A Wholesale Apple Transition To ARM: On This ‘Forecast,’ I Wouldn’t Bet The Farm (Or Even The Barn)

A Wholesale Apple Transition To ARM: On This ‘Forecast,’ I Wouldn't Bet The Farm (Or Even The Barn)

I am, lest there be any doubt, a member of the technology press. Unlike seemingly 99.9% of the technology press corps, however, I actually have a educational-plus-professional technology background; a BSEE from Purdue University, to be precise, along with three years as a working engineer at Magnavox's Government Defense Systems Group, followed by eight years at Intel Corporation. Not to mention almost 15 subsequent years-to-date as a technology analyst and journalist.

All of which gifts me with an abundance of informed perspective. Not to mention no shortage of shame for my peers. For, just as I suspect is the case for some lawyers, some politicians, and some used car salespeople, I'm sometimes embarrassed to admit how I earn a paycheck. Now is one of those times. For recently, cyberspace has been awash with claims that Apple is poised to convert its entire product line from x86 to ARM CPUs. Give me a break.

The so-called “evidence” seemingly comes from three “sources.” An interview published online at Wall Street Journal on August 10 ( http://blogs.wsj.com/digits/2011/08/10/intel-sets-300-million-fund-to-spur-ultrabooks/) included the following quote, referring to Greg Welch, director of Intel’s Ultrabook group:

“Welch said Apple informed Intel that it better drastically slash its power consumption or would likely lose Apple’s business. “It was a real wake-up call to us,” he said.”

Three months earlier, Barclays Capital hardware analyst Ben Reitzes had offered some “suggestions” to PC makers struggling with poor profit margins ( http://blogs.barrons.com/techtraderdaily/2011/05/09/aapl-might-switch-to-arm-says-barclays-so-should-dell-hp-too-chrome-anyone/):

“Shift from Intel (INTC) microprocessors to chips based on ARM Holdings (ARMH) designs. Those chips might cut processor cost by a third, which would save $25 per PC, 5% of the total cost of the machine, and would add 55 cents a share to HP’s annual profit and $0.45 to Dell’s profit, roughly 10% and 20%, respectively, of their total annual profit.

“Switch from Windows to Chrome. Microsoft’s (MSFT) software is $75 per desktop and notebook, on average, for the vendor, and $50 for the average consumer PC. Switching to Google’s (GOOG) “Chrome” OS would save $45, 10% of the bill of materials. That would add $1 extra in profit per share per year for HP, and add 80 cents to Dell’s annual EPS.”

Specifically, Reitzes prognosticated, “We believe that Apple will be the first in our sector to embrace ARM for some Macs, as early as C2H12.” And a few days before that, long-time industry gossip-monger Charlie Demerjian, currently at SemiAccurate and previously for many years at The Inquirer, published a writeup with the provocative title “Apple dumps Intel from laptop lines,” and containing lines such as the following ( http://semiaccurate.com/2011/05/05/apple-dumps-intel-from-laptop-lines/):

“Word has reached SemiAccurate that Apple is going to show Intel the door, at least as far as laptops are concerned. It won't be really soon, but we are told it is a done deal. The short story is that Apple is moving the laptop line, and presumably desktops too, to ARM based chips as soon as possible.”

Although I'm admittedly of the opinion that Demerjian is prone to link-baiting (translation: saying provocative things predominantly-to-completely to drive fiscally lucrative traffic to his advertising-flush website), I also don't underestimate the breadth and depth of his industry sources. As such, I suspect there's at least a nugget of truth in his words. Apple likely has developed an experimental MacBook Air ultra-thin notebook prototype (http://www.macotakara.jp/blog/index.php?ID=12774), and the company may even decide to bring it to production. After all, the lack of a physical keyboard is a notable complaint that some folks seem to have with the ARM-based iPad, and an ARM-based “laptop” in a MacBook Air form factor (complete with a touchscreen-inclusive display) would neatly address that particular issue. Not to mention the fact that such a product would lock users into the lucrative-to-Apple iTunes Store distribution scheme…and that Apple has already paved a path to this likely scenario by giving latest-generation Mac OS 10.7 “Tiger” iPad-like user interface features such as the App Store, multi-touch support, compatible-direction scrolling capabilities, and the Launchpad.

