It will be to split the "MacOS" ecosystem in 2 and maintain two branches for x86 and ARM. And while they also have a team of pretty good software engineers with x86 and ARM experience they are already having major quality control problems lately. Having to maintain a dual ISA MacOS will be a challenge. Then they have to provide the legacy compatibility in order to convince people to buy into something with little to no support for most of what's out there today.
But you know... we could use some diversity in this space. I wish them all the best.
The problem is that as established as ARM is they really don't have a meaningful presence in the desktop/laptop ecosystem. You have probably tens of thousands of apps for the mobile space but even if Apple magically adapted them for a non-touchscreen, keyboard and touchpad world the vast majority still won't cater to the regular or professional user wanting to buy a laptop, not a glorified tablet. And that's a hard sell. Why am I buying a MacBook if all I get is an iPad with a keyboard? And don't get me wrong, having used both iPads and CoreM ultraportables (Apple or otherwise) the iPad offered the better experience more often than not. But if I compare it to the regular CPU it just can't compete. The productivity with mobile apps is not where it should be to convince a normal buyer to switch, let alone a more demanding user.
Yes. Not established. That’s what I said.
Look at another way - Apple moved from PowerPC to Intel. It's not that Motorola ISA weren't well established but they simply weren't as well established and didn't have access to the same mindshare in terms of high performance hardware and software engineering.
So in this sense it's a step back the other way - except for perhaps Intel seem to be stuck in the mire these days.
They've done that twice before (68k -> PPC, PPC -> Intel), and IIRC, seem to be preparing to do it again (Bitcode).
I think the biggest risk for Apple is that they may not be able to keep up with x86 on the desktop performance front. They have a much smaller market to amortize the R&D over than Intel does.
(Fun fact: even classic Mac OS was running on Intel at one point before the 680x0-to-PowerPC transition: https://en.wikipedia.org/wiki/Star_Trek_project)
So there's good reason to think that they could pull this off.
What could possibly go wrong!