A week ago I read the rumors that Apple might be considering a move to Intel, and all the naysayers saying, “It’ll never happen”, “Can’t happen” or “No way”. I wrote a small article or two, and responded in a few forums that said, “careful guys, never say never”, then I laid out a few cases for why it could be real; business, technology, emotion, history, and so on. This is not a “Nya, Nya – I told you so” sorta article — because I never said it would happen, I just said it could. Frankly, I didn’t know, and most of those that said they did know are liars or fools. I just speculated on all the ways that Apple could do it and why they might. Apple still surprised me in how, or in actually doing it.
So there I am, at WWDC, waiting for the Keynote, and I go over and chat with one of the smartest most insightful people I know; Bill Chin. He runs a WebObjects / Web Development consulting firm in Virginia, called M Dimension. As usual, I walked up to him and called him “Eric” by mistake. (Sigh). I had a friend in High School named Eric Minami, who was the smartest kid in the school, and he reminds a lot of Bill — and so I often slip, and call Bill by the wrong name, and feel like an idiot. He gracious ignored my “faux pas”, and we had an hour or two to kill before the keynote, and we sat down and chatted. I laid out all the reasons why Apple going to Intel could be a possibility, and he intelligently and reasonably shot them down in diplomatic flames. By the time we walked in for the presentation, I had far more doubts. I kept expecting Steve to say, “Gotcha” in the presentation, but he didn’t. When the slide came up that said “Transitions” you could have heard a flee sneeze; but in that nanosecond, I realized this was really happening, and the computing universe just got whacked hard enough to dent it.
There’s of course a mix of shock, fear, anger, excitement, people pretending to be cool and blasé towards the whole thing, and the zombies walking around thinking, “my reality has just been distorted”. Some very smart people are sitting in corners, rocking back and forth, saying to themselves, “how could Apple do this? Why? Why? Why? Well, here’s my two cents on why I would have done it if I was Apple or my name was Steve Jobs.
Why would Apple do this?
Some are saying it is just ego, that Steve is mad at IBM for some reason or another. Look, we all know that ego plays a part in many executive decisions; then those executive make a business case to back it up. But the other way to look at it is that business plays a part, and sometimes ego backs it up. If Steve is “mad” at IBM, Motorola or the PowerPC, the first question you need to ask is why. And the answer kicks you in the face. Steve made a promise that he’d have 3GHz machines in a year, two years later, he still can’t deliver. Taken by itself, that’s a pretty minor thing; the whole industry had problems with the last die shrinks and power issues and so on. But look at the bigger picture. They failed to deliver.
Apple just spent 8 years fighting Motorola to keep up with Intel and to produce the chips they needed, and they failed. Apple moved to their second source supplier, IBM — and IBM has failed for a couple of years to deliver. IBM produces great high-end server chips, and great low-end embedded chips; Apple is stuck somewhere in between. Jobs doesn’t have G5 laptops, and we don’t know what IBM told him; but probably something that Apple just isn’t that big a priority to them. The PowerPC division has larger contracts with Sony and Microsoft, and used Apple to get them — but now just doesn’t care about Apple as much. So sure that steps on Steve’s ego; but why? Because that gets in the way of his vision for Apple. So it is OK to be mad if your supplier doesn’t support you, if they give your competitors better products than you, or if they can’t deliver what you need. If you are a business, and you can’t trust your suppliers to deliver, then you and your customers get hurt. Period. Excuses are made for failures. Deliver. Motorola had 8 years of incompetence, IBM is showing signs of not being enough better or to give Apple the diversity of products or speed/performance/power that they want, so it is time to move to someone that will.
Intel is an obvious and safe move. Apple can stop fighting the tide. Our processors are the same as everyone else’s. They’ve got two fantastic sources; Intel first, and AMD as a backup. Enterprises or government agencies that have “Intel only” in their requirements will consider x86 based Macs in the future. Apple has access to more suppliers for everything; chips, motherboards, people, and so on. Intel (and competitors) are focused on the exact same markets as Apple, and it is time to stop fighting the impedance mismatch between Apple’s goals and IBM/Motorola’s (now Freescale), and align with someone that is headed in the exact same direction as them.
