What Is Starlink? How Satellite Internet Actually Works

If you have spent any time watching the launch schedule, you will have noticed something: an enormous share of the world's orbital launches carry the same payload. Starlink missions dominate the manifest to a degree that would have seemed absurd a decade ago. SpaceX now launches its own satellites more often than most countries launch anything at all.

So it is worth understanding what all those satellites actually do, why there need to be so many of them, and what their presence in orbit means for everyone else.

The Problem Starlink Was Built to Solve

Satellite internet is not new. It has existed since the 1990s, and for most of that time it was genuinely unpleasant to use. The reason was orbital geometry.

Traditional communications satellites sit in geostationary orbit, roughly 35,786 km above the equator. At that precise altitude a satellite circles the Earth once every 24 hours, which means it appears to hang motionless above one spot on the ground. That is enormously convenient: your dish points at one fixed patch of sky and never moves again. Three such satellites can cover nearly the whole planet.

The catch is distance. A signal travelling from your home to a geostationary satellite and back to a ground station covers something like 70,000 km before the reply even starts its return trip. Radio waves move at the speed of light, but light is not infinitely fast. That round trip imposes a floor of roughly 500 to 600 milliseconds of latency that no amount of engineering can remove. Web pages load sluggishly. Video calls are painful. Anything real-time — gaming, remote control, live conversation — is effectively impossible.

Why Low Earth Orbit Changes the Maths

Starlink satellites orbit at roughly 550 km — about one sixty-fifth of the geostationary distance. The round-trip delay drops to somewhere in the range of 20 to 50 milliseconds, which is comparable to a mediocre cable connection and entirely usable for video calls and most gaming.

But low orbit imposes its own penalty, and it is a severe one. A satellite at 550 km is not stationary relative to the ground; it is moving at roughly 27,000 km/h and crosses the sky in a matter of minutes. To keep a continuous connection you need another satellite arriving overhead the moment the last one leaves. To do that everywhere on Earth, all the time, you need thousands of them.

This is the central trade of the whole system. Geostationary buys you simplicity at the cost of latency. Low Earth orbit buys you latency at the cost of needing an enormous constellation — and therefore an enormous number of launches. Starlink is viable largely because SpaceX also owns the cheapest reusable rocket in the world and can launch its own satellites at internal cost.

How the System Actually Works

Your dish — SpaceX calls it a user terminal, and the community calls it Dishy — is a phased array antenna. Rather than physically rotating to track satellites, it steers its radio beam electronically by adjusting the timing of signals across hundreds of small antenna elements. It has no moving parts for tracking, which is why it can lock onto a satellite crossing the sky in minutes, then hand off seamlessly to the next one.

Early in the programme, every connection had to route from your terminal up to a satellite, then straight back down to a ground station within the same satellite's view. That worked in populated areas but was useless over oceans or remote wilderness, where there is no ground station in range.

Newer satellites carry laser inter-satellite links, which let them pass data directly to their neighbours in orbit. A connection can now hop across several satellites before descending to a ground station on another continent. This is what made mid-ocean and polar coverage possible, and it is a genuinely impressive piece of engineering — aiming a laser precisely enough to hit a target moving at orbital velocity, thousands of kilometres away.

Where It Genuinely Matters

The most compelling case for Starlink is not urban consumers who already have fibre. It is places where laying cable is uneconomic or impossible: rural properties, mountain villages, research stations, ships at sea, aircraft, and remote industrial sites. For these users the alternative is often not slower internet — it is no internet.

It has also proved valuable in disaster response, where terrestrial infrastructure is destroyed and terminals can be flown in and switched on within hours. Its role in Ukraine following the 2022 invasion demonstrated both the strategic usefulness of the system and, uncomfortably, how much influence that concentrates in a single private company's hands.

The Legitimate Objections

Astronomers raised concerns early and have not stopped. Starlink satellites are bright, particularly shortly after launch before they raise to operational altitude, and they streak across long-exposure images. SpaceX has made real mitigation efforts — darkening coatings, sun visors, adjusting orientation during orbit raising — and these have reduced but not eliminated the problem. Wide-field survey telescopes are the worst affected, because they photograph large swathes of sky precisely when satellites are most visible.

Radio astronomy faces a separate and arguably harder problem, since the frequencies used for satellite downlink can bleed into bands reserved for observing faint natural radio sources.

Then there is orbital congestion. Thousands of satellites in a relatively narrow band of altitudes means constant collision-avoidance manoeuvring and a meaningful increase in the long-term risk profile of low Earth orbit. Starlink satellites do manoeuvre autonomously to avoid conjunctions, and they are designed to deorbit at end of life — but the sheer number involved has changed the character of that orbital shell permanently.

The Competition

Starlink is the largest constellation but not the only one. Eutelsat OneWeb operates a smaller network at higher altitude, aimed primarily at enterprise, government, and maritime customers rather than consumers. Amazon's Project Kuiper is building a competing consumer-facing constellation, backed by enormous capital and Amazon's existing infrastructure, though it started years behind. Several national and regional programmes are in various stages of development.

Competition here is unusually capital-intensive. Building a useful low-orbit constellation requires thousands of satellites in orbit before you have a single paying customer with reliable service. That barrier is why the field has so few serious entrants — and why SpaceX's ownership of its own launch vehicle is such a decisive structural advantage.

What to Take Away

Starlink is best understood not as a satellite company but as a consequence of cheap launch. The physics of low-orbit internet were always understood; what changed is that putting several thousand satellites into space stopped being prohibitively expensive. Nearly everything interesting about the current space economy follows from that same shift.

It is also the clearest example of a broader tension. Cheap access to orbit lets us build genuinely useful infrastructure up there — and it lets us fill a shared, finite environment faster than anyone has worked out how to manage.