Why Your Internet Bill Is What It Is: The Last-Mile Monopoly Problem
The internet's backbone is fiercely competitive and absurdly cheap. The wire to your house is neither. How the economics of the "last mile" decide what you pay, why a third of Americans have one real choice or none, and what $42 billion is trying to change.
Evergreen explainer. Figures are checked against the primary sources listed at the end. Corrections policy.
Here's a pricing puzzle hiding in plain sight on your monthly statement. The actual internet — the global mesh of fiber backbones and exchange points that moves your data across continents — is one of the most brutally competitive, rapidly deflating markets ever created. Wholesale long-haul bandwidth gets cheaper nearly every year. And yet the bill for reaching that internet from your living room mostly goes one direction, and it isn't down.
The resolution of the puzzle is geography. Your payment doesn't really buy the internet. It buys the last mile: the specific physical wire — coax, fiber, or copper — running from a neighborhood node to your home. And the last mile has economics unlike almost anything else you purchase.
The last mile is a construction project wearing a tech costume
Strip away the branding and an internet service provider is mostly a trenching-and-cabling company. To serve your house, someone had to dig up streets or string cable along poles, past every single address in your neighborhood, whether or not those addresses ever subscribe. Industry and government cost estimates for fiber construction run from several hundred dollars per home passed in dense suburbs to many thousands in the rural expanse — before a single customer pays a dime.
That cost structure produces the defining feature of the market: the first network to a neighborhood is hard to justify, and the second is usually irrational. If building costs, say, $1,000 per home and half of homes subscribe, the first entrant spreads $2,000 of construction across each customer and can recover it over years of bills. A second entrant faces the same construction bill but can hope to win only a fraction of the homes — while its arrival cuts prices for everyone, shrinking the prize it's fighting for. Economists call this a natural-monopoly tendency; it's the same logic that gives you exactly one water main and one electric wire, and it's a cousin of the huge-fixed-cost math that shapes highways and rail lines.
America's twist is that we ended up with not one wire but, in much of the country, two — and that history explains the current map. The telephone network and the cable-TV network were built separately, decades apart, for different purposes, each under its own regulatory deal (phone companies as regulated utilities; cable operators under municipal franchise agreements). When both were retrofitted for broadband, most metropolitan addresses inherited a duopoly: the phone company's line versus the cable company's line. For years cable's coax simply outran DSL on speed, which in practice made many "two-provider" markets one-fast-provider markets.
What the referee's scorecard says
The FCC is required to report periodically on the state of competition, and its own numbers are the cleanest picture of how much choice actually exists. Per the 2024 Communications Marketplace Report, measuring fixed broadband at the 100/20 Mbps benchmark (100 megabits down, 20 up — the FCC's current definition of baseline broadband):
- About 66% of households nationally could choose between two or more providers as of the end of 2023.
- In urban areas, roughly 75% of households had at least two options; in rural areas, about 35%; in Tribal areas, about 49%.
- Put the other way: more than a third of the country has one provider or none at that speed tier.
How many Americans actually get a choice
Households with 2+ fixed providers at 100/20 Mbps, end of 2023 — %
Competition scholars will note that "two providers" is still thin — in most markets two players compete on promotions more than on list price. The economics point is subtler than "monopolists gouge": even a well-behaved last-mile operator prices to recover enormous sunk construction costs from a limited subscriber base, plus ongoing network upkeep, plus content and transit costs, plus a margin its investors demand for having risked the trench. Where a rival wire exists, promotional pricing appears; where none does, the list price simply is the price. Your bill is a map coordinate.
It's also why your bill behaves the way it does over time: a low promotional rate (customer acquisition against a rival), stepping up after 12 or 24 months (harvesting customers who won't switch), with equipment and fees layered on. Switching requires a technician visit and a Saturday, and providers price around that inertia — the same switching-cost playbook you'll recognize from how airlines price seats.
The ways competition is actually arriving
The interesting development of the past few years is that last-mile competition, where it's growing, mostly isn't coming from a second company digging a parallel trench. It's coming from technologies and programs that dodge the trench entirely.
