Infrastructure Ownership Is the Variable That Broadband Policy Arguments Usually Skip

In late 2023, a Swiss ISP called Init7 began selling 25 Gbps symmetrical residential broadband for CHF 199 per month. This is not a lab demonstration or a marketing stunt. It is a commercial product available to ordinary households in Zürich, Bern, and other Swiss cities. A customer pays roughly USD 220 per month and gets 25 gigabits per second in both directions, no data cap, no throttling, with a dedicated /48 IPv6 prefix.

The source article frames this as evidence that the American “free market” approach to broadband has failed. That framing is correct, but it undersells the specificity of what went wrong. The story of Swiss 25 Gbps versus American cable internet is not fundamentally about the merits of markets versus government. It is about a single structural question that most broadband discussions avoid: who owns the last mile, and on what terms can others use it.

Why 25 Gbps Is Technically Trivial on Swiss Fiber

Init7 does not own the fiber going into Swiss homes. The company is a retail ISP operating on top of networks built and owned by municipal utilities, primarily EWZ (the City of Zürich’s electricity and infrastructure utility) and its equivalents in Bern, St. Gallen, and Basel. These networks use active Ethernet architecture, meaning each customer gets a dedicated fiber strand from their premises to an aggregation switch at the utility’s point of presence. There is no passive splitter, no shared medium, no TDMA scheduling. The fiber is simply glass.

The consequence of this architecture is that upgrading speed requires swapping transceivers. A connection running at 1 Gbps uses a 1000BASE-LX SFP at both ends. Upgrading to 25 Gbps means installing an SFP28 25GBASE-LR transceiver at the customer premises and a matching port on the aggregation switch. The fiber itself, single-mode G.652.D, is rated for far higher speeds. Init7 ships customers a 25GbE transceiver module. Customers plug it into a compatible switch, such as a MikroTik CRS326 or similar, and the upgrade is complete. The bottleneck at that point is internal storage, not the internet connection.

This is categorically different from what most American customers have. Around 65% of US residential broadband runs over DOCSIS cable infrastructure, which is a shared medium. A coaxial node in a neighborhood serves anywhere from 250 to 500 homes, and all of them share the available bandwidth. DOCSIS 3.1 can deliver impressive peak downstream numbers, 1 to 2 Gbps during off-peak hours, but upstream bandwidth on cable infrastructure is structurally scarce. The US median fixed broadband upload speed sits at roughly 27 Mbps according to Ookla’s Global Speedtest Index. Switzerland’s median upload is approximately 180 Mbps. That disparity is the honest comparison. The download headline numbers are misleading; the upload numbers reveal the infrastructure.

The Utility Model and Why It Creates Retail Competition

EWZ is an electricity utility. It laid fiber across Zürich as a natural extension of its existing civil infrastructure and right-of-way position. The fiber network is owned by the city, operated as a neutral carrier, and available to any ISP that signs a wholesale agreement. Init7, Yallo, Salt Mobile, Sunrise, and others all operate on the same physical infrastructure. They compete on price, service quality, IPv6 support, and routing decisions. The infrastructure owner does not participate in retail competition.

This structural separation, sometimes called “functional unbundling” or the utility model, is precisely what Switzerland’s high-performing cities have and what the United States systematically dismantled in the early 2000s.

The Swiss federal regulator BAKOM enforces mandatory unbundling on Swisscom’s copper and fiber infrastructure, and Swisscom itself is roughly 51% owned by the Swiss federal government, aligning regulatory and ownership incentives toward broad deployment. Neither condition exists in the US.

The 2002 Decision That Locked In the American Model

The critical inflection point in US broadband history is not the 1996 Telecommunications Act, which at least attempted to mandate unbundling of telephone infrastructure. It is a pair of FCC decisions in 2002 and 2005.

In 2002, the FCC under Chairman Michael Powell classified cable modem service as an “information service” under the Communications Act rather than a “telecommunications service.” This classification mattered enormously because telecommunications services are subject to common carrier obligations, including mandatory access for competitors at regulated wholesale rates. Information services are not. The Supreme Court upheld this classification in Brand X Internet Services v. FCC (2005). In 2005, the FCC extended the same classification to DSL broadband, eliminating unbundling requirements from telephone company internet service as well.

In a single decade, the regulatory structure that might have enabled competitive access to broadband infrastructure was dismantled. The result was a duopoly in most US markets: the local cable company and the local telephone company. In many suburban and rural areas, only one of those two actually offered high-speed service.

