Bharat Electronics: One Customer, Who Owns the Shop
If you owned the whole company
Buy all of Bharat Electronics today for its market capitalisation of ₹3,11,031 crore. Last year — the year ended March 2026 — it earned ₹6,062 crore of net profit. That's about ₹2 of annual profit for every ₹100 paid, a P/E of 51.3. Fifty-one times earnings is a price that says: this profit stream is safe, growing, and will keep growing for a long time. The safety claim, as we'll see, is unusually strong. The price claim is the part that would make us sit on our hands.
But first admire the machine itself. Return on capital employed — what every ₹100 of money inside the business earns before it's split between lenders and owners — is 36.5%. There are almost no lenders: total borrowings are ₹65 crore against ₹23,257 crore of reserves. This is a business that mints high returns using almost nobody's money but its own.
What does this business actually do?
BEL, born in 1954 as a government defence factory, makes the electronic nervous system of India's armed forces: radars, sonars, missile guidance electronics, communication sets, electronic-warfare systems, night-vision devices. When the Navy needs a sonar for a new frigate or the Air Force needs a radar for a missile battery, BEL designs and builds it. The customer, overwhelmingly, is the Government of India — which also happens to own 51.14% of the company, appoint its board, and write the defence-procurement rules. Keep that sentence in mind; it is the strangest and most important fact in this chapter.
The science underneath
Charlie here. One paragraph of physics, as promised, because a radar is a beautifully simple idea executed with fiendish difficulty.
A radar shouts and listens. It fires a pulse of radio waves — light's longer-wavelength cousin, travelling at the same 3 lakh kilometres per second — and waits for an echo. Distance is just the stopwatch reading: time for the round trip, times the speed of light, divided by two. Direction comes from where you pointed; speed of the target comes from the Doppler shift, the same physics that makes a train horn drop in pitch as it passes you. Simple. The fiendish part: the echo returning from a small aircraft hundreds of kilometres away is unimaginably faint — the signal fades with the fourth power of distance, since it spreads out going and spreads again coming back. Doubling the range means detecting a signal one-sixteenth as strong. Pulling that whisper out of a screaming storm of noise and enemy jamming is where the decades of engineering live.
Now the chain to economics. Because the physics is unforgiving, a serious radar takes ten-odd years of accumulated design know-how, test ranges, and failure lessons — knowledge that lives in teams, not textbooks. Because the product is secret, the customer cannot buy it off a shelf abroad without strategic risk, and no country wants its air-defence source code held by a foreign vendor. So defence electronics everywhere on Earth collapses into a handful of trusted national champions. Science creates the difficulty; secrecy converts the difficulty into a licence.
But this castle has a peculiarity found almost nowhere else in commerce: it has exactly one paying visitor. The economics of one buyer even has a name — monopsony, monopoly's mirror twin. A monopolist sets prices because it's the only seller; a monopsonist sets prices because it's the only buyer. BEL cannot raise prices on a customer who writes the procurement rules — margins are, in effect, administered. And here the customer is also the majority owner and the regulator. The same hand signs the purchase order, pockets the dividend, and drafts the rulebook. When that hand is friendly, the arrangement is wonderful: guaranteed demand, patient contracts, protected turf. The moment national priorities shift, the same hand can squeeze margins, demand dividends, or open the gate to private rivals — and there is no court of appeal above your own majority shareholder.
The moat test
Hand a capable rival ₹3,11,031 crore and ten years of patience. Can they take the castle? Start with what money buys: factories, yes; engineers, some. What it cannot quickly buy: fifty years of classified design heritage, security clearances, test data from decades of field trials, and — decisive — the trust of a defence establishment that already owns the incumbent. A government choosing between its own company and a newcomer for a critical radar is not a neutral referee; it's the incumbent's parent. Add genuine switching costs: a radar is not a purchase but a thirty-year marriage of spares, upgrades and maintenance, and armed forces hate mixed fleets.
