[sneak]

Amazon is also planning an LEO internet constellation; being dependent upon a direct competitor for core infrastructure probably isn't a good idea, especially when that specific competitor (Amazon) has been busted before abusing its position as vendor to appropriate their customers' private information and enter their markets directly as participant.

If you are going to compete with Amazon, don't use AWS to do it.

I also doubt AWS ground station locations are numerous enough to provide the kind of service that Starlink needs.

[supernova87a]

Ah, that is a good point, of course.

What kind of ground station density does Starlink need in its eventual state?

[ac29]

The ground stations I saw on the map had 2.1GHz of uplink and 1.3GHz of downlink. If we assume 4b/hz (about as efficient as LTE/WiFi), thats 8.2Gbps up and 5.2Gbps down. I'm not too sure to what extent they can use MIMO techniques, but if they can, throughput could be multiplied by the number of MIMO channels.

[GekkePrutser]

I don't think you'd have much benefit of MIMO on a connection that's line of sight like this. MIMO is all about making the most of separate connection paths through reflections from walls, buildings etc. But in this case there'll be only one path.

[WJW]

Not necessarily: you could connect to several satellites and multiply your bandwidth that way.

[GekkePrutser]

Good point, once they have enough that could work.