Digging Up the Street Is the Entire Investment Case

Two Numbers, Constantly Confused

Fiber operators report two numbers that sound almost identical and mean something completely different. Past houses Every address the network physically reaches counts meaning a technician could show up tomorrow and activate service. Connected homes count the addresses that actually pay for it. Divide the second by the first and you get penetration and penetration is the number that decides whether the last decade of digging was worth it

Here's the part that trips people up the first time they see it. The expensive part of this business is passing through not connecting. Once the trench is dug the conduit is buried and the fiber strand is spliced into a neighborhood adding one more paying customer costs a truck ride and a box on the side of the house. The big capital has already been spent whether no one signs up or every single one does. The first time I read a fiber operator's annual reportI found that asymmetry disorienting because almost nothing in stock research works this way. The cost of a software company increases roughly with usage. The cost of a retailer increases with units sold. The cost of a fiber network is charged almost entirely up front with revenue coming in later at the rate the neighborhood decides to hire

What It Actually Costs to Dig

The construction cost per approved home sometimes reduced to cost per approved premises in industrial shorthand varies greatly by method and geography. Hanging fiber from existing utility poles called aerial deployment is much cheaper than drilling or trenching under the pavement called underground deployment because the crew is connecting the cable to infrastructure that already exists rather than opening a street. Density also helps. A crew running forty houses per mile of cable in a dense suburb is spreading the same costof trenching in four times the directions of a crew working on a rural route with ten houses per mile

The industry tends to talk about a range from several hundred dollars per house on a favorable overhead installation to a thousand dollars on a difficult underground one plus an additional installation cost for each house that actually signs up. What surprised me when I looked at some of these builds is that excavation is often not the bottleneck at all. The negotiations over connecting poles the legal process for getting permission to hang new wires on poles owned by someone else the deadlines set with local government and something the industry calls doingBefore the job is done which means rearranging existing cables and equipment on a pole to make room for the new line they can stall a construction for months before a single foot of fiber reaches the ground

Penetration Does Almost All the Work

Because the cost is fixed at the step the return is close to linear penetration and it is a kind of relentless linearity

Doubling penetration from 15 percent to 30 percent doesn't just double revenue. It roughly halves the effective capital invested per paying customer because the same ditch and buried glass now support twice as many bills. That's why operators obsess over what the industry calls the penetration ramp the rate at which a newly built neighborhood fills up after the crews leave and why marketing budgets are concentrated in one zip code in the early days.months rather than spreading out evenly over years. A cohort that grows slowly tends to stay slow because the households most eager to leave the incumbent cable provider typically convert in the first month. Anyone who left after that is a harder sell

A Worked Example: Turning Take Rate Into a Payback Period

Let me specify this with round and clearly illustrative numbers not with actual numbers from any company

Suppose a fiber builder passes houses in a medium-density suburb at a cost of $900 per house passed about half the band described above. Let's call it take rate another name for penetration 35 percent. That means that for every 100 homes the team connects 35 homes eventually become paying customers

The total construction capital for those 100 homes is 100 times $900 or $90,000. Add an installation cost of $150 for each of the 35 homes that actually connect since someone has to run a drop line to the house and install equipment inside. That's 35 times $150 or $5,250. The total capital deployed across the entire cohort is90,000 plus 5,250 which equals $95,250

Divide that by the 35 paying subscribers and the effective cost of capital per connected customer will be 95,250 divided by 35 which is approximately $2,721. Each subscriber who eventually signs up has approximately $2,721 of capital buried under their backs whether they know it or not

Now the revenue side. The average monthly revenue per subscriber is $75 a reasonable illustrative figure for a fiber broadband plan. Call the gross margin on that revenue i.e. what's left after the direct cost of serving the customer such as network operations and customer support 80 percent which is plausible because the marginal cost of sending more bits over fiber that's already built and paid for is genuinely small. The annual gross profit per subscriber is $75.multiplied by 12 months that is $900 multiplied by 80 percent which gives $720 per year

