Wonder if they will manage to keep it up with GalCiv4 around the corner.
Indeed. And let's not forget Distant Worlds 2, either. Granted, Slitherine doesn't have the same marketing acumen (or resources0 as Paradox, or even Stardock, but I think it's still going to make a bigger splash than its predecessor did.
My gut feeling is that ultimately, Stellaris will be able to stay ahead of the others, if only because it has Paradox's marketing machine -- and gaming community -- behind it. So long as Moregård and his team manage to release at least one new expansion a year, I think that will be enough to maintain Stellaris' current hold atop the space 4x sub-genre.
...I won't be upset if I'm proven wrong, however.
I don't think there's so much more they can improve.
I can think of quite a bit, myself.
* Factions still don't do much. (There remains much clamoring for a DLC that focuses on internal politics, and possible civil wars.)
* Sectors still suck, and badly need a revamp.
* There are lot of ways in which wars still don't work very well.
* Leaders aren't truly meaningful, unlike in Total War or Crusader Kings. That's something a lot of folks would like to see changed/improved.
* Despite the 2.6 update and Federations DLC, there are a lot of ways in which diplomacy still doesn't really work like players wish it did.
* Including Federations. It's clear they still don't really function as how anyone -- players or developers -- really would like.
* Although the 3.0 update and the addition of the intel/spy system helped, the mid-game is still rather boring.
* Towards that end, The Great Khan needs to be fixed/improved, and more mid-game crises/events added.
* More end-game crises would be welcome as well.
* Also: Why do all crises have to be military in nature? Why can't we have crises like a galaxy-wide plague or nanite storm -- one that requires a certain amount of research points, Unity, etc., that would require multiple empires combining their resources in order to successfully defeat? Not every problem can be solved at the tip of a laser cannon...
* And of course, the AI can always, always stand to be improved. There remain far too many examples of it doing inexplicably foolish things.
And these are just the things I can think of off the top of my head. Doubtless there are many more issues (some more fundamental than others) that could/should be improved upon.