How Starlink's Laser Links Work

Before laser crosslinks, a Starlink satellite was a mirror. Data went up from your dish and came straight back down to a ground station within the same footprint. If there was no ground station in view, the satellite could do nothing for you.

That constraint drew the map of where Starlink could work. Crosslinks erased it.

What a crosslink actually is

Each satellite carries several optical terminals — laser transceivers on gimbals that can point at another satellite and hold that pointing while both are moving at 7.6 km per second.

A typical satellite maintains links in two directions along its own orbital plane, forward and backward to the satellites ahead and behind it, plus links across to satellites in adjacent planes. The result is a mesh: data can hop from satellite to satellite around the constellation without touching the ground.

Each link carries on the order of 100 Gbps. Multiply by the number of simultaneous links per satellite and by thousands of satellites, and the constellation becomes a routed network in its own right rather than a collection of independent relays.

• Multiple optical terminals per satellite, on steerable mounts
• Links in-plane, forward and aft, to immediate orbital neighbours
• Links cross-plane, to satellites in adjacent orbital planes
• Roughly 100 Gbps per link, forming a mesh network in orbit

The pointing problem

This is the hard part, and it is worth appreciating how hard. Two satellites thousands of kilometres apart, each travelling at 7.6 km/s in different directions, must find each other with a laser beam narrow enough to be useful and then hold that alignment continuously.

A laser beam that spreads too much wastes power; one that is tight enough to be efficient is correspondingly unforgiving about pointing. The terminals solve this in stages — a coarse acquisition scan using predicted positions from onboard orbit knowledge, then a fine tracking loop that locks onto the received beam and corrects continuously.

The links also have to be re-established regularly. As satellites move through their orbits, the optimal neighbour changes, so terminals break and re-acquire links routinely as part of normal operation.

Optical crosslinks are a pointing problem first and a communications problem second. The bandwidth was never the hard bit.

Why light is faster in space than in fibre

This surprises people, and it is genuinely true. Light travels at about 300,000 km/s in vacuum. In glass optical fibre it travels roughly a third slower, because the refractive index of glass is around 1.47 — that is what a refractive index means.

So a signal crossing a given distance through space beats the same distance through fibre by a substantial margin. Space paths are also straighter: undersea cables follow seabed routes with detours around geography and politics, while a laser mesh can approximate a great circle.

The catch is the ends. Getting up to orbit and back down costs latency that a terrestrial route does not pay. For short distances that overhead dominates and fibre wins comfortably. For very long intercontinental routes the vacuum advantage can catch up and, on some paths, overtake — which is why low-latency financial networking has taken an interest in satellite mesh routing.

What crosslinks made possible

The capability list is basically a list of places with no ground stations.

• Mid-ocean maritime service, thousands of kilometres from any gateway
• Aviation over oceans and polar routes
• Antarctic research stations, where a terrestrial gateway is not an option
• Remote wilderness and small islands with no local infrastructure
• Resilience — traffic can route around a ground station that is congested, weathered out or offline

Why they are not used for everything

Every crosslink hop adds latency and consumes capacity that could serve users. If there is a ground station in your satellite's footprint, the shortest path is straight down — and that is what the network will use.

Crosslinks are for when the direct path does not exist or is a bad choice. This means their value is concentrated exactly where terrestrial infrastructure is absent, which is also where Starlink's competitive position is strongest.

It also means the number of ground stations still matters enormously. A dense gateway network keeps most traffic on one hop, and the crosslink mesh handles the remainder.

The part you cannot see

Optical crosslinks are invisible from the ground in every sense. The beams point sideways between satellites rather than downward, they operate at infrared wavelengths outside human vision, and nothing about a satellite's appearance changes when its lasers are active.

You can, however, see the structure that makes them work. The constellation's shell organisation — satellites evenly distributed across defined orbital planes — is exactly what a mesh network needs, because it makes each satellite's neighbours predictable. The globe on this site plots that geometry live, and the regular spacing along orbital tracks is the crosslink topology made visible.