Fixed wireless. Mobile carriers discovered they could sell home internet over spare 5G capacity — no construction to your address at all, just a receiver in your window. The FCC's report shows fixed wireless as the fastest-growing connection type of recent years. Its capacity per neighborhood is limited, but as a pricing constraint on the incumbent duopoly it has been the most effective new entrant in decades.
Satellite constellations. Low-earth-orbit systems now offer usable service at any address with sky view — transformative at the rural margin where the wired case never closed, and a fallback ceiling on prices everywhere. The physics still charges rent: satellite capacity is shared across huge footprints, so it disciplines the price of rural monopoly service more than it threatens urban fiber. But "some competitor exists at every address in America" is a genuinely new sentence in the history of this market, and it changes the worst-case bargaining position of every rural household.
Fiber overbuilds. In dense, affluent, or pole-friendly markets, fiber builders do sometimes overbuild cable incumbents, because fiber's operating costs and speeds make the second-wire math workable in the right zip codes. The FCC data shows real movement here: the share of households with three or more choices at 100/20 jumped from roughly 19% to 29% in a single year.
Choice is thickening — from a thin base
Households with 3+ fixed providers at 100/20 Mbps — %
Public money for the unprofitable mile. Where no private case works at all, Congress in 2021 created the largest broadband construction subsidy in U.S. history: the BEAD program, $42.45 billion administered by NTIA and allocated to states in proportion to their unserved locations, to pay the capital costs of reaching addresses the market skipped. It is, functionally, the rural electrification playbook applied to fiber and fixed wireless: the government buys down the trench, then a private operator runs the service. (Some cities and cooperatives run their own municipal networks too — though roughly a third of states restrict or condition municipal broadband by statute, a live policy fight we'll simply flag as contested.)
One friction deserves special mention because it decides whether overbuilds happen at all: poles and conduit. A new entrant rarely digs where it can string cable on existing utility poles — but every pole belongs to someone (usually the electric utility or the incumbent phone company), and attaching requires surveys, engineering, "make-ready" work to shift existing lines, and fees, all governed by a lattice of FCC and state rules. Incumbents have little incentive to hurry a competitor onto their poles, and disputes over make-ready timelines are a staple of broadband-deployment dockets. It's a reminder that the last mile's true bottleneck is often not fiber or financing but rights to physical space — the same scarce-corridor problem that shapes every other linear infrastructure in American life.
Reading your own bill like an economist
So, mechanically, what is your monthly number made of? Stack it up: a share of the original construction of the wire on your street (the dominant long-run item), the electronics and upkeep of the local network, "transit" and content costs (small), customer service and billing, taxes and pass-through fees, and a margin calibrated to exactly how easily you could leave. The internet itself — the part with the satellites and the sea cables — is nearly a rounding error in that stack.
The line items below the headline rate follow the same logic. Equipment rental is margin dressed as hardware (a purchased modem typically pays for itself within a year). "Broadcast TV" and "regional sports" fees on bundled bills are content costs deliberately broken out so the advertised price can stay low. And the taxes-and-fees block mixes genuine government charges with provider-invented recovery fees that merely sound governmental — a distinction the FCC's newer "broadband nutrition label" rules force providers to display more honestly at the point of sale.
Which yields the practical takeaways, all of them mechanical rather than moral. If your address has a fiber overbuilder or fixed-wireless option, you have leverage, and the incumbent's retention department knows it — list prices are for the un-annoying. If your address has one wire, your bill reflects that arithmetic, and no amount of phone-tree persistence changes the map. And whether your street's map changes in the next decade depends on three things you can actually watch: 5G capacity in your area, fiber construction crews in your county, and — for the hardest miles — how the BEAD build-out proceeds, address by unserved address, on the FCC's public broadband map.
The wire is the story. It has been since the first telegraph pole went up — and it will be until something makes the last mile cheap, which nothing yet has.
Primary Sources
Reads the primary documents — agency data, GAO reports, court opinions — and explains what they actually say.
No invented credentials: the sourcing is the credential.