The question with these engines is how scalable they are. This is in terms of team size, ease of use, project size and complexity, editor performance/scalability, runtime performance, quality of presentation...
Unreal, Unity, CryEngine, Frostbite etc. have shown that they are tools capable of supporting huge productions. This means that the engine teams have invested a lot of work to remove a lot of barriers that the game teams ran up against when scaling their productions up. New users now get to profit from that when they start out on one of these engines.
As far as I know, Godot hasn't been thrown into that particular gauntlet yet (please correct me if I'm wrong here!). From experience with complex projects like this, I can tell you that you will always find new bugs when you stress them in novel ways. For a game production, this means that you take on a higher risk that you need to dive into the engine and bend it to your will, costing engineering resources.
