Stardate 20020722.2109

(Captain's log): David writes:

It's been 5 years now since Steve Jobs took over Apple again and dropped all Mac-clone licensing and the CHRP platform program. Was the PPC doomed anyway or did Apple's own decision to drop CHRP cooperation do it in?

In my opinion, that particular decision was lose-lose. Either way Jobs took it, Apple was fucked. Basically, though, he could choose short-term fucked or long-term fucked, and he chose the latter.

The main reason that Motorola was willing to make the investment in trying to keep the PPC competitive (and that investment was massive) was because it was expected that the PPC would become the basis for a commodity platform (the Common Hardware Reference Platform) to compete with the x86 PC commodity platform. It would run not only MacOS but also WinNT and also, I believe, Solaris. (NT 4 shipped with a version for PPC to run on the CHRP.) As such, the expectation was that volume would rise to the point where Motorola could actually make a profit on that investment.

But in the short term, the main market for PPC was Mac users, and the cloners were stealing customers from Apple instead of finding new customers. The cloners could create the same hardware as Apple and sell it for less than Apple, because Apple needed to divert money from hardware sales to fund development of MacOS, which they were selling at a loss to the users of the clone platforms.

Killing off the cloners made it so that everyone who bought a platform which could run MacOS paid their share of the bill for MacOS development via higher hardware prices. But it also kicked Motorola in the teeth, and led directly to the legendary 18-month clock-rate stall. Instead of continuing to work on high performance PPCs for what had become a single customer which had demonstrated itself to be untrustworthy, Motorola instead concentrated on making PPCs which ran the same speed as before but which used much less power, which is what Motorola's embedded customers wanted. That's always been Motorola's volume market for the PPC, and unlike Apple it wasn't ruled by a single customer. It just made more sense to invest there.

The result at the end of the 18 months was the 7450, which was only marginally faster than the processor which had come before but which used about half as much power.

That was great for the embedded customers; it was exactly what they wanted. But it was nearly useless for Apple. And in fact, Motorola has never been seriously dedicated to making a high performance processor since then. They've been working on it half-heartedly, but there has never been a big enough market for high performance PPCs to make it economically viable for Motorola to make the necessary investment. Motorola has been losing money on every processor it sells to Apple for several years now, mainly because the volume isn't there for economy of scale to kick in. (It's been a major contributor to Motorola's dismal financial performance recently.)

Processors costs the same huge amount of money to design no matter how many get sold, but with lower volume you have to amortize more engineering cost per unit. At the expected volume, a fair price (i.e. break-even for Motorola) would have been too expensive to be competitive with AMD and Intel, who had vastly higher volume to pay for their engineering.

So far as I know, there's nothing inherent in the PPC architecture which has kept its speed down. If anything, there's every reason to believe that it could have kept up with sufficient R&D investment. What made it slow was market conditions, not technical limitations.

If Apple hadn't booted the cloners, I suspect that there would be a healthy market for PPC-based commodity computers using the CHRP. And I suspect nearly all of them would now be running NT or Solaris or BeOS (!), but not MacOS because I think Apple would have gone out of business by now.

There was a third choice, but it would have required Jobs to "think different". He could have divided Apple into a hardware company and a software company. The software company would sell equally to every maker of hardware, and the hardware company would have become one of the sources of commodity PPC systems.

But it was never really possible, because the "different" it would have required Jobs to think would have been to admit that he'd actually made a mistake, and that all-under-one-roof had become commercially non-viable.

I think it would have worked, but we'll never know now.