ARPU tells you if monetisation is actually working.
Telecom builds a hugely expensive network of towers and spectrum, then earns a small amount from each subscriber every month. Once the network is built, every extra rupee of revenue is almost pure profit, so the whole game is filling the pipe and charging more per user. Revenue per user is the honest scoreboard: adding cheap subscribers while that number stalls is just competitive bleeding. Spectrum auctions are brutal cash outflows that can starve a player of the money it needs to keep the network good.
Airwaves are bought at auction, towers turn them into a network, the network fills with subscribers, and each subscriber pays a small amount every month. Then comes the catch: most of that money goes straight back into the next round of spectrum and the next generation of network before the owner sees any of it. The whole investment case is whether anything escapes the loop.
Rising ARPU, self-funded capex, a survivable balance sheet and quality subscribers. In telecom that short list separates the one or two investable companies from the rest of the industry.
Answer these six and you have understood the industry. The rest of the page is the detail behind them.
Strip away the jargon and a telco is a landlord of invisible pipes. It spends enormous sums up front to build a network, then rents access to it, one subscriber at a time, for a monthly fee.
Building towers and buying spectrum costs tens of thousands of crores before a single customer connects. Once that network exists, serving one more subscriber costs almost nothing extra, so the business has huge operating leverage. That is why market share swings hit profit so hard in both directions, a subscriber gained or lost falls almost straight to the bottom line. This is also why the industry consolidates into a handful of players rather than staying fragmented like a grocery store chain.
A telco does not sell you a phone call, it sells you a subscription that renews every month whether you think about it or not. That means the real product is retention: keeping a customer paying long after the network cost is sunk. A customer who stays 5 years is worth many times a customer who leaves after 5 months, because the acquisition cost (SIM, marketing, dealer commission) is paid once but recovered slowly.
Here is the paradox worth sitting with before anything else. Telecom sells one of the most dependable products in any economy: a mobile connection is now as basic as electricity, people pay through good times and bad, and each user consumes more data every year. Wonderful demand, terrible history. Indian telecom has destroyed enormous amounts of shareholder money despite everyone using its product all day, every day. The rest of this page explains that paradox, and the explanation transfers to every industry: dependable demand is worth nothing to the owner if pricing is broken and the cash never stops leaving. Demand pays the customer's bill. Structure decides who keeps the money.
Every industry on this site has one number that tells you most of the truth. In telecom it is ARPU, and it deserves its own section because everything else hangs off it.
Average Revenue Per User is simply monthly revenue divided by subscriber count: how much money the company collects per customer per month. It is the cleanest read on whether a telco is winning quality customers at fair prices or buying cheap subscribers with discounts. Rising ARPU on a stable base means real pricing power at work. Huge subscriber growth with flat ARPU means the company is purchasing market share and calling it growth. Subscriber counts flatter easily; the average revenue each one pays does not.
Because the product renews monthly, ARPU is reported every single quarter, which makes it a running commentary on the company's power to price. When it falls for quarters on end, the message is unambiguous: customers are being kept only by charging them less, and the business is shrinking per head even as the subscriber line grows. This is the single most important thing to watch, and it is handed to you in every results presentation. No modelling required.
A telco lifts ARPU three ways. Tariff hikes, which only stick when competitors follow, which is why they only happen once an industry has consolidated. Mix, moving customers from prepaid to postpaid and from basic packs to bigger data bundles, postpaid customers pay more and leave less. And attach, layering extras onto the same connection: streaming bundles, cloud, enterprise lines, home broadband. The first depends on the industry structure. The second and third are the company's own work, and they are how a good operator grows revenue even between tariff hikes.
Now the cost side, and the reason this industry eats cash. A telecom network is not built once. It is built forever.
Spectrum, the slices of airwaves that carry every call and byte, is the industry's scarce resource. The government owns it and sells it through auctions, and because no telco can operate without it, auctions become forced purchases at fearsome prices: tens of thousands of crores, recurring every few years, usually funded with debt. Think of it as rent for the air, paid in advance, in lump sums large enough to bend a balance sheet. When you hear that a telco is cash-rich going into an auction year, wait for the auction before believing it.
Every technology generation, 3G to 4G to 5G, forces telcos to rebuild the network at enormous cost, and here is the trap: customers do not pay more just because the technology changed. The new capacity earns its keep only years later, once usage grows into it and tariffs catch up. So a telco's reported profits right after a big capex cycle understate the business, and its profits late in a cycle, when the network is sweated and paid for, overstate what the next cycle will allow. Investors who buy right after the spending, before the payback, are paying for a network that has not yet earned anything.
Put the auctions and the upgrades together and you get the defining test for any telco: put cash generated from operations next to purchase of fixed assets plus spectrum payments, side by side, for five years running. In a healthy business the first line comfortably exceeds the second and the surplus reaches the owner. In most of telecom's history, the second line kept swallowing the first, so however fast revenue grew, nothing was left over. This is arithmetic, not opinion, and it sits in every annual report's cash-flow statement waiting to be read.
In 2016 India ran the most brutal natural experiment in telecom economics anywhere in the world. Understanding it explains the entire industry you see today.
In a high fixed-cost business, a new entrant with deep pockets can price below everyone's cost, because its own cost per user falls as users pile in, and its losses are an investment in market share. Jio launched in 2016 with free voice and data cheaper than anyone could match. Every incumbent had the same choice: match the prices and bleed, or hold prices and watch subscribers walk. Everyone matched. Industry revenue per user roughly halved, and a dozen operators became three. Fixed-cost industries do not stay crowded; a price war is how they thin out, and the entrant with the strongest balance sheet writes the rules.
