I remember the Motorola to Intel transition very well. I remember it left some scar on many Mac users, some hoarding the G5 Mac Pros well after usefulness due to software. I'm not sure if OSX marketshare is "drop in the bucket". I think it's quite significant. If you treat Apple's OSX computer sales as a PC ODM, they are the biggest one by far. It's nowhere the success that iOS is, but it is still pretty big. Big enough that Intel would concoct a special CPU SKU just for Apple when asked. And I also don't know about X86/64 reaching any limits. The progress has reached limits of Moore's law, but that fate will also fall on ARM as well. Only Intel really has been on the bleeding edge of Moore's law train all this time if you think about it. ARM was late to ride the train and Apple is the only one to really push the limits of the ARM SoC in transistor count. Now that Apple as gone over 3 billion with A9, they also have stagnated in transistor count with A10 and A11, relying mostly on architectural and clock improvement for performance increases last two iterations. Intel will do the same going forward, but they have slightly more room for tacking on transistors without getting too big for decent yields and profitability. And the fact remains that Intel's 2 billion transistor mobile processor outperforms Apple's 3 billion transistors convincingly due to being able to scale to 4GHz+/45W without melting down. I don't know how much you can push the thermal envelope in A series architecture when 2.4GHz/8~10W seems to be about what Apple is comfortable with even in battery-less applications like Apple TV. Granted, a big chunk of the 3 billion transistors are turned off for sake of battery life architecturally but that should allow for higher clock in Apple TV. Apple TV should be a lot more like nVidia Shield TV, where focus on games are big, but it's just another 4K HDR box with light gaming, like any other sub $100 Chinese quad core A53 boxes.