They also have the most silicon to work with. If you are designing a chip with x resources vs y resources its kind of unfair to say one design is better than the other. One design is optimized for one thing vs another. Arm needs to sell to wide set of customers who might be more more price sensitive than apple. If I could create a chip with more decoders, alu units, cache, etc. I can design a faster chip, but it wouldn't necessarily mean I am a better designer than an engineer who works at company x.
It's sort of a little bit of both. Like for sure their raw gate count is a huge part, but if you go look at a bunch of their acquisitions (particularly looking at Intrinsity here https://en.wikipedia.org/wiki/Intrinsity ), they have a lot of really cool techniques for making their designs count even gate for gate. Intrinsity's work is neat, because they didn't really make any uarch changes or increase the gate count really, they just made their Cortex-A8 core way better which lets you judge the work in an apples to apples (lol) way that you normally don't get.