Here is the transferable lesson: in telecom, pricing power does not come from the product, which is identical across operators, it comes from how many players are left. A crowded market prices at the floor. Once three players remain and the weakest can no longer afford to fight, the survivors can finally raise tariffs together, and every hike falls almost straight to profit because the network is already paid for. This is why the investment case for Indian telecom flipped after 2019: the same companies, the same towers, but a structure that finally allowed prices to rise.
A three-player market with one player on life support behaves like a two-and-a-half player market. The weak operator cannot invest in its network, so it loses its best subscribers each quarter; it cannot let go of price, so it cannot rebuild ARPU; and its debt keeps compounding regardless. It survives only as long as lenders and the government prefer delay to burial. For an investor, this is the sharpest sorting question in the sector: are you buying the consolidation winners, or the company the consolidation happened to?
A telco's biggest costs and its right to operate both come from the government. That makes regulation not a background risk but a first-order one, and one judgement in 2019 proved it.
Indian operators pay the government a share of their adjusted gross revenue, AGR, as licence and spectrum fees. For over a decade, the operators and the government disputed what counted as revenue: only telecom income, or everything including interest and asset sales. In October 2019 the Supreme Court ruled for the widest definition, retrospectively, with interest and penalties compounding on 14 years of dues. The bills landed in the tens of thousands of crores, for years the companies had already closed. Airtel booked a provision of about ₹28,450 crore and reported a record quarterly loss. Nothing about the network, the customers or the tariffs had changed. The rules had.
The lesson is not that telecom is uninvestable; it is that in any industry where the government sets the biggest costs, the balance sheet must be judged against the regulator's power, not just the market's. A strong operator absorbs a rule change, raises capital, and carries on. A leveraged one is pushed to the edge by the same ruling. Before buying any telco, ask what an adverse ruling or a surprise levy would do to this specific balance sheet, because the sector's history says such things arrive roughly once a decade. The same test applies to power, banking, and every other licensed industry on this site.
The failures in this industry follow one script closely enough that you can learn it once and recognise it anywhere.
Spectrum and networks are bought with debt, and debt does not shrink when a rival cuts prices. A telco caught with heavy borrowings in a price war faces a closing trap: it cannot cut capex without losing customers to the better network, and it cannot keep spending without drowning in interest. ARPU falls, the debt stays, and the gap between what the business earns and what it owes widens every quarter until lenders decide the story is over. Note what the killer is: not losing customers, but carrying obligations sized for prices that no longer exist.
Airtel made the classic error too. In 2010 it paid $10.7 billion, borrowed, for Zain's African operations, and net debt swelled toward ₹60,000 crore. From that day, the first claim on every rupee belonged to lenders, and the stock went nowhere for a decade. So why did Airtel survive what killed Jet Airways, the same debt-funded acquisition mistake? Because of what the debt sat on. Jet's acquisition added losses to a business with no pricing power and evaporating revenue. Airtel's sat on subscriptions that renewed every month through the entire ordeal; the revenue never fell away, so the debt could be serviced, however painfully, until consolidation restored pricing. The transferable lesson: the same mistake is survivable on recurring revenue and fatal on cyclical revenue. Judge the error by what carries it.
Telecom valuation has two rules that beginners skip, and skipping either one leads to buying the wrong company at the wrong time.
Because telcos carry enormous and unequal debt, market capitalisation alone is meaningless for comparison. Always work with enterprise value, market cap plus net debt, because that is the real price of the business including the obligations you inherit. The standard yardstick is EV against EBITDA, the operating profit the network throws off. Two telcos with the same market cap can be wildly different purchases once the debt is counted; one of them, you are mostly buying the lenders' problem.
EBITDA in telecom flatters, because it ignores precisely the two things that eat this industry alive: capex and spectrum. The number that matters is what remains of operating cash after both, free cash flow after spectrum, because that is the money that can actually reach you. A telco can report fat EBITDA margins for a decade while this number stays at or below zero, and its shareholders will have owned a growing network and earned nothing. Pay for telecom cash flows only when the structure, consolidation, rational pricing, a funded network, lets them actually exist.
Every sector reduces to a demand metric, a pricing metric, an efficiency metric, a capital metric, and a risk metric. For telecom, these are the ones that matter.
What each number tells you, and how to read it.
| Metric | Why it matters |
|---|---|
| ARPU (Avg Revenue Per User) | Pricing power per subscriber. Rising ARPU means monetisation works; flat ARPU in a growing base means competitive pressure. |
| Subscriber Growth | Scale, which brings fixed-cost leverage. Quality (postpaid versus prepaid) matters as much as quantity. |
| Churn | Customer loyalty. Low churn signals a sticky network effect and better coverage or brand. |
| Data Consumption / User | Monetisation potential. Higher data usage is the path to higher ARPU tiers and add-on services. |
| Spectrum Costs | Capital intensity. Auctions are massive cash outflows; heavy spectrum debt constrains network capex. |
| EBITDA Margin | Efficiency on fixed infrastructure. Telcos should post 40-50% EBITDA margin at scale. |
| Net Debt / EBITDA | Survival gauge. Spectrum is bought with debt; this ratio says whether the next auction or ruling is absorbable or fatal. |
| FCF after Spectrum | The only honest profit in telecom: operating cash minus capex minus spectrum. Until this is positive, owners are funding the network, not the reverse. |
Telecom is a fixed-cost machine that rewards its owners only once consolidation gives the survivors the power to raise prices.