Telecom: ARPU, EBITDA after leases, and the limits of sector lenses

Building a network, renting it by the month

A mobile operator spends enormous sums up front — spectrum licences, towers, fibre, radio equipment — and then rents access to that network, month by month, to hundreds of millions of customers. Most of its costs are fixed; most of its revenue is small and recurring.

That shapes the lens:

• Revenue is customers × ARPU (average revenue per user). Growth comes from more customers, or from each paying more — through tariff increases or by moving to richer plans (postpaid, more data).
• EBITDA margins look huge, because the biggest cost — the network — sits mostly in depreciation, amortisation and interest, below EBITDA (earnings before interest, tax, depreciation and amortisation).
• Capex never stops. Networks have to be upgraded (4G to 5G) and expanded, so a large share of EBITDA goes straight back into the network.

The formulas

Mobile revenue    =  average mobile customers  ×  ARPU  ×  months in the period
ARPU              =  mobile revenue  /  average customers  /  months

Growth split      ≈  (1 + customer growth) × (1 + ARPU growth) − 1

EBITDAaL          =  EBITDA after leases
                     (EBITDA minus the depreciation and interest on leased
                      assets such as tower sites)

Capex intensity   =  capex  /  revenue

Why “after leases”? Under the lease accounting standard (Ind AS 116 — Indian Accounting Standard 116), rentals on leased assets — for an operator, largely tower and site rentals — are no longer shown as operating expenses. They reappear as depreciation of a “right-of-use” asset plus interest on a lease liability, both below EBITDA. So reported EBITDA excludes a real, recurring cash cost. EBITDAaL puts it back, which makes it closer to the old, pre-standard EBITDA.

Worked example: Bharti Airtel, quarter and year to March 2026

From Airtel’s media release of 13 May 2026 (consolidated, Ind AS). ₹ crore unless noted. Historical, for illustration only.

Step 1: ARPU and customers

1. Mobile ARPU in India was ₹257 a month in the March 2026 quarter, against ₹245 a year earlier — up 4.9%.
2. India mobile revenue grew 8.3% over the same period. Using the growth split, that implies average mobile customers grew roughly 3.2%. So more than half of the mobile revenue growth came from each customer paying more, and the rest from more customers. (Rough: ARPU is rounded to the rupee and quarters differ in length.)
3. ARPU rises for more than one reason. The release credits “portfolio premiumization” — 80% of mobile customers now use smartphone data, and postpaid customers reached 29 million — as well as higher realisations. Mix and price both push ARPU.

Step 2: EBITDA, EBITDAaL, EBIT

Item	Q4 FY26	Share of revenue
Revenue	55,383	100%
EBITDA	32,038	57.8%
EBITDAaL (after leases)	28,647	51.7%
EBIT (earnings before interest and tax)	18,156	32.8%
Capex	16,066	29%

1. Lease costs alone — the gap between EBITDA and EBITDAaL — were ₹3,391 crore, about 6.1% of the quarter’s revenue.
2. For the full year, EBITDA was ₹1,21,268 crore and EBIT ₹68,100 crore. The difference — mostly depreciation and amortisation of the network and spectrum — was about ₹53,168 crore, or 25.2% of revenue. An EBITDA margin of 57.5% became an EBIT margin of 32.3%.
3. Capex in the quarter was ₹16,066 crore — about 29% of revenue and 50.1% of EBITDA. Roughly half of every rupee of EBITDA went back into the network that quarter, before interest, tax, lease payments or spectrum dues.

The release gives two leverage figures: net debt to EBITDA of 1.29 times (from 1.86 a year earlier), and net debt excluding leases to EBITDAaL of 0.79 times. By implication the first keeps lease liabilities in the debt and lease costs out of EBITDA, and the second takes them out of both — each pair consistent, which is the point: see net debt/EBITDA and never mix the two.

None of this is a view on Airtel. It’s one year’s release, read through the sector’s metrics.

🧒 Explain it like I'm 10 (optional — skip if this is already clear)

You build a giant water slide for ₹10 lakh. Each kid pays ₹100 a month to use it as often as they like.

Your “average revenue per user” is ₹100. You earn more if more kids sign up, or if you get kids to pay ₹150 for the fast lane.

The monthly money looks like almost pure profit — the slide is already built. But slides wear out, and next year everyone wants a taller one. If you forget that, you’ll think you’re rich right up until the slide needs replacing.

Common mistakes

• Reading EBITDA margin without capex and leases. A 57.5% EBITDA margin isn’t three times as good as Britannia’s 17.8% (FY25). Subtract leases, depreciation and ongoing capex first.
• Treating every ARPU increase as a price increase. Customers moving to postpaid or richer data plans lift ARPU with no tariff change. Read the company’s explanation of mix before assuming pricing power.
• Comparing ARPU across companies without checking the definition. Operators can count customers differently (for example, all SIMs versus active ones), which changes ARPU without changing revenue. Compare each company’s ARPU with its own history first.
• Mixing lease treatments in leverage ratios. Net debt with leases against EBITDA, or net debt without leases against EBITDAaL — never one of each.

What sector lenses can’t do

This module has used one real filing per sector to show that the same ratio means different things in different businesses. It’s worth being just as clear about what a sector lens can’t tell you.

• Most sector metrics are the company’s own. Utilisation, underlying volume growth, cement profit per tonne, value of new business, ARPU — each is defined and calculated by the company, and many sit outside the audited financial statements. They’re useful, but they deserve the same scepticism as any adjusted number. The red-flags post still applies.
• They describe one past year. Every example here is backward-looking. A strong ARPU year says nothing certain about next year’s tariffs, competition or regulation.
• They say nothing about price. This module deliberately never asked whether any of these companies is cheap or expensive. A lens explains the business; it doesn’t value it.
• Businesses don’t stay in one sector. A telecom company adding data centres and a lending arm, an FMCG company building foods and digital brands, a pharma company buying a consumer-health business — each needs more than one lens, applied segment by segment.
• The metric that matters can change. A new regulation, a technology shift or a price war can make last decade’s key number less relevant. The lens is a starting checklist, not a permanent one.

Takeaway: In telecom, revenue is customers times ARPU, and a big EBITDA margin has to pay for leases, depreciation and a network that’s never finished. Across every sector, the lens explains how a business makes money — it doesn’t audit the numbers, predict next year, or tell you what the company is worth.