Start with the mechanism, because it's simpler than any one country's politics. Communication -- voice, then data, then everything -- has to physically travel through something: a switchboard, a numbering system, a gateway router, a cable on an ocean floor. Whoever controls the physical thing it travels through can, if they choose to, observe it, throttle it, or cut it off entirely. That's not a statement about any particular government's intentions. It's a statement about physics and ownership. A chokepoint doesn't need to be built FOR control to become a tool of it -- it only needs to be built, and then to change hands, or change purpose, once.

The oldest version of this sits inside the US itself, built by a company, not a state. AT&T and Bell Labs developed the North American Numbering Plan in 1947 for Direct Distance Dialing -- the system that gave the US and Canada the "+1" country code and every area code inside it, still the backbone of how a phone call physically routes today.^[1] AT&T built and ran nearly the entire American phone network as a government-tolerated near-monopoly for most of the 20th century, a status formalized as early as the 1913 Kingsbury Commitment.^[1] That monopoly wasn't broken by AT&T choosing to share power. It took a federal antitrust suit, settled as a 1982 consent decree overseen by Judge Harold Greene, to force the actual breakup -- AT&T divested into the regional "Baby Bells" on January 1, 1984.^[1] The company didn't build the numbering plan to control American communication. It built it to run a phone network efficiently. The monopoly that came with owning the only network at all was a separate fact, one the government eventually had to use its own power to break.

The most direct version of the same mechanism, aimed at population control on purpose, runs through Iran. Iran's international internet connectivity routes through essentially two state-controlled gateways -- the Telecommunication Infrastructure Company (TIC) and the Institute for Research in Fundamental Sciences (IPM).^[2] That narrow, designed chokepoint is what made a near-total nationwide blackout possible during the November 2019 fuel-price protests -- connectivity fell to near zero for roughly a week, and Amnesty International later documented that the shutdown was used specifically to conceal the mass killing of protesters while it was underway.^[3] It happened again after Mahsa Amini's death in September 2022: Instagram and WhatsApp cut first, then nationwide mobile shutdowns, with Kurdistan province -- the region of Amini's own death -- hit with a total regional blackout that year.^[4] Two gateways is not an accident of geography. It's a designed chokepoint, and it gets used exactly the way a chokepoint can be used.

China runs the same mechanism at a different scale, permanently rather than episodically. The Great Firewall doesn't work by policing every website individually -- it works by routing essentially all of mainland China's international traffic through a small number of state-linked gateway points concentrated in five cities (Beijing, Guangzhou, Shanghai, Urumqi, Xi'an), operated by state-connected carriers.^[5] That's the same structural logic as Iran's two gateways, built as permanent, standing infrastructure instead of an emergency measure -- filtering, by design, more than a billion people's access to the outside internet as a normal condition, not a crisis response.

None of this is confined to authoritarian states, which is the part that's easiest to miss. India, a democracy, led the world in the sheer count of internet shutdowns in 2024 -- 84 of them, across 16 states and territories, mostly regional and tied to unrest or exam security.^[6] Globally, 2024 set a record for internet shutdowns -- 304 of them across 54 countries -- and 2025 broke that record again.^[6] And the fight isn't only about shutting a network off. In 2025, the UK government reportedly ordered Apple to build a backdoor into UK users' iCloud end-to-end encryption; rather than comply, Apple disabled Advanced Data Protection for UK accounts entirely.^[7] The European Commission published its own roadmap in June 2025 for "lawful access to data," with a further 2026 roadmap aimed specifically at finding ways around end-to-end encryption -- a mandatory client-side-scanning proposal was dropped after member-state pushback, but the underlying effort continues.^[7] A democratic government reaching for the same kind of access an authoritarian one reaches for isn't a scandal exclusive to one system of government. It's the same chokepoint logic, available to whoever is positioned to use it, regardless of what else that government is.

Underneath all of it sits the most physical chokepoint of all, and it's more concentrated than most people realize. Roughly 95-99% of international internet traffic runs through undersea fiber cables, not satellites.^[8] Egypt alone carries an estimated 17% of the entire world's internet traffic through just 16 cables crossing its territory.^[8] And this chokepoint doesn't have to be used on purpose to matter -- it just has to be found. Since 2022, roughly ten undersea cables in the Baltic Sea region have been cut or damaged; seven of those happened between November 2024 and January 2025 alone, including the BCS East-West Interlink and C-Lion1 cables (Nov 17-18, 2024, a Chinese vessel implicated) and the Estlink 2 power cable (Dec 25, 2024, a tanker linked to Russia's "shadow fleet" implicated).^[9] Official attribution remains unconfirmed in most of these cases -- which is itself the point. A chokepoint that can plausibly be damaged without anyone having to admit responsibility is, if anything, a more useful one than a chokepoint you have to openly seize.

There's a fifth, newer version of the same mechanism, and it's domestic: owning a piece of the company that runs the chokepoint instead of owning the chokepoint outright. "The equity stake used to be rare" documents the real, current pattern -- the US government taking direct equity positions in Intel, MP Materials, GlobalFoundries, and Westinghouse since July 2025. None of those are communication chokepoints in the AT&T/Iran/China sense. But the mechanism is the same one this piece is about, wearing different clothes: a government doesn't need to build a chokepoint, or seize one outright, to end up with leverage over it. An ownership stake in the company that operates it gets there by a third route -- not the state building the gate, not the state seizing the gate, but the state becoming a shareholder in whoever already holds it.

None of this is an argument that infrastructure shouldn't exist, and treating it as one would miss the actual point. A phone network, an internet gateway, an undersea cable, a critical-materials supplier -- all of them have to exist for a modern country to function at all, and almost none of them were built with control as the founding purpose. AT&T wanted to run a phone company. Iran and China's own designed gateways are the exceptions that prove the general rule, not the norm. But the general rule is the actual finding: physical concentration creates leverage whether or not leverage was the original intent, and that leverage sits there, available, for as long as the concentration does -- waiting only for whoever ends up holding it to decide to use it. The chokepoint doesn't check anyone's motives before it works. It never has to.