Broadband's Smiling Curve
America's expensive broadband stems from physical monopolies, not geography, proving that a competitive free market requires regulating the dirt.
By Ben Thompson
Sparked by Why Switzerland has 25 gbit internet and America doesn't · discussion
To read the Hacker News reaction to Stefan Schueller’s viral post about Swiss broadband is to be immersed in the geographic density argument. The consensus is straightforward: America is vast and suburban, Switzerland is dense and alpine, therefore any direct comparisons of internet pricing are inherently unfair.
If you want to see this density-as-destiny argument fall apart, though, you only need to look at Australia's NBN — a state-directed project covering a continent far emptier than America, which struggled not because of kangaroos and vast distances, but because of early political compromises on the physical architecture (relying on existing copper instead of pure fiber). Or, conversely, look at dense US suburbs, where houses sit practically on top of each other, yet residents still suffer under single-provider monopolies charging exorbitant rates. The density argument is, to be fair, a compelling narrative. Laying fiber in rural Montana is fundamentally a different capital expenditure equation than wiring an apartment block in Zurich. The economics of American broadband, though, have less to do with the physical distance between houses and everything to do with who owns the trench.
Consider the value chain of an American internet service provider. To actually deliver a packet of data to a consumer, an ISP must acquire municipal rights of way, dig the trench, lay the physical fiber optic cables, light that fiber with optical networking terminals, and finally, run the routing software to connect that local loop to the broader internet. In the United States, a single corporate entity typically executes all of these steps.
This is a model characterized by staggering upfront capital costs and total vertical integration. Because one company owns the entire stack from the dirt to the router, they must recoup the massive CapEx of the physical buildout through the monthly service fee. More importantly, because digging trenches is prohibitively expensive, that single entity inevitably holds a local monopoly or, at best, a duopoly. The barrier to entry isn't software expertise; the barrier to entry is the literal cost of moving earth. That is the fundamental reality of telecom economics: the marginal cost of routing a packet is near zero, but the fixed cost of burying a conduit is astronomical.
The natural economic result is visible in the extreme pricing structure of high-end American internet. Take Ziply Fiber's $900/month 50Gbps tier. That eye-watering price tag is not merely a reflection of the cost of high-end routing equipment. Rather, it represents the unified rent extraction required when a single company has to monetize the entire physical infrastructure stack. When you are paying for an American gigabit connection, you are paying a premium for the software and routing, but you are primarily paying off the debt of the trench. There simply is no way to decouple the two; you cannot buy your physical connection from Comcast and your routing software from a nimble startup.
The market is defined by the dirt.
This brings us to the actual mechanism that makes the Swiss market different, which was thrown into sharp relief last month. In April, the Swiss Competition Commission, known as COMCO, issued an 18 million CHF fine against Swisscom.
The reason for the fine is the crux of the entire broadband debate. Swisscom was not penalized for price gouging or traditional monopolistic behavior at the retail level. They were fined for attempting to build point-to-multipoint networks instead of point-to-point networks.
To understand the difference, a point-to-multipoint architecture aggregates multiple homes onto a single fiber strand using passive splitters in the street. This makes it technically impossible for a competitor to lease access to just one specific household's line without intercepting the data of neighbors. A point-to-point architecture, conversely, runs a dedicated, individual glass strand from the central telephone exchange directly to the living room. It is more expensive to build initially, but it guarantees complete unbundling.
The Swiss government demands the latter. This is literal state-mandated physical modularization — a structural intervention at the lowest possible layer — ensuring any provider can lease the raw, unlit fiber. Swisscom can dig the trench and lay the glass, but they are legally compelled to let any competitor lease that dark fiber on perfectly equivalent terms.
This is not a free market emerging organically from the goodness of telecom executives' hearts. It is the heavy hand of the state reaching down to the very bottom of the technology stack and forcing open-access modularity at the physical layer.
To understand the profound economic effects of this mandate, it is useful to apply a framework originally coined in the early 1990s by Stan Shih, the founder of Acer: The Smiling Curve.
Shih observed that in the personal computer manufacturing industry, the value chain was shaped like a smile. On a chart, the Y-axis represents "Value Added" or profitability, while the X-axis tracks the sequence of production. The curve starts high on the left edge with research and development or core component design — think Intel or Microsoft. It dips severely in the middle, representing the low-margin, commoditized work of physical manufacturing and assembly. The curve then swoops high again on the right edge, representing the high-margin realms of software, marketing, and customer experience.
The primary lesson of the Smiling Curve is that profits flow to the edges, while the undifferentiated middle is a brutal, low-margin utility.
The traditional American ISP model prevents the Smiling Curve from forming. Because the incumbent ISP owns the trench (the middle) and refuses to lease it, they artificially elevate the profitability of that middle segment. They use their monopoly over the physical dirt to extract rents across the entire chain, preventing any specialized competitor from attacking the right edge. To put it another way, companies like Comcast are not making their margin because they offer a transcendent routing experience; they make their margin because you literally cannot get a wire to your house without them.
The Swiss policy, by contrast, forces the Smiling Curve into existence. When the state mandates that the physical infrastructure must be open-access and point-to-point, the trench plummets into the undifferentiated middle of the curve. It becomes a pure, commoditized utility. The left edge is the initial capital to secure rights of way, the middle is the raw unlit fiber, and the right edge is the actual internet service.
Once that middle is commoditized, the barriers to entry at the right edge of the curve completely collapse. An aspiring ISP in Switzerland does not need to raise billions of dollars to negotiate with local municipalities and dig up sidewalks. They only need to rent space in a central exchange, install a high-capacity router, and lease the dark fiber leading directly to the customer's home.
This structural modularity is what enables the numbers that spark so much envy on technology forums. Consider the Swiss ISP Init7, which currently offers a jaw-dropping tier of CHF 777 annually for 25 Gbps.
Init7 is not a vertically integrated behemoth. They operate as an entirely distinct type of business — a completely horizontal one living exclusively at the right edge of the Smiling Curve. Because they do not have to dig trenches or lay glass, they can focus all of their capital and engineering talent on pure software, packet routing, peering agreements, and customer experience.
The result is a product that simply could not exist under an integrated American model. Init7 is essentially a software company that happens to route packets, riding atop a state-enforced commodity physical layer. They push the bounds of what consumer broadband looks like because their entire business model depends on differentiating on service, not resting on a localized monopoly of dirt.
This brings us back to the premise of Schueller’s post, and the fundamental tension of telecom policy. The Hacker News crowd is wrong to blame America's pricing entirely on its geography, but they are right that "the free market" is a loaded term here. True competition in broadband does not emerge from deregulation; in fact, deregulation in a market defined by massive fixed costs inevitably leads to localized monopolies.
If you want a vibrant, competitive market at the service layer, you have to ruthlessly commoditize the physical layer. You have to force the Smiling Curve into existence. The paradox of Swiss broadband is that the most fiercely competitive, low-priced, consumer-friendly market is not the result of unchecked laissez-faire capitalism. It is the direct downstream consequence of aggressive state intervention, literally dictating the architecture of the wires in the ground.
If you want the free market to thrive in the cloud, you have to regulate the dirt.