Netflix takes the world by storm, grows quickly, and becomes reliant on AWS. Amazon enters the video streaming market as a much weaker competitor and builds upon their own tooling.
In this scenario, I presume Amazon is making more money from Netflix using their service than they are from running their own streaming service. Their first goal is to not disrupt the money they make from Netflix and their second goal is to create a competitor. Later, if the math says that hardening their contracts to the detriment of Netflix would allow them to capture such market share as to recoup the investment with their own streaming service, they would do so, but as they're late to the party that is not the case.
In the Starlink scenario, it's the other way around. Starlink is already dominating the market using their own, SpaceX tooling. The incumbent, AST SpaceMobile, is allowed to use SpaceX rockets. It's not hard for SpaceX to disrupt AST SpaceMobile's operations without hurting SpaceX's bottom line as there's a backlog of launch demand to replace AST SpaceMobile. If the math says that AST SpaceMobile is likely to begin encroaching on Starlink then they deny a contract, replace it with the next person in line, and don't bat an eye.
The order in which things occurred between the companies means there aren't these deeply entangled, mutual benefits that allow competitors to coexist.