The honest cracks: India's "Make in India" defence policy is deliberately nurturing private competitors, and global majors partner with Indian firms for offsets. The moat is wide today. Its width is set by policy, and policy has no chemistry holding it in place — Charlie notes that a moat you rent from your customer is still a moat, but read the lease. Wide, with a landlord.
The numbers Warren would check
| What we check | Why it matters | BEL |
|---|---|---|
| ROCE | Return on all capital inside | 36.5% |
| Return on equity | Profit per ₹100 of owners' money | 27.6% |
| Borrowings Mar 2026 | Debt | ₹65 crore (essentially nil) |
| Sales growth, 10 yr | Demand trend | 14% a year |
| Profit growth, 5 yr | Earnings trend | 24% a year |
| Operating margin | 17% (FY15) → 29% (FY26) | rising |
| Dividend payout | Cash mailed to owners | ~30–46% range across the decade |
| Debtor days (Screener) | How fast the customer pays | 170 days |
| P/E | Price of admission | 51.3 |
Sales grew from ₹7,093 crore in March 2015 to ₹27,610 crore in March 2026; profit from ₹1,197 crore to ₹6,062 crore. The operating margin — profit from operations per ₹100 of sales — rose from 17% to 29%, remarkable for a price-administered business. The equity capital line jumped from ₹80 crore to ₹731 crore over the decade; don't mistake that for dilution — these were bonus issues, the company splitting the same cake into more slices for the same owners.
Now the wart, and it's a big one: the customer pays late. Screener flags debtors of 170 days — nearly six months of sales sitting as IOUs — and working-capital days rising from 79 to 135. You can see the consequence in cash from operations, which lurches from ₹5,093 crore (FY21) to ₹1,199 crore (FY23) to ₹4,659 crore (FY24) to ₹587 crore (FY25). The profits are real; the cash arrives on the government's calendar, not BEL's. A monopsonist doesn't just set your prices — it sets your payment terms, and there is nobody to complain to.
What could go wrong
Invert: what kills BEL? Not bankruptcy — you cannot bankrupt a debt-free company whose owner prints the currency. The threats are subtler. One: policy competition — the government deliberately handing slices of defence electronics to private players to build a wider industrial base; BEL's protected share of the pie shrinks even as the pie grows. Two: margin compression by decree — administered pricing can be re-administered; the same signature that let margins reach 29% can walk them back. Three: technological leapfrog — electronic warfare moves fast, and a state-owned enterprise's pay scales make it hard to hoard the best chip designers against private and global bidders. Four: peace, oddly — defence budgets follow threat perception. Five, and nearest: the price. At 51 times earnings and 13 times book value, the market has extrapolated a golden decade far forward. A wonderful business bought at a price that assumes perfection is how careful people get mediocre results.
What management must do to keep the castle
- Chase exports. A second, third and tenth customer is the only lasting cure for monopsony; every foreign order is a vote of confidence and a negotiating counterweight.
- Fight for payment terms. 170 debtor days quietly taxes the owners; press the customer-owner for milestone payments the way private vendors do.
- Out-spend on R&D within reason — the moat is know-how; the day a private rival's radar wins a competitive trial, the policy umbrella starts folding.
- Keep the balance sheet boringly debt-free and the dividend flowing; it disciplines capital allocation.
- Poach and retain top engineers even if it breaks public-sector pay convention; the castle is made of people who understand fourth-power signal loss.
The verdict
Wide moat — earned by physics and secrecy, fenced by policy, and owned by its only customer. The numbers are the prettiest in this book's neighbourhood: 36.5% return on capital, no debt, margins rising for a decade, profit compounding at 24% for five years. The peculiar risk is that the moat, the pricing, and the payment terms are all set by the same government that owns the majority stake — a benevolent arrangement today, an administered one forever. And at 51 times earnings, an owner is paying now for many years of continued benevolence. Dinner-table version: BEL is the only shop licensed to sell radar to a customer who owns the shop — a lovely franchise, currently priced as if the licence were carved in granite rather than written in policy.
Written in the style of Buffett & Munger for the Omaha Investments book project. Educational material, not investment advice. Numbers from Screener.in and live NSE data via Angel One as of the date above.