Absolutely! In fact, support for “more software” had very little to do with it. Other factors were far more critical.

The AIM Alliance

The AIM (Apple, IBM, Motorola) alliance was founded in 1991 as a collective devoted to building an industry standard architecture for PowerPC. Unlike the ad-hoc standard of the IBM PC, this was planned from the get-go to be Apple compatible but also to include features to make support of both hardware and software easier. They were very much teaming up to go head to head against Intel and Microsoft.

The AIM alliance intially got other companies involved in the Mac-compatible hardware business. The first of these were Mac “clones” in a very real sense: many of these used the same Motorola-made modular motherboards used in some Macintosh computers. I had a startup company, PIOS AG, based in Hildesheim Germany back in the day, making this sort of computer. In fact, we had the first 300MHz Mac-compatibles shipping anywhere.

Early on, the AIM alliance agreed on a standard called PReP (PowerPC Reference Platform) for compatible motherboards going forward. The goal was to be able to allow everyone to do their own designs, rather than relying on using Apple-derived designs. This is the original version of the PIOS One main board.

The advant of Mac “clones” pushed MacOS compatible systems to over 6% of the desktop PC market.. but Apple’s already small business shrunk a bit more. I wasn’t paying all that close attention to this small fact, but on a visit to Apple in early 1997, I did get a bit of an unsettled idea. Apple was reported to be having problems, financially, but you’d never know it from the people and excitment in the open architecture group. Oh, yeah, by then, Apple had decided they were just kidding about the PReP specs, and instead everyone had to revise their designs to support CHRP (Common Hardware Reference Platform), or perhaps PPCP (PowerPC Common Platform).. but hey, those were at least the same thing with names that jumped around a bit. They even had a Motorola Starmax there, runing MacOS, which was touted as “the fastest Mac we ever tested”.

Uh-oh.

In the summer of 1997, Apple brought Steve Jobs back as CEO. With decades of hindsight, it’s obvious SJ not only brought Apple back to health, but made it the world’s most successful company. Of course, that didn’t happen with the Macintosh. But SJ also managed to kill the PowerPC that summer, though it took awhile for everyone to understand that’s what happened. He did this, clearly with a team of lawyers at his side, by re-naming MacOS 7.7 as MacOS 8.0, which apparently let Apple completely renegotiate MacOS pricing. So they were going to jack the price way up to, well, make up for the fact everyone else made better CHRP computers (at least as I saw it). Which effectively killed any reason for anyone to make a Mac-compatible.

There were other uses for the PowerPC out there in the world. Cisco was using older desktop-class PPCs in their high-end network switching gear, and in fact, selling to those guys after Apple lost interest was pretty much how Motorola, at least, justified making PPCs at all. Motorola had full lines of embedded CPUs, so they moved higher-end embedded with spinoff processors. Intel made some realy low-end parts and got some success in places like ABS Brake systems in automotive computers. But there was no other customer for desktop-class PPC. Not really. Motorola made a go for a little while, selling StarMax computers for use with Windows NT. But even Microsoft didn’t bother with support of their tools on non-x86 hardware in those days.. they wanted someone to pay for that port. Just as Motorola was paying for the NT development.

Speed, Bits, and Going It Alone

After Apple became the only company making PowerPC desktop computers that run MacOS, their sales went up. However, the overall sales of MacOS PCs went down. And it was never quite as healthy as it might have been. Even SJ saw this, and certainly gets credit for diversifying, first with the iPod in 2001.

But back in 1997, at least part of that was that Apple’s downright treason against both IBM and Motorola’s intents on growing their desktop CPU business and delivering systems that could run MacOS didn’t just affect those business units. It made PowerPC much less important to both companies… in fact, in this time, Motorola has jettisoned their whole chip business, to become the stand-alone Freescale. They didn’t have enough incentive to keep boosting the PPC750 (G3) and PPC7400 (G4) architectures. Cisco wasn’t demanding faster CPUs… they wanted faster I/O.

So Apple had to push… with money. IBM announced the first 64-bit PowerPC for desktops in 2002, the PowerPC 970. This was derived from 2001’s Power 4 architecture used in IBM’s large machines. This processor was the result of a huge effort, with funding from Apple as well as IBM. The PowerPC had never really kept up much after its initial launch with the performance of the best of the 80x86 world, but it had really been falling behind in the 2000s.

Apple dubbed this architecture “G5” and introduced their first G5 desktop as “The World’s First 64-bit Personal Computer” in June of 2003. Ok, so sure, the Nintendo 64 also used a 64-bit processor, but it is true that standard Intel-based PCs were still 32-bit. As were all of Apple’s other offerings.

