A poor workman & his tools
The biggest challenge in game development, I suppose, is simply finishing something. I've certainly stumbled across that particular impediment repeatedly. An apt saying goes "it's a poor workman that blames his tools", and yet I'll admit that the second biggest challenge I, personally, have had is simply choosing which tools to use. I know Gamemaker Studio 1 & 2 (GMS) quite well from years of practice with it, and I'm comfortable with its proprietary Gamemaker Language (GML). It doesn't quite fit my brain, though, and I struggle to keep the right pieces of a project in mind as I'm working. It also has a rather unfortunate business model in comparison to engines like Unity and Unreal, both of which offer more for free. At this writing, GMS requires a not insubstantial $400-500 to export to mobile platforms, then yearly fees above that for each console one might want to target. And those consoles usually have developer fees as well. I'm not doubting the effort that goes into maintaining and updating a virtually seamless export path to dozens of platforms. But then these other guys, who are pretty much industry standards now, are doing so without charging small indie developers and hobbiests at all.
I'm currently working with Unity to see how well I adapt to it because of the cost issue, as well as its use of C# (an attractive feature for me), and the fact that I have a couple of projects in mind that would benefit from its 3D (and optional 2D) tooling. However, I think that it shares with GMS a certain IDE-ness that may be part of my problem. Despite (or perhaps because of) a pronounced lack of tooling, I think I was actually more productive coding in Javascript in Notepad++. Look, I'm not trying to suggest that "real men" code in a text editor or anything, but juggling blocks of text, confined to seperate module files, seems to suit me better than GUI windows, toggles, and buttons, especially when working in a more object-oriented language that simply "clicks" with me. There are game engines and frameworks that use Javascript, of course, and ultimately maybe I should go that direction. I'm also interested in Love 2D because I want to work with Lua a bit and it lends itself to this code-editor kind of work flow. The Monogame framework is another option that might suit.
The answer, really, should be "whatever allows you to finish." Successful indie developers like Concerned Ape (of Stardew Valley) do seem more focused on completing a vision than porting it; in fact, Star Dew Valley received ports after it was wildly successful. Which sort of argues for continuing with GMS, I suppose, but I'm also leery of the sunk-cost fallacy and pouring more time into an engine with few upsides other than that I know it.