Moving the Mac to ARM

Speculation about Apple transitioning some or all of its Mac computer lineup from Intel to Arm is running hot and heavy again. I've already written all about how Steve Jobs' famous keynote announcing the previous PowerPC to Intel transition could be replayed almost note by note for an Intel to ARM transition. So, now I want to touch on what happens next: The transition itself.

MacBooks

A decade and a half ago, Intel was already making commodity x86 chips for everything from laptops to desktops to workstations. In other words, everything Apple needed.

Right now, publicly, Apple is only making custom ARM chips for tablets, phones, and even lower power devices.

Sure, those ultra-mobile tablet chips are, as they say, screamers, and in terms of power-efficiency put everything in the same class to utter, desolate shame, including and especially Intel. But, they're still ultra-mobile chips. So, as is, they could probably power a new MacBook or MacBook Air really, really well. With incredibly fast performance, taking full advantage of neural engines, encode/decode blocks, accelerators, secure enclaves, and all of that, as well as phenomenal battery life.

Same for a Mac nano, or whatever Apple would call the box that would blend the very best characteristics of the Apple TV and Mac mini.

In fact, that story tells itself so well it's not hard to imagine Apple forgoing any awkward developer boxes, like during the Intel transition, and just announcing a Mac ARM SDK that developers can run on existing iPads Pro or Apple TVs to get their apps ported over with way less fuss and much higher availability muss.

MacBooks especially avoid a lot of the software transition pain as well, since most of the people who buy them aren't doing so to run DaVinci Resolve or Adobe After Effects, Pro Tools or Maya.

Office was one of the biggest pain points back then but now Microsoft is busy working on Windows 10 X — 10 10? — and its own ARM transition, and there's Office Online, and GSuite, and, hell, Apple announced iWork for ARM back with the original iPad in 2010. Office for iPad has been around for years already as well and so have a baker's dozen of Adobe apps, and so many indie apps that make such incredible use of Apple's Core's and Kits and Metal that, with Catalyst now and SwiftUI soon, could make more than enough tools available for anyone using an ultra-mobile ARM-based Mac going forward.

Especially since Apple hard killed 32-bit Mac apps this year, greatly reducing what would have to be ported, much less emulated, over.

MacBook Pro & iMac

The MacBook Pro and the iMac are more interesting. Here, people would not only want but need their Xcode and Final Cut Pro, and all the pro apps.

Johnny Srouji and Apple's platform technologies org could have designs and architecture already all laid out for the level of silicon needed to support them.

Apple's in-house Pro apps could be done and ready to go at launch. Others, like Adobe's, might take a more annoying amount of time and/or emulation.

Apple could simply flip the table on the current status quo: Instead of Core i7 or i9 with a T2 co-processor, they could have a T7 or T9 with Intel deprecated down to the co-processor slot. But, unless there's a strict phase-out schedule announced upfront, that might make compatibility faster but the transition take longer.

Just look at what happened with the decades-long 32 to 64-bit transition.

Then there's the question of GPU. Apple is already making its own graphics cores for mobile, and modern Macs already handle dispatch between Intel processors, ARM co-processors for things like H.265, and AMD GPU. I assume they'd work just as well with Intel demoted or removed from that chain. Especially since, unlike Nvidia, AMD is happy to let Apple work down to the metal… and with Metal. Currently, the Metal 2 framework, as an abstraction layer to make processing less a bunch of separate silicon and more a unified set of resources to be targeted on a task-by-task basis.

The Mac Pro and iMac Pro are, of course, the most interesting. Apple just released the brand new Mac Pro. And, because it's so modular, it could easily last a decade. That means the workstation could have both the biggest challenges but the least immediate pressure.

Again, Johnny Srouji's team could very well have massively multicore ARM-based blades ready and chomping at the literal bits. And, given the modularity of the Mac Pro with everything from how it handles GPUs to custom accelerator cards like Afterburner, Apple could again flip things around and offer x86 on a card. For as long as anyone needs it.

They've also worked around their loggerheads with Nvidia by getting more and more high-end app makers to port to AMD for the Mac Pro, so maybe even a workstation transition won't take as long or be as painful as it might otherwise have been.

Yes, Apple absolutely mismanaged the Mac from like 2015 to 2018, took too many wrong turns, and sacrificed too many resources to more popular products. But it's also fair to say many of the products that did ship were delayed, constrained, and compromised far beyond reason by Intel's years and years of mission roadmaps, failure to die shrink, pushing off feature implementation, and otherwise doing the exact opposite of why Apple transitioned to them to begin with.

And that's what the ARM transition will fix — by giving Apple the one thing it lacked on the Mac, the same thing that's made it so successful with iPhone and iPad — control of its own silicon destiny.