The payback period is capital per subscriber divided by annual gross profit: 2,721 divided by 720 which is about 3.8 years. Call that roughly four years of bill collection before the cost of building the cohort is paid. Every dollar after that is close to being a pure contribution since the trench is already dug and will continue to be dug for decades

How Sensitive Is Payback to Take Rate

Here's why penetration isn't just one input among several. It's effectively the entire investment case. Keep the $900 transfer cost and $150 setup cost fixed and only move the take rate

If you move the acquisition rate from 20 percent to 50 percent less than a factor of three the payback period goes from about six and a half years to less than three a swing of almost four years on an otherwise identical buried asset.The industry spends much more time modeling marketing conversion curves and competitive win rates in the first eighteen months after a build than it does modeling the construction bill itself. The cost of construction is a known number. The acquisition rate is a bet on human behavior and is the bet on which all performance depends

Why a Second Crew Rarely Shows Up

This is also why digging up the street takes up the entire moat not just the total cost. Once a fiber network is built and reasonably penetrated a second company looking at that same street runs into a brutal problem: The neighborhood can only support so much total broadband spending and a second network doesn't grow that fund it splits it up. Two fiber operators serving a street with a combined 40 percent penetration typically don't reach a stable truce in the end.20 percent each. Price becomes the weapon since the product itself symmetrical fiber speeds is difficult to differentiate once both sides have it and neither side can walk away because the trench is already sunk. That's the natural logic of monopoly that infrastructure investors love: It's not that competition is legally blocked but that the economics of a second entrant digging up the same street rarely works once someone got there first

Fixed wireless access that is broadband delivered over the spectrum of a cellular network instead of a cable adds an extra layer of competition at the low end without anyone doing any research since the operator is repurposing towers it already built for phones. I'll come back to that because it's a real crack in this thesis and not a minor one

Case Study: Google Fiber Learns the Hard Way

Google Fiber is the case study I keep coming back to because it's a company with essentially unlimited capital that still stumbled headlong into this arithmetic. Google announced its fiber-to-the-home effort in 2010 and Kansas City became the first city with live service in 2012. Google was upfront about the ambition: gigabit speeds comparatively low prices and a plan to expand city by city. For a few years it genuinely looked like a threat to traditional cable operators in every market it served.mentioned

Then starting in 2016 Google Fiber pulled out. It halted planned launches in several announced metropolitan areas cut a significant portion of its staff and its leadership at the time left. What emerged from that period was a fairly candid recognition that trenching an entire city block by block was slower and more expensive than the original plan assumed and that penetration in some already built-up neighborhoods was increasing more slowly than expected once the initial wave of enthusiastic changes had already subsided.had signed up. Parent company Alphabet shifted some strategy toward technologies that avoid the trench almost entirely including wireless approaches through its acquisition of Webpass precisely because excavating an existing street beneath an existing city with all its permits and friction with poles turned out to be a much more difficult and time-consuming business than building in a greenfield suburb

I read Google Fiber as proof that this is a construction and penetration problem rather than a technology problem. Google had the engineering talent and balance to build a truly great network. It ran into the same wall every fiber overbuilder runs into: the cost and calendar time of physically excavating an existing city and the uncertainty of how many homes would actually be moved once the crews leave. Money alone doesn't make a street easier to dig

Subsidy Money Redraws the Map

Rural territory generally fails this arithmetic on its own merits and no amount of smart financing changes that if the underlying inputs remain the same. There simply aren't enough homes per mile of ditch for a realistic acquisition rate to erase the recovery calculations mentioned above. Public subsidy programs exist to change that equation directly: a government pays part of the construction cost up front in exchange for a coverage commitment which takes territory that doesn't pass the recovery test and turns it into passing territory

The practical implication for anyone reading a fiber operator's disclosures is that announced construction targets within a subsidized footprint should be compared to the terms of the subsidy not the ordinary unit economics described above. A change in how a subsidy program is structured who qualifies how much they pay per household what speed it requires can substantially expand or shrink a company's footprint without changing anything in the underlying technology or the underlying demand of households

Where the Moat Cracks

I have argued that trenching a street is almost an unbeatable ditch. Let me honestly argue the other side because each of them is a real failure mode not a hypothetical one

