On July 7, SpaceX filed an application with the Federal Communications Commission seeking authorization to deploy 100,000 third-generation Starlink satellites. The number grabbed the headlines — 100,000 is a staggering figure, roughly twenty times the total number of operational satellites currently orbiting Earth. The predictable reactions followed immediately: cheers from connectivity evangelists, alarm from astronomers, and the usual round of regulatory hand-wringing about orbital debris and spectrum allocation.
But the most important word in the filing isn’t “100,000.” It’s “backhaul.”
Buried in the technical annex of the Gen3 application is language stating that the constellation is designed to route internet backhaul traffic — the mid-mile infrastructure that connects local networks to the global internet — and that the system aims to handle “the majority of the world’s internet traffic.” This is not a filing about beaming Netflix to rural cabins. It is a filing about replacing the physical backbone of the internet with a privately owned orbital mesh.
The Filing Says ‘Backhaul’ — and That Changes the Arithmetic
Backhaul is the unglamorous plumbing of the internet. It is the fiber trunks running alongside railways, the undersea cables snaking across ocean floors, the peering facilities where networks exchange traffic in windowless buildings outside Ashburn, Virginia. It is expensive, slow to build, and — crucially — owned by a distributed patchwork of telecoms, consortiums, and governments. No single entity controls it.
SpaceX’s Gen3 satellites are specced to change that. Each spacecraft, weighing between 2,000 and 2,500 kilograms and deliverable only via Starship’s 100-ton payload capacity, will push 1 terabit per second of downlink throughput. That is a tenfold leap over the Gen2 hardware. At 100,000 units, the constellation’s aggregate capacity enters territory where it can plausibly absorb a meaningful fraction of global internet traffic — not just the last-mile connections Starlink is known for, but the transit paths between continents.
A network architect at a major cloud provider, reached via Slack after the filing dropped, put it bluntly: “We just finished a five-year undersea cable project. I’m looking at the Gen3 throughput numbers and wondering if we built the last dinosaur.”
One Company, One Backbone, One Point of Control
The internet’s resilience has always been architectural. It routes around damage because no single entity owns the critical mass of paths. That property is not a law of nature; it is an artifact of how the physical layer was built — by hundreds of independent operators making independent investment decisions over four decades.
A constellation that carries the majority of the world’s internet traffic collapses that diversity into a single operator’s routing tables. The implications extend well beyond the familiar net neutrality debates, which were always about last-mile ISPs throttling content. A backbone operator with dominant market share can shape traffic flows in ways that are far harder to detect and far more consequential — prioritizing certain cloud providers, degrading certain national networks, or simply making routing decisions that advantage its own content and services.
National governments that have spent years building domestic internet infrastructure and negotiating peering agreements would find their leverage sharply reduced. A country can regulate the fiber under its soil. It has far fewer options when the traffic is passing through a mesh of satellites owned by an American company and routed through ground stations of that company’s choosing.
The Regulators Are Debating the Last War
The FCC’s current approach to satellite licensing, as the Benton Institute noted this week, is moving toward an “assembly-line process” designed to speed approvals. The framework was built for a world where satellite constellations competed with terrestrial ISPs for consumer broadband subscribers — a market with established competitive dynamics and a playbook for oversight.
A single company seeking to carry the majority of global internet traffic is not that market. It is a structural transformation of how the internet’s physical layer is owned and operated, and it raises questions that the FCC’s consumer-broadband framework was never designed to answer. Should any single entity — however competent, however well-capitalized — be permitted to control the backbone that carries most of the world’s data? What happens to peering disputes when one party owns the sky? And why is the International Telecommunication Union, which has governed global spectrum coordination for 160 years, largely absent from this conversation?
SpaceX’s filing is technically impressive. The engineering is real, the capacity numbers are extraordinary, and the ambition is, in its own way, bracing. But the filing is also a test — of whether our regulatory institutions can recognize a category shift when they see one, or whether they will process this as just another broadband application, stamp it, and move on to the next item on the docket.
The number that matters isn’t 100,000. It’s one. One backbone. One operator. One point of control. That is the question the filing actually poses, and so far, nobody in a position to answer it seems to be asking.