The net neutrality debates that followed, from the 2010 Open Internet Order through the 2015 Title II reclassification, the 2017 reversal under Chairman Pai, the 2024 reinstatement, and the subsequent court vacatur, were downstream of this foundational decision. The question of whether ISPs must treat all traffic equally is secondary to the question of whether customers have any choice of ISP at all.

Chattanooga Proves the Model, Which Is Why It Got Contained

The strongest domestic evidence that the utility model works is the EPB network in Chattanooga, Tennessee. The Electric Power Board, a city-owned utility, launched 1 Gbps residential fiber in 2010, the first city in the US to offer it commercially. The network was built for approximately $330 million, partially funded by a $111 million federal stimulus grant. By 2015, EPB had upgraded to 10 Gbps availability. Pricing starts around $68 per month for 1 Gbps, well below what Comcast or AT&T charge where those services exist.

The economic impact studies are not subtle. Independent research has attributed roughly $865 million in economic development to the EPB network over its first decade, including attraction of technology employers to a city that had no particular claim to them beforehand.

The response from the incumbent carriers was not to compete. It was to lobby. AT&T and Time Warner Cable supported the drafting of North Carolina Senate Bill 346 in 2011, which imposed severe operational and financial restrictions on municipal ISPs and effectively prevented Wilson, NC’s Greenlight network from expanding into surrounding counties that wanted the service. Tennessee passed similar legislation containing EPB’s geographic reach. When the FCC attempted to preempt these state laws in 2015, the Sixth Circuit ruled in Tennessee v. FCC (2016) that the agency lacked statutory authority to do so.

As of 2024, 17 to 19 US states have laws restricting or prohibiting municipal broadband to varying degrees. The American Legislative Exchange Council, funded significantly by AT&T and other carriers, produced model legislation that multiple state anti-municipal-broadband laws trace directly to.

The OECD Picture Is Not Ambiguous

Across comparable developed nations, the countries with the best broadband outcomes share a structural feature: either state ownership of the infrastructure layer or mandatory wholesale access that allows retail competition on privately owned infrastructure. South Korea built government-funded fiber in the late 1990s and early 2000s under national plans and now has some of the fastest and cheapest residential broadband in the world. Japan’s NTT is majority government-owned and subject to mandatory wholesale requirements. Sweden’s Stokab, owned by the City of Stockholm, operates an open-access dark fiber network that any ISP can use. The UK’s Ofcom spent years pressuring BT to operationally and then legally separate its wholesale infrastructure arm, Openreach, from its retail business.

The OECD Broadband Portal documents this consistently. Countries with structural separation or public infrastructure ownership outperform countries that rely on private vertical integration to self-regulate toward competition. The mechanism is straightforward: a vertically integrated ISP that owns both the physical infrastructure and the retail service has no incentive to make the infrastructure attractive to competitors. An infrastructure utility that earns revenue by selling wholesale access to multiple retail ISPs has every incentive to expand coverage.

What It Would Actually Take

Fix this in the United States requires either Congress or the FCC to reclassify broadband as a Title II telecommunications service and impose meaningful unbundling requirements, or a significant expansion of federal funding for open-access municipal fiber with preemption of state laws that restrict it. The Infrastructure Investment and Jobs Act of 2021 allocated $65 billion toward broadband, but most of that funding flows through programs that do not require open-access conditions on funded infrastructure. Private ISPs can receive federal money to build networks they will then operate as closed systems.

The Swiss outcome is not an accident of geography, culture, or unusual public appetite for infrastructure spending. Switzerland is a small country but not a uniquely collective one. What it has is a regulatory framework that treats the physical fiber as a utility layer separate from the services running over it, a framework built partly through deliberate policy and partly through the historical accident of municipal electricity utilities that had right-of-way and civil engineering capacity already in place.

The US has EPB Chattanooga. It has UTOPIA in Utah, where multiple ISPs compete on shared municipal fiber across several cities. It has Longmont’s NextLight offering 1 Gbps symmetrical for around $70 per month. These are not isolated curiosities; they are proof of concept. The reason they have not scaled is not technical, financial, or geographic. It is legislative, and the legislation was written by the companies that benefit from the current arrangement.

Init7’s 25 Gbps product is a transceiver swap on top of infrastructure that a city utility built and made available to any ISP that wants to use it. The question of why American consumers do not have equivalent options has a precise answer: the regulatory structure that would have required comparable infrastructure access was dismantled between 2002 and 2005, and the companies that benefit from its absence have spent reliably to keep it that way.