The first opportunity is wireless substitution. Fixed wireless access delivered through a cellular operator's existing towers and low-Earth orbit satellite service commercially led by Starlink deliver broadband to a home without anyone touching that home's street. Neither match symmetrical fiber speeds at the top end and satellite in particular still lags in latency and capacity per user. But for a home that primarily wants reliable broadband at a decent price and doesn't mindIt matters especially whether the speed test says 300 or 1000 fixed wireless is a genuine substitute and shows up in a fiber operator's numbers as a household that simply never converts. Every acquisition rate assumption I discussed above silently assumes that the household's only realistic alternative is the existing cable operator. That assumption weakens every year that wireless technology improves

The second crack is that overbuilders show up more often than the natural monopoly story allows. AT&T for example has been overbuilding parts of its own legacy copper footprint with fiber in territories where a traditional cable operator already had a network in part because a large incumbent can sometimes absorb years of split penetration that would sink a smaller more leveraged entrant. The math above assumes that a rational competitor stays off a busy street. It doesn't hold up.If a competitor has a strategic reason defend a national footprint match a rival city by city which outweighs the independent economy of that street

The third crack is regulatory and it's the one I consider most underrated. Some jurisdictions require the owner of last-mile infrastructure to rent access to competitors at a regulated wholesale rate an arrangement often called open access. Utah's UTOPIA Fiber network is a real-life example built on this model from the beginning: the network operator builds and maintains the physical fiber and multiple retail Internet providers sell services through the same buried glass competing on price and support rather than infrastructure. If regulationmoves a normally closed network toward that model after the fact the company that dug the trench may end up sharing the same asset that was not supposed to be copied

How I Actually Read a Fiber Filing

The way I'd actually use this if I were sitting through a fiber operator's presentations: I skip the subscriber growth number almost entirely because a company can grow subscribers simply by going through more homes and that alone says nothing about whether a given cohort of construction spending is working. I go straight to penetration by era that is penetration is broken down by the time since each neighborhood was built when the company discloses it. A three-year-old footprint stuck at one penetration20 percent is a much bigger problem than a footprint from six months ago at 20 percent and a combined headline figure between the two will mask that difference entirely

Then I want to know the take rate assumption for new build underwriting and I test it the same way I tested the payback table above. If a company's plan only works at 40 percent penetration and a competitor announces plans to build the same footprint I treat that build as something to watch closely rather than one to model to the letter because the risk of overbuilding shows up in the take rate line before it shows up anywhere else. My takeThe honest truth is that this sector rewards patience and punishes anyone who chases current quarter subscriber additions and I say this as someone who initially found the entire decade-long capital-heavy payback structure a little boring next to a fast-growing software story. It took me a while to realize that boring bond-like cash flow is exactly the goal for infrastructure funds to buy these deals. They're not looking for the next high-growth story. They're looking for a buried asset that will allow thempay a predictable check for twenty years and they will pay a premium for the clumsiness that it took me a while to respect

The Bottom Line

Fiber economics boils down to a structure that is easy to formulate and brutally difficult to execute well: spending a large lump sum to dig up a street once and then collecting a high-margin annuity from whichever part of that street decides to subscribe to it. I reviewed a practical example where a cost of $900 per home was approved and a payback rate of 35 percent produced approximately a payback in 3.8 years and where moving that same payback rate from 20 to50 percent reduced the payback from about six and a half years to less than three. That sensitivity is the entire investment case in a sentence. The real risk is overbuilding not obsolescence because two networks on one street divide a fixed group of willing customers rather than adding to it. Google Fiber's withdrawal from block-by-block digging shows that even a company with essentially unlimited capital can't avoid the problem of construction and penetration. And the moat has real cracks: fixed networksWireless and satellite miss the ditch entirely overbuilders sometimes come in anyway for strategic reasons and open access regulation can force the digger to share the same asset that was supposed to be exclusive. Anyone reading a fiber operator's disclosures should go straight to penetration by cohort and skip the core subscriber count because that's where this whole gamble pays off or doesn't pay off