More generally, the Apple-on-ARM believers point to two fundamental historical data points in rationalizing their futuristic claims:

• Apple had previously transitioned from PowerPC to Intel x86 CPUs, beginning in June 2005 and completed with the production of Intel-based Xserve rack units in December 2006 (http://en.wikipedia.org/wiki/Apple%E2%80%93Intel_transition), and

• Apple had subsequently migrated Mac OS X to ARM in developing the memory footprint- and ARM-optimized iOS, which initially appeared in the January 2007 first-generation iPhone

All true. But let's look at them in turn. Intel CPU support first appeared in mid-2005 Mac OS 10.4 “Tiger”; along with it came “Rosetta” emulation technology that enabled running PowerPC-compiled binaries on the Intel-compiled O/S variant and associated hardware. PowerPC support within the O/S itself didn't disappear until four-plus years later, with Mac OS 10.6 “Snow Leopard,” which still retained Rosetta support (as an optional installation) for PowerPC-compiled apps. Only with just-released Mac OS 10.7 “Lion,” which started shipping in late July, did Apple completely close the door on its PowerPC past…six-plus years after the transition process publicly began.

Translation: CPU conversions are a big, expensive, time-consuming deal, for Apple and its developer partners alike. Speaking of which, what about that Mac OS X-to-iOS transformation? First off, to be clear, it represented an expansion, not a conversion like the previous example, since the ARM processors running iOS went into new products (the iPad, iPhone, iPod touch, and second-generation Apple TV) versus new versions of existing products (the MacBook series, Mac mini, iMac, Xserve, and Power Mac-now-Mac Pro). And secondly, anyone who has run an iOS-based product is already familiar with its notable, numerous functional limitations versus a full-featured computer, both in terms of the O/S itself and of available applications running on it.

Apple and its developer partners have made notable progress in closing the iOS-vs-Mac OS X gap since early 2007, but ubstantial disparities remain, in some cases by Apple's choice. Take Adobe Flash support, which Apple seems to remain stubbornly determined to keep off iOS. Long term, HTML5 cognizance within browsers will probably take over some portion of the content delivery burden currently handled by Flash, although DRM'd material may not be able to make the leap. To that point, anyone who has tried to surf the Web using an iOS-based product already knows how pervasive Flash usage is. I'm reminded of a saying; while people admittedly tend to be too pessimistic in making long-term predictions, they also tend to be too optimistic in the short term. Such is the case, I suspect, with the HTML5 fan club. And for this and innumerable other reasons, it's ludicrous to believe that a notable percentage of Mac OS X-based computer users will tolerate a forced migration to iOS-based hardware, even of a comparable form factor.

When Apple finally decided to make the jump to x86, it was faced with an intractable performance-vs-power consumption problem; PowerPC CPUs required expensive liquid cooling subsystems to achieve even a reasonable fraction of the capabilities of Intel counterparts, which (among other things) made PowerPC an infeasible option for the booming mobile computing market. That's not the case today. ARM processors arguably deliver more attractive power consumption profiles, but at a performance deficit versus x86 alternatives, and nothing I've yet seen suggests that this discrepancy will notable shift in ARM's favor in the years to come. If anything, the converse may be the case.

So does Apple regularly inform Intel that “it better drastically slash its power consumption or [it] would likely lose Apple’s business?” I have no doubt that it does, but then again what smart customer doesn't periodically threaten its primary supplier with alternative sources? And does Intel know that, just as it was the beneficiary of a past CPU conversion, it might theoretically lose out to another conversion in the future? Of course. But far more likely, I believe, is that Apple might broaden its x86 supplier base beyond Intel to AMD and/or Via. And anyone who is convinced that a wholesale conversion from x86 to ARM is, like those who believe that Microsoft Windows will sooner-or-later be usurped by Linux (specifically the Linux-based Google Chrome O/S), fueled predominantly by distaste for market segment share leaders, not by any sort of technical competence-therefore-insight.

You're regularly the recipient of all sorts of “information” delivered by all sorts of “sources.” Admittedly, some of it might be entertaining, in a National Enquirer sort of way. But before you act on that “information,” do some serious research on the “source's” credentials. What you'll frequently find, I suspect, will substantially underwhelm you. Instead, I'd humbly-as-possible suggest that you prioritize sources with credible engineering credentials and lengthy technology journalism careers (therefore perspectives), such as those of us here at Electronics Products .

Brian Dipert

1

editor