Losing Boot Camp and Windows emulation will be a deal-breaker for some but most people run Macs for macOS.
Currently, Windows isn't so much emulated as virtualized. At this point in modern implementations the substantive difference between the two is that virtualization support is built into all modern x86 CPUs. Only a very narrow set of kernel calls are trapped by the hardware and "managed". Normal application doing things that operate in user space run at full speed. In that user space more nothing is emulated at al.. It is the same exact execution/processing.
x86 version of windows OS and apps on ARM would largely loose most of that 0% overhead. So far there is no 3rd party Windows OS ARM images to buy ( they all come completely bundled to the device. Which is the general mainstream deployment for Windows 10. Win-on-ARM came with 10 and shows no sign of wide support for the pre-10 models of buying/purchasing so far. )
So while the standard ARM implementation has very similar virtualization support... getting a version of Windows to virtualize is a probably a stopper right now ( as Apple is extremely unlikely to start paying Microsoft for windows copies to distribute with their hardware. )
Now if ARM implementations were 10-15% than all x86 implementation, then I don't think many folks would care if they gave up 5-6% on emulation. (there would still be net positive performance increase. And why the 68K->PPC moved worked well. And the PPC->x86 did also. Neither of those emulations were done with direct CPU instruction support. ). Emulation is only a big deal if barely managing to get to parity.
I don't see this as a good enough reason for Apple to stay with Intel.
In and of itself? No. Combined with other relevant factors ( only parity in speed, no licenable Windows ARM , and/or not any less expensive. ) it could be one of several "tie breakers".
Nobody predicted Apple was going to ship the 64-bit Arm A7 chip back in 2013, it took the whole industry by surprise and forced Qualcomm to have to step up to compete.
That was entirely driven by ARM delivering up the 64 spec and far more so Qualcomm kicking that down the road. That was primarily more because Apple was laser focused on doing a Phone chip. At the time most folks wre focus on 64 bit ARM being some kind of "intel killer" in the server space. Apple looked at it as a chance to put some of the laegacy parts of the ARM instruction set in the review mirror. It was nothing about 64 bit data addressing and all about getting to a cleaner instruction set. (The vast majoirty Apple's iOS still aren't past the 4GB limit. iOS actually won't let user apps allocate more than 1-1.5GB of memory . They are actively suppressing greater than 32bit address space by user apps!!!! )
But where is the laser focus in trying to chase 2-3 desktop implementations when nobody else is. Apple's move was to do
less vacations than everyone else; not more. Part of the laser focus is to drive volume through their chip development efforts. The entire Mac line up really doesn't motivate the same level of volume at all.
This year's Mac Pro will almost certainly run on Intel but at WWDC 2019 we are going to learn all about how developers can port their iOS apps to macOS.
Really that is going to be about porting both ways now with two different underlying chip implementations. Apple doesn't require shifting Macs to ARM to carry out that plan at all. Nothing particular in Apple plan so far points to some unified single blob that is delivered to both os implementations. That was not present in their presentation at WWDC 2018 and that isn't likely to change in 2019. It doesn't "have to" change in 2020 either.
There is very little in the talks about Apple using ARM in macs that points to it being a rapid 100% wholesale dumping. The previously transitions were "big bang" like but there is about zero that definitively points to that now.
The major moves in the desktop space are moving opposite of what Apple has been doing over last 5 years of trying to integrate the whole system into one chip with their ARM SoC. Ryzen is GPU less ( Xeon workstation and AMD workstation/server already were ) Apple's chips are fundamentally pull the GPU in 'closer' to the CPU memory system. Apple has done about zero work here. There is an ARM server targeted version they could license as a jumps start, but it isn't a great fit for single user desktop. And it is still largely unmotivated by volume and economically. It might be a way to fart a sizable chunk of the money pit holding down the drain..... but better return on investment as several other more rational approaches ... probably not.
Once Macs run on Arm processors and the big apps get ported Apple will allow users to run unported iOS apps on their Arm Macs and this will become a selling point differentiating the Mac platform.
Apple has said nothing about unported iOS running on macOS at all. Zip, nada, nothing. (besides what already happens in XCode. )
[ ChromeOS is trying to drag in lots of Android apps but that is primarily because ChromeOS has about zero native apps. macOS doesn't have that "problem". Apple spending tons of effort to bring over the flotsam of iOS apps to macOS is not really a significant value add at all. Folks who wan't to put in the effort to port the right way... yeah but those are ports. Lazy folks who just want to dump a mismtached UI onto a window on the macOS screen..... macOS needs that like another hole in head. It isn't going to help much at all. ]
As far as this plan being "several years off in the future" it's more like 2 years away, it's coming, and sooner than you think.
To the MacBook in 2 years or so? more than a good chance it could be. To the iMac Pro , iMac , Mac Pro in 2 years or so? highly likely not (at least using Apple chips. ).
P.S. IMHO there is a higher chance Apple would get into a either a cellular modem and/or discrete mobile GPU before trying to do a desktop focused CPU. Most likely they are just going to throw the iPad Pro chip over the fence and do a subset of Mac laptops with it.... and that will be about it.
