Fibre is built per home passed and paid for per home connected, so penetration decides whether the first spend is ever repaid.
Penetration is the share of homes passed that become paying subscribers, and the rate a fibre-to-the-home network needs is found by solving its cash flows for the target return. On an illustrative 50,000-home build at 750 per home passed, with 600 per connection and 45 a month of revenue per subscriber, 10 per cent unlevered needs 49.2 per cent penetration. At the 40 per cent the plan assumes, the network earns 7.43 per cent.
Worked in full in The Digital Infrastructure Investor by Julian R. Sterling, with every figure reproduced in a free workbook.See the book on Amazon →
An FTTH network spends twice. It spends once to pass every home in the footprint, whether or not anyone buys, and again to connect each home that does. Revenue arrives only with the second. Penetration therefore decides whether the first spend is ever repaid, and the cost per home passed decides how much penetration is needed. The case is illustrative: unlevered, pre-tax, a 25-year life and no terminal value unless stated.
| Input | Value |
|---|---|
| Homes passed | 50,000 |
| Cost per home passed | 750 |
| Build cost, millions | 37.5 |
| Cost per connection | 600 |
| Revenue per subscriber, per month (ARPU) | 45.0 |
| Variable operating cost per subscriber, per month | 15.0 |
| Fixed network operating cost, millions a year | 1.5 |
| Maintenance capex, share of build cost a year | 1.0% |
| Years to reach terminal penetration, linear ramp | 5 |
| Planned terminal penetration | 40% |
Subscriberst = homes passed × terminal penetration × min(t / 5, 1)
EBITDAt = subscribers × (ARPU − variable cost) × 12 − fixed operating cost
Free cash flowt = EBITDA − new subscribers × connection cost − maintenance capex
Required penetration = the terminal penetration at which IRR(−build cost, FCF1 … FCF25) = 10%
In Excel: put terminal penetration in one input cell, compute =IRR(C20:AB20) on the free cash flow row, and use Goal Seek to set the IRR to 10% by changing the penetration cell.
| Year | Subscribers | Revenue | EBITDA | Connection capex | Free cash flow |
|---|---|---|---|---|---|
| 1 | 4,000 | 2.16 | -0.06 | 2.40 | -2.83 |
| 2 | 8,000 | 4.32 | 1.38 | 2.40 | -1.39 |
| 3 | 12,000 | 6.48 | 2.82 | 2.40 | 0.05 |
| 4 | 16,000 | 8.64 | 4.26 | 2.40 | 1.48 |
| 5 | 20,000 | 10.80 | 5.70 | 2.40 | 2.93 |
| 6 to 25 | 20,000 | 10.80 | 5.70 | 0.00 | 5.33 |
The build cost of 37.5 is spent at year 0 and the network does not cover its own cash costs until year 3. Connection capex adds 12.00 over the ramp, and from year 6 the network throws off 5.33 a year. The IRR on that profile is 7.43 per cent. Solving for 10 per cent gives a terminal penetration of 49.2 per cent, or 24,612 subscribers; break-even at a zero return needs 22.0 per cent, and an 8 per cent return needs 41.9.
| Terminal penetration | IRR |
|---|---|
| 25% | 1.66% |
| 30% | 3.92% |
| 35% | 5.80% |
| 40% | 7.43% |
| 45% | 8.88% |
| 50% | 10.20% |
| Case | IRR at 40% | Penetration for 10% |
|---|---|---|
| Cost per home passed 600 | 9.41% | 41.8% |
| Base: 750 per home, ARPU 45.0 | 7.43% | 49.2% |
| Cost per home passed 900 | 5.86% | 56.6% |
| ARPU 40.0 | 4.88% | 61.6% |
| ARPU 50.0 | 9.68% | 41.0% |
| Ramp of 8 years instead of 5 | 6.00% | 57.1% |
| Terminal value of 10.0x final-year EBITDA | 8.98% | 43.9% |
Every 150 of cost per home passed moves the required penetration by about 7.4 points. Five units of ARPU move it by more, and the ramp matters as much as the destination: reaching 40 per cent in eight years instead of five costs 1.43 points of IRR. A terminal value helps, but even at 10.0 times final-year EBITDA the plan's 40 per cent does not reach 10.
Penetration is the hinge because the build cost is spent per home passed and recovered per home connected. At 40 per cent, each subscriber carries the build cost of 2.5 homes, 1,875, before its own 600 connection.
Turning the question round changes the diligence. An IRR at an assumed penetration invites an argument about the IRR; a required penetration invites the right argument, about whether the market can deliver it. Three checks follow directly from the 49.2 per cent:
Leaving out connection capex, or treating it as opex in a single year. It is incurred per subscriber as the network fills, so it scales with the penetration it is meant to test. Ignore it and the plan shows 9.69 per cent at 40 per cent penetration and needs only 40.8 per cent for 10, a target that looks within reach of the plan when it is not. The second mistake is quoting penetration on homes passed in the whole footprint while the cost per home passed is quoted on a narrower, cheaper first phase.
Model the two spends separately, solve the cash flow for the target return, and report the answer as a penetration rather than an IRR at an assumed one: 49.2 per cent for 10 per cent here, against a plan of 40. The book's Oakmere Fiber network works the same question on a half-built network with penetration by cohort and churn; its grid of penetration against cost per home passed is in the free workbook for this book. For how a tower portfolio's return turns on a comparable occupancy measure, see how the tenancy ratio drives cell tower returns.
There is no single benchmark; the rate that matters is the one the cost per home passed requires. In the illustrative case 22.0 per cent only returns the capital, 41.9 per cent earns 8 per cent and 49.2 per cent earns 10. At 600 per home passed the 10 per cent threshold falls to 41.8.
It is spent on every home in the footprint, so it must be recovered from the subscribers alone. In the illustrative network, raising it from 750 to 900 cuts the IRR at 40 per cent penetration from 7.43 to 5.86 per cent and lifts the penetration needed for 10 per cent from 49.2 to 56.6.
Yes, as capex incurred per new subscriber, because they scale with take-up. Omitting them in the illustrative case overstates the IRR at 40 per cent penetration at 9.69 instead of 7.43 per cent and understates the penetration needed for 10 per cent at 40.8 instead of 49.2.
This article is one calculation from The Digital Infrastructure Investor. The book takes the same case from first principles to the decision, chapter by chapter, and every figure it prints is a live formula in the free companion workbooks.
Get the book on Amazon →Free companion files
Also on Amazon UK · Amazon Germany · Amazon France · Amazon Canada
Reading guide: infrastructure, data centres and energy → · All 324 articles →
If this book helped, or didn’t, a few lines on Amazon are worth more than they look: they are what the next reader goes on. Write a review. The workbook stays free either way.