the register: foreign institutions have trimmed from 25.29% to 18.78% since Jun 2023 while domestic institutions climbed from 38.01% to 43.32%, and the shareholder count grew from about 14.2 lakh to 18 lakh. The company is passing, hand to hand, into the ownership of the very country it builds.

Ways a builder can fall

The structural risks, in rough order of gravity:

One bad megaproject. The industry's classical death. Size caps and risk pricing are the guardrails, and both live in culture, not in spreadsheets.
Client concentration in the state. Governments are the ultimate customer for infrastructure, and governments dispute bills, defer payments, and change policy. The receivables cycle is a permanent tax on the trade, and a fiscal squeeze anywhere becomes a cash squeeze here.
Commodity and wage inflation inside fixed-price contracts — a short position in surprises, renewed with every bid.
The conglomerate temptation. Capital drifting from building (where the bench earns) to owning (where it merely sits). The last decade's discipline must be re-chosen each cycle.
Labour. The business ultimately runs on hundreds of thousands of hands at height, in heat. Safety, housing and skilling are not welfare lines; they are the production function.

The country as a construction site

The thirty-year view is unusually legible, because the customer's intentions are published: India means to urbanise several hundred million more people, and every one of them arrives with an unfunded order for concrete, steel, power, water and transit. No nation has industrialised without this build-out, and the build-out cannot be imported — a bridge must be made where it stands. The energy transition adds a second continent of work on top of the first: solar parks, transmission corridors, hydrogen electrolysers, nuclear units — all of it, whatever the technology, resolving at ground level into engineering, procurement and construction.

Technology will change the site more than the business. Design is already migrating into digital twins; modules are increasingly made in factories and assembled rather than built; automation will thin the crowds on the deck. Each shift favours the contractor with the deepest engineering bench and the balance sheet to retool — advantages of the incumbent, if the incumbent stays awake.

The permanent truth of the trade, though, was visible back on that monsoon bridge. Concrete takes the compression, steel takes the tension, and the builder takes the risk. The first two are covenants of physics and can be trusted. The third is renewed by human judgment at every bid — which is why this company's moat, real as it is, must be re-poured, like its concrete, one project at a time.

An Omaha Investments chapter. Educational material, not investment advice. Figures from Screener.in and NSE data via Angel One as of the date above.