Why Apple won't dump Intel x86 for its own ARM chips in MacBooks and the Mac Pro

A recent article penned by former Apple executive, BeOS founder, and PalmSource executive Jean-Louis Gassée has stirred the internet by predicting that Apple could break up with Intel at any time and swap its entire product line over from x86 to ARM-based chips by using future iterations of the A7 SoC.

It's an interesting argument, and it brings up some good points, particularly given that Intel charges Apple a high premium for the laptop chips it uses -- from Apple's perspective, it could pocket that revenue itself. Unfortunately, the article -- published by Quartz(Opens in a new window) -- also reprints some laughably untrue assertions about the ARM-x86 comparison, writing "The aging x86 architecture is beset by layers of architectural silt accreted from a succession of additions to the instruction set. Emerging media formats demand new extensions, while obsolete constructs must be maintained for the sake of Microsoft’s backward compatibility religion."

It's a pretty story, but it's not true. Independent research has demonstrated(Opens in a new window) that the power consumption differences between ARM and x86 come down to architectural choices, not intrinsic inefficiencies in the ISA. Furthermore, power consumption at the device level is driven by multiple components, not just the CPU.

In an exclusive interview with ExtremeTech a couple of years ago, Intel's then-mobile chief Mike Bell clearly stated that the concept of an "x86 tax" simply isn't true. “There is nothing in the instruction set that is more or less energy efficient than any other instruction set,” Bell said. “It’s all about the implementation and the process technology; whether you target power, or speed, or both."

None of this means you can't build an ARM processor that's more efficient than the x86 competition -- but it means that doing so will be a function of the chip's design or its process technology rather than an intrinsic property of the proposed instruction set. And yes, Apple's newest A7 Cyclone SoC is a beast -- but it's a long, long way from being in any way competitive with Intel's Core chips in terms of performance.

That aside, there are three good reasons why Apple won't adopt ARM across its entire product line anytime soon -- and one good reason why Intel should be worried about the long-term roadmap.

Next page: Three reasons Apple won't dump x86 for ARM any time soon

Three reasons Apple won't adopt ARM for OS X any time soon

The chart above shows the ratio of Windows PC sales to Mac sales for the relevant periods. It peaks in 2004, drops slightly when the G5 is introduced, and drops more sharply when the Intel changeover is announced in 2006 (follow the lightest green area to see the Mac-PC comparison area once the iPad and iPhone are also introduced(Opens in a new window)). Data backs up anecdotal evidence -- once Macintosh systems could run Windows, more users adopted the Mac platform.

Absent a major sea change from Microsoft governing how Windows RT is adopted and sold, any shift to ARM-based Mac hardware is going to fragment the user base. Buyers who view Windows compatibility as necessary may make a distinct minority of Apple's total revenue stream, but Windows remains extremely important in corporate IT.

Apple's ability to shift to a new 64-bit approach for OS X was increasingly hampered by a 32-bit last-generation CPU with no serious reason to believe the situation would change in the near-term future. Intel's failure to offer a mobile chip for iOS isn't really equivalent -- not considering that the tablet and PC space continue to serve customers with different needs.

It's worth noting that Apple didn't just leap for Intel when Intel was competitive -- benchmarks from 2006(Opens in a new window) show the new Intel Mac Pros outperforming the old G4 PowerBooks by up to 63% in single-threaded tests and by nearly 30% in certain Quicktime encoding benchmarks.

Much of the discussion around Apple's long-term plans revolves around the idea that its A-series SoCs have proven themselves sufficiently advanced to make the leap wholesale from iPhone to MacBook to Mac Pro -- but this ignores the way that Apple historically operates. It didn't split the OS X family between two architectures previously and if it makes such a shift again, it'll do so across the entire product family. This is where we hit the sticking point -- regardless of whether Apple can design a chip to replace Intel at every SKU, it hasn't done so yet.

Before Apple can build SoCs that take on the top of Intel's product stack, it's got to ramp up support for multi-core products, drivers, on-die I/O, PCI Express, and a host of other features to ensure that its product family can scale from the iPhone all the way to the Mac Pro. The question of whether or not Apple is capable of making that transition is different from whether or not it can do so quickly (or if it makes sense to do so in the first place).

And the one reason Intel should be paying attention

The one reason why Apple might seriously build its own ARM core for high-end computing is that such an approach would give it more control over its product family, its roadmap, and its future -- and JLG isn't wrong that yes, historically, Apple strongly prefers that control.