BTW, I want to give full credit to my source: https://www.youtube.com/watch?v=giQ8xEWjnBs
It also does a much better job of explaining it than I have but it is a convincing argument for how Starlink can be useful and why traders would be willing to pay a lot for it.
That is nonsense, the route is longer. You are confusing stuff, light travels faster in air compared to fiberoptic cable, so the latency will be lower, but distance is actually longer.
Also they're only orbiting about 200 miles above Earth, so unless you're already relatively close to your target location, you'll need to bounce your traffic across a whole bunch of satellites, each adding a fractions of a ms just for hardware processing + light speed to the next hop. In fact over any sufficiently long distance, doing this will actually resemble a longer curved arc circumscribing the arc of the Earth.
People who really need those extra few ms have little choice but to stick to the ground, near the exchange.
(I recommend reading the whole series, it's fascinating!)
The tower location does not have much of an effect on latency because the microwave network in this case always terminates at NYSE. If you build your microwave tower a few feet closer to the transmitter (Chicago), then you have to run a few more feet of cable to the colo.
[1] https://www.bloomberg.com/news/features/2019-03-08/the-gazil...
"Everyone" is pretty questionable here - it's quite expensive to rent rack space in an exchange, and not everyone gets to do it. That said, I think basically anyone who has any business doing latency-sensitive trading is probably capable of getting a rack.
Of course, that price isn't insignificant per se, but are minimal for anyone who actually achieves any benefit from having co-located space.
(I know you meant server colo but I wanted to point out that this is yet another case of the back and forth swing Of distributed and centralized)