Yah, this is something really worth discussing. I have my reservations about SpriteKit even though I'm practically married to it at this point. I did two releases within the last half-year, one of which was a major one - and while I'm happy SpriteKit served me well with these, it did take a long time to get it to do what I wanted. I'm talking years here. Also, support and peer discussion seems pretty lacking.
As SpriteKit is highly inspired by Cocos2d, it was a natural transition for me having worked with Cocos2d for a couple of years. I've also been an iOS developer from the very beginning, 10 years in total, so I love that I can use the same tools (Xcode) and keep to the same language (now Swift). In short, being in the iOS ecosystem with SpriteKit allows me to make simple games with potentially great returns while at the same time strengthening my career in UIKit. So if I choose to go to a high-profile UIKit job, all this counts as experience. It works for me since I'm not likely ever going to join a game company (benefits are worse, hierarchy is greater, crunching is practiced, etc.), and would rather join an app company if I get bored of entrepreneurship.
However, if I were planning to go fully into games, I would wholeheartedly pick Unity over SpriteKit. The cross-platform support alone is enough to choose it when we're talking about games that are independent from iOS practices (most of them). In Unity you can now even export to Nintendo Switch, let alone all the other consoles and handhelds, so I feel like nowadays iOS exclusiveness needs to be argued well (because there are certain cases where being an Apple darling is great).
A good example of this to me is GNOG (https://www.gnoggame.com/). It's very experimental, artistic and just great, and it represents this sort of artistic indie branch of gaming that at least attracts me. GNOG would do great as an iOS exclusive and could easily be Apple's favorites (in fact it kind of is), and you could build it with SceneKit if you wanted to. However with Unity they can have PS4, Nintendo Switch and everything in-between. Even if the increased financial market isn't the benefit you're looking for, it also just expands your userbase to a whole other type of people and communities. And as a VR-able game, GNOG can support AR on mobiles and VR on consoles and PC.
Then there's content creation. If anything, Unity beats SpriteKit/SceneKit in this. If you're interested in doing ARKit, you'll likely be working with 3D. Setting up an AR scene isn't so terribly difficult in SpriteKit/SceneKit, and the re-emergence of AR with ARKit gives you a market. However, AR is only the enabler and it's now "hot" and open to everyone. In the future it's not going to be the hardware that you compete with, it's going to be what kind and quality of content can you provide. Already now, the first boost of ARKit has passed, users have tried the gimmick apps and will need content next if they plan to stay interested. Ask yourself: what development platform should I choose when my main goal is going to be creating an immersive, rich 3D experience, with perhaps a progressing storyline, sophisticated AI, game mechanics, plugins, etc.? Unity, of course.
At the end it depends on what your goals are. If you want to do small apps and you happen to like being native, lowish-level and create systems from scratch (I do), then SpriteKit is fine. However, if you wish to reach more people, create bigger experiences, focus more on content than systems and/or target a job in a game company, Unity is a better choice by far. And I'm not knocking SpriteKit. After all, my apps do provide me my living. I also now have a fantastic template for a game project along with plug-and-play classes for SpriteKit (after years of toil), which makes developing apps super-easy. However, if I could go back in time to 2010, I would tell myself to just go with Unity over Cocos2d to avoid years of pain. Even now, switching to Unity (starting from scratch) might be a reality for me in the next few years, since I'm getting quite a lot of pressure from partners to be cross-platform and have an easier development platform/environment.
Sorry for the long diatribe 🙂. I hope this is helpful and I sure do welcome any rebuttals 😀.