Starlink vs Kuiper vs OneWeb: How the Constellations Differ

Comparisons of satellite constellations usually turn into a satellite-count contest, which misses the point. The interesting differences are in orbit height and business model, and those two choices constrain everything downstream.

Starlink: scale and vertical integration

Roughly 550 km, thousands of satellites, direct-to-consumer. Starlink's defining advantage is that SpaceX owns the rocket. Launching your own hardware at internal cost changes what constellation sizes are economically sensible, and it is why Starlink could commit to a satellite count nobody else could match.

It is also the only one of the three that started as a consumer product. You can order a dish, plug it in, and be online — a distribution model closer to a consumer electronics company than a traditional satellite operator.

The trade-off is capacity sharing. A consumer service in a popular area gets congested, which is why waitlists and priority tiers exist.

Amazon Kuiper: the well-funded challenger

Kuiper targets a constellation of a few thousand satellites in shells around 590 to 630 km — somewhat higher than Starlink's main shells, at inclinations chosen to concentrate coverage on populated mid-latitudes rather than reaching the poles.

Amazon's structural advantages are different from SpaceX's. It does not own a rocket, and has contracted launches across multiple providers, which is more expensive and schedule-dependent. What it does own is AWS — an enormous global cloud and networking footprint that a satellite network can plug directly into, and an existing relationship with a very large number of enterprise customers.

The consumer terminal strategy also mirrors Amazon's usual approach: hardware sold at aggressive prices as a route to the service. Whether that translates into meaningful market share depends largely on how quickly the constellation can be deployed, which is the constraint launch capacity imposes.

• Orbit: roughly 590–630 km, in shells at mid-latitude inclinations
• Scale: a few thousand satellites planned
• Launch: contracted across multiple providers rather than in-house
• Advantage: AWS integration and existing enterprise relationships
• Constraint: deployment rate is bounded by purchased launch capacity

OneWeb: higher, fewer, and not aimed at you

OneWeb, now part of Eutelsat, took the opposite approach on almost every axis. Its constellation sits at about 1,200 km — more than twice Starlink's altitude — in near-polar orbits at around 87.9° inclination, and totals several hundred satellites rather than thousands.

Higher orbit is a deliberate trade. Each satellite covers far more ground, so global coverage needs far fewer of them, which means far fewer launches and a much smaller replacement treadmill. The cost is latency: at 1,200 km the round trip is roughly twice as long as Starlink's, putting typical figures in the 70–100 ms range rather than 25–60.

That is perfectly acceptable for what OneWeb sells. It is not a consumer service. OneWeb sells capacity wholesale to telecom operators, governments, maritime and aviation customers, and enterprises — buyers who care about coverage, reliability and contract terms far more than about whether a video call feels snappy.

OneWeb is not losing a race with Starlink. It entered a different one — wholesale capacity, not consumer broadband.

Why altitude is the fundamental choice

Every other difference cascades from orbit height, and the trade is unavoidable.

Lower orbit means lower latency and stronger signal, which means smaller cheaper terminals. It also means each satellite covers less ground and decays faster, so you need many more of them and you must keep replacing them. That is only viable with very cheap launch — which is precisely why the company that owns a rocket is the one that chose the lowest orbit.

Higher orbit means more coverage per satellite and longer satellite life, so far fewer launches. It also means higher latency and a larger, more expensive terminal. That is a sensible choice if your customers are buying wholesale capacity rather than consumer broadband.

• Starlink at ~550 km — lowest latency, most satellites, highest replacement rate, consumer market
• Kuiper at ~590–630 km — similar approach, slightly higher, enterprise and consumer, purchased launch
• OneWeb at ~1,200 km — higher latency, far fewer satellites, wholesale and government market

The ones not on the list

Telesat's Lightspeed targets a small constellation aimed squarely at enterprise and backhaul customers, with no consumer ambitions at all. China's Guowang and Qianfan programmes are building constellations at very large planned scales, driven by strategic as much as commercial considerations. Several regional and national projects are at earlier stages.

Two things follow from this. Low Earth orbit is getting considerably more crowded than any single operator's plans suggest, which sharpens every question about collision risk, debris and astronomical interference. And the astronomy concerns in particular cannot be addressed by one operator's mitigations, because the cumulative effect is what matters.

What you can track here

This site tracks Starlink specifically, and the reason is data quality: the constellation is large, its elements are published frequently, and it is by far the most likely thing to be crossing your sky on a given night.

The concepts transfer completely, though. Orbital shells, inclination setting maximum latitude, twilight visibility windows, shadow entry — all of it applies identically to Kuiper and OneWeb satellites. If you learn to read a pass prediction here, you can read one anywhere.