Its less about the instruction set and more about the specific implementation choices and market they're competing in. Apple has been doing ARM SoC design for a long time and was an early 64-bit adopter, buys tons of bleeding-edge capacity from fabs, and is willing to design in more expensive features like tons of onboard cache that they can make up on margins in other ways because Apple sells complete devices, while Qualcomm etc are mostly in the business of selling chips. So while there are fundamental differences between x86 and ARM, the bigger differences are more "Apple vs other ARM SoC designers" than something fundamentally different about ARM that Apple takes advantage of.

1 - The instruction set is more interesting because it is smaller.

2 - Arm also implements weak ordering which is type of memory model. It allows instructions to be separated into groups based on whether they affect other instructions. This allows these groups to skip some waiting lines. (Magic). X86 has strong ordering, which can make it slower in specific scenarios.

If the company is selling just the chip, it's in their best interests to hold the die size down as much as possible because doing so directly effects their profit margin.

If the company is selling the whole device, including the chip, they can afford to throw more transistors and die area at performance with a much more minimal impact on their bottom line.

TSMC tech helps a lot, but is not exclusive to Apple. A relatively modern and sensible 64-bit ARM architecture helps too, but on top of that Apple seems to be really good at designing the chips.

OTOH Qualcomm is constantly lagging behind Apple, promising that next year they will be almost as fast as a previous-gen Apple chip. The most generous explanation I've heard is that Qualcomm is specializing in quantity over quality, and focuses on making CPUs cheap, rather than fast.