Actually, between IBM’s PPC 970 announcement and Apple’s G5shipping, AMD introduced the Opteron processor and their 64-bit 80x86 instructions. By the fall of 2003, they had also released the Athlon 64, and the faster processors in these product lines were alreadly beating the IBM PPC970 on popular benchmarks. So Apple was “the world’s fastest desktop PC” for less than a season.

And this kind of illustrates the problem Apple faced. They spent heavily on a competitive desktop processor for a relatively tiny market. Consider too that most of the Mac PCs sold in 2003–2006 were not G5 models, as those were only the high-end units, and never in a MacBook. They were competing with a market that could sell over 300 million units in a year. There was simply no way IBM was going to foot the bill for Apple, and Apple couldn’t pay for it themselves at these volumes.

And in fact, the 80x86 jump was good for Apple. Note that their transisition started in 2006. Look where the graph of sales starts ticking up. It wasn’t simply that Apple knew that PowerPC had no future — everyone knew it.

Apple had their shot at making the PowerPC market work, but they killed it when they killed the open MacOS. Apple was never going to make enough Mac PCs to justify the cost of competitive desktop processor development. That’s not to say that there was any problem with the architecture. Obviously, IBM kept the Power Architecture improving. The problem was simply that, in order to serve a market of 4 or even 8 million PCs per year, you don’t get PC-class prices on competitive processors.

And even if Mac Compatibles had remained in the market, Apple’s deal with IBM kept the PowerPC 970 and subsequent versions exclusive to Apple for some period after release. So other companies wouldn’t have had access to those parts for a year or two. Apple was the only customer, and they had 1/50th the market that Intel and AMD were supporting.

Ironically, it was Apple experience in their unusually smooth transition from 68K to PowerPC that informed their move to Intel. But make no mistake, this move was a hardware move: the only option Apple had at the time to remain both performance and cost comparable with the “commodity” PCs in the market. Sure, Apple sells a premium personal computer, and makes some of the highest profit margins in the business, but this gave them no room to keep PowerPC competitive. And if PPC couldn’t be an advantage, why carry it forward as a disadvantage?

One of the key factors in the success of the PC industry was the fact that all system vendors have had level access to all of the critical chips. Proprietary solutions were often an early advantage to a company, but rarely remained so in the long term. Custom CPUs made sense in very low quantities when they could offer a competitive advantage in workstations and large computers, but even much of that business has been overtaken by more advanced standard parts.

Apple’s made custom processors work for their mobile personal computers because of the volume they do. And they eased into custom. The first iPhone and iPad were using fairly off-the-shelf parts made by Samsung, perhaps with a feature or two left untested to lower cost. Later on, Samsung made them semi-custom parts. Eventually, Apple made their own, built on a budget that, much like the PC industry in the PowerPC days, allowed for full development of a new version every year. Apple can sell over 50 million units in a quarter, all using essentially the same system-on-chip.

And Yeah, Some Application Stuff, Too.

So yeah, Apple did some benefit from x86 in the software world. But I’m not sure it had anything to do with their transition to x86, and it certainly wasn’t inevitable. PowerPC didn’t have to fail — it failed only because Apple had decided to go it alone.

Sure, you can run Windows on a Mac PC these days, and Apple enabled that by offering Boot Camp. Why not? This did allow the want-to-be Mac user to buy a Macintosh even if they needed one or two applications not available on MacOS. And the move installed far more confidence within Apple’s market. Sure, there were a small number of people who regarded Intel and 80x86 as “The Enemy” or “inferior,” but by 2005 those were basically just religious arguments, nothing grounded in fact.

The one place “support for more software” might be arguable is on some of the custom media software being written in those days. A company like Adobe might have done some pretty tight manual coding of x86 special instructions (MMX, SSE, etc) in order to optimizing some very critical sections of code for video, photography, etc. That made support of PPC in the same way difficult: either do another complete set of optimizations for a tiny subset of customers, or ignore it and just let the Mac version run slower. Neither is a great answer.

However, it certainly did become true that over the years, the MacOS version of applications like Photoshop and Premiere were increasingly late-to-market compared to the Windows version, and sometimes left out features. How much of that was really due to PowerPC, and how much was due to Apple moving directly into competing software like Final Cut and Aperture, however, is known only to TPTB at Adobe.

Read More

History of the ISA: Processors, the PowerPC, and the AIM Triple-Threat

Today in Apple history: 'Turnaround artist' Gil Amelio gets the boot

Four years later: Why did Apple drop PowerPC?

10 years on, Apple's risky move to Intel Macs is one of its biggest successes