The Last Mile Is Always Political: What Swiss Broadband Gets Right That America Refuses To

A few weeks ago I came across a post from Stefan Schueller laying out why Switzerland has access to 25 Gbit residential fiber while large parts of the United States struggle to hit 100 Mbps. The argument isn’t new, but the framing is sharp: American broadband underperformance isn’t a market outcome, it’s a policy outcome dressed up as one.

I want to push the technical and economic case further, because the usual discourse around this gets stuck in left-versus-right framing and never examines the actual infrastructure decisions that produce the gap.

The Natural Monopoly Problem Nobody Wants to Name

Digging fiber to every home in a city costs an enormous amount of money. Once that fiber is in the ground, the economics of the last mile strongly favor one provider. The cost to lay a second set of fiber conduit to the same street is nearly identical to the first, but the revenue from the second network is split with the incumbent. This is the textbook definition of a natural monopoly, and it applies to roads, water mains, and electrical wires for the same reason it applies to fiber.

Truly free markets don’t function well in natural monopoly conditions. The outcome is either a monopoly that extracts rent, or a regulated utility with mandated access. Both the US and Switzerland figured this out, but they made opposite bets on which structure to use.

The US went with franchised geographic monopolies. Under the Cable Communications Policy Act of 1984, municipalities could grant exclusive franchise agreements to cable operators. The theory was that guaranteed market exclusivity would incentivize infrastructure buildout. It did, in the sense that cable got built. But it also locked in an ownership structure where the same company that owns the conduit and the fiber also sells you the service and has no structural incentive to compete with itself on price or speed.

Switzerland, along with most of northern Europe, built or inherited a different structure. The physical infrastructure, the conduit and fiber, is owned and operated separately from the internet service layer. Swisscom, which is majority state-owned, built out a massive FTTH network but is required under Swiss telecommunications law to provide open access to that infrastructure. Other ISPs, including smaller regional providers, can run their services over Swisscom’s fiber. Meanwhile, many Swiss cities have their own municipal fiber utilities, extensions of the same public works departments that run water and power. Zurich’s EWZ, Bern’s Energie Wasser Bern, and similar Stadtwerke (city utilities) across the country have been laying fiber for decades.

This is not socialism in any meaningful sense. The service layer, where ISPs compete on price, speed tiers, and service quality, is a genuine market. Competition happens at the layer where competition actually makes sense. The physical pipes are treated like roads: public infrastructure that enables commerce rather than commerce itself.

What 25 Gbit Actually Requires

The headline number is striking enough to be worth examining on its own technical terms. Residential 25 Gbit service is being offered by Swiss ISPs including Init7, which has been operating a full BGP-routed flat-rate fiber product for years. Their fiber7 product has been a reference case in European networking circles because it’s a genuinely routed, non-oversubscribed connection rather than a “up to” marketing figure.

Deploying 25 Gbit to a home requires point-to-point fiber, not GPON (Gigabit Passive Optical Network), which shares bandwidth across 32 to 128 subscribers on a single fiber strand using optical splitters. GPON is how most US fiber deployments, including AT&T Fiber and Verizon’s older FIOS infrastructure, handle residential service. The physics of GPON cap aggregate downstream at 2.5 Gbit in the older XG-PON standard, or 10 Gbit in the newer XGS-PON revision, shared across all subscribers on that tree.

Point-to-point fiber runs a dedicated strand from the street-level termination point to each premise. It’s more expensive per drop because you can’t amortize a single fiber across multiple homes, but each subscriber gets the full capacity of their wavelength. This is the architecture that makes 25 Gbit per household physically plausible, and it’s what municipal networks in Switzerland have been building.

The US mostly got GPON because GPON is cheaper to deploy per home passed, and private capital always optimizes for that metric first.

The American Municipal Broadband Exception

The US actually has a handful of municipal broadband success stories, and they’re instructive precisely because they exist within the same country as the horror stories.

Chattanooga, Tennessee built a city-owned fiber network starting in 2009 through EPB, the city’s electric utility. By 2010 it was offering symmetric gigabit service, years before Google Fiber launched in Kansas City and made that headline. EPB’s network currently offers 10 Gbit residential tiers and operates at costs well below what Comcast and AT&T charge in comparable markets. The infrastructure investment has been credited with attracting data centers and tech employers to the city at rates that outpaced regional peers.

The story that doesn’t get told alongside Chattanooga is how hard the incumbents fought to prevent others from doing the same thing. Comcast, AT&T, and their state-level lobbying arms have successfully passed laws in at least 18 US states that restrict or outright prohibit municipalities from building their own broadband networks. These are not obscure regulations buried in utility law; they’re explicit preemptions passed in states like North Carolina, Tennessee (yes, the same state that has Chattanooga), Texas, and Michigan, specifically to stop cities from becoming ISPs.

When you see that list, the “free market” framing collapses. Incumbents lobbying to prevent government competition is not a free market position. It’s rent-seeking using legislative power to maintain monopoly conditions.

Infrastructure Investment and Its Discontents

The 2021 Infrastructure Investment and Jobs Act included $65 billion for broadband, channeled through the BEAD (Broadband Equity, Access, and Deployment) program administered by NTIA. That’s a significant number, but the implementation has been characteristically messy. States had to submit deployment plans, which required defining “unserved” and “underserved” coverage using FCC maps that were notoriously inaccurate because they relied on ISP self-reporting. The FCC’s own process for challenging those maps is cumbersome enough that rural counties in many states are still fighting over whether their existing service actually meets the 25/3 Mbps definition of “served.”

The threshold itself, 25 Mbps download and 3 Mbps upload, is the FCC’s longstanding benchmark for broadband adequacy. The asymmetry reflects a consumption model that’s at least a decade out of date. Remote work, video conferencing, and home servers all push meaningful upload traffic. Switzerland doesn’t have an “upload” problem because symmetric fiber is the deployment standard, not an upsell tier.

The Dig-Once Policy That Could Change The Math

One structural reform that doesn’t require choosing between public and private ownership is a mandatory “dig once” policy, requiring conduit installation whenever a road is opened for utility work. The physical act of trenching is most of the cost of fiber deployment; the fiber and electronics are relatively cheap. Requiring conduit installation during road maintenance effectively pre-builds the fiber pathway at a fraction of the standalone cost.

Several European countries and a growing number of US cities have adopted dig-once requirements. It’s a genuinely market-compatible policy that doesn’t mandate who owns or operates the infrastructure, only that the pathway gets built while the ground is already open. The lobbying resistance to even this modest intervention tells you a lot about who benefits from the current deployment pace.

Switzerland’s advantage isn’t that it’s small, or wealthy, or that its population is conveniently clustered. The density argument doesn’t survive contact with the US cities, including dense urban centers like Los Angeles and Atlanta, where gigabit fiber is still unavailable in large residential areas. The advantage is that Switzerland separated infrastructure ownership from service provision early, and built a regulatory structure that treated the physical layer as a public utility rather than a product.

That’s a political decision, not a technical one. The technical decisions, fiber type, GPON versus point-to-point, open access protocols, symmetric capacity planning, all follow from it.