Isn’t the PowerPC better?
Some say, “Isn’t PowerPC better, then why go to Intel?” This isn’t an either/or or a black/white thing. It can be both — the PowerPC can be better in design, but not better enough in implementation. Or the PowerPC could be better in design, but the suppliers fail Apple enough, that Apple could want to settle with equal to the other PC manufacturers, if it means they can get the parts they need. And so on. Let’s assume the PowerPC is a better instruction set architecture (ISA), um, because it is. First does that matter? Not as much as you might think. It did 10 years ago, when there were 3 million transistors on a chip, now that we are at 150 million and climbing, the few million dedicated to the ISA matters far far less. As Moore’s law has marched on, and doubled capacity every 18-24 months, it meant that inefficiencies in design mattered less and less over time as well. And always what matters more than efficiency of design is efficiency of the implementation.
Technology is never the whole story. And good engineering is balancing tradeoffs. I like cars; so let’s look use some analogies. A Corvette on some levels is lousy technology; they use old 1960’s like push-rod truck motors, and they actually have a leaf spring in their cars suspension. Why can’t they have modern DOHC designs and more advanced suspension like everyone else? Engineering nerds can say, “how out of date”. Now drive the new C6 Corvettes. It beats cars that cost twice as much, with much better technologies. It is a fantastic car for its segment if you look at the system and not just individual parts. The new Mustang’s uses a buggy axle for a rear-end. SRA (Solid Rear Axles) went out in the 70s as the rest of the world went to independent rear suspensions. Now drive it and price it in dollars for horsepower or performance. I like technology; Continuous Variable Valve Timing, Direct Injection, Super/Turbo Charging, active suspensions and so on. Good stuff. But engineering is about tradeoffs, and cars are systems or solutions; look at the whole, not just the parts. Most drivers shouldn’t care how the engineers build the cars, if the entire solution is good and reliable value. The same with computers, only more so as in many ways they are far more complex.
So Apple’s going backwards on ISA, sideways on process; not a huge win. Now look at the processor as a system by itself. It’s been keeping up with PowerPC every step of the way. PowerPC is better at Vector Processing (a little), and historically, fallen short in Scalar performance. Overall, they have stayed fairly close. But Apple got to see IBM and Freescale’s best pitches on what they are doing (their roadmaps), and weigh that with their trust of the companies past promises. And they got to see Intel, and weight that against their deliveries. We know Intel slips, and sells fiction and marketing too, but for 10 years PowerPC was saying how they could deliver twice the performance and would beat CISC designs; yet have they? Users shouldn’t care about what motor is in the car, they should care about what they get in the way of performance and features in the whole car!
Now look at the bigger picture. Apple isn’t choosing one processor, they’re choosing many. Look at the number of processors or I/O chips they get with x86. The infrastructure is bigger, the choices more varied. Apple will get to make the best products in whatever segments they want to go into. They have a much richer family of choices, and a much more reliable supply chain. That’s huge.
Apple isn’t making the move on the processors out today, but the next family of processors, which are 64 bit, multiple core, and maybe even tuned for some of Apple’s needs. Intel has never had a competent customer that could say, “We need X to make the OS and hardware run better”. Dell, et al, are hardware repackagers and don’t know software. Microsoft is a software company, that doesn’t tune hardware. Apple can help Intel, and vise versa. Imagine Apple getting some exclusivity in time for certain processors, or special additions to features that add value. (Not sure I like the implications of that — but we don’t have the full story). Either way, I suspect that Intel can compete, and with the entire industry Intel based, Apple got tired of fighting the tide.
But what about all the Cell Processor, etc.
Now Cell Processor looks interesting. A ton of potential. But as I said in my analysis; can the numbers back it up. I suspect Apple had it in labs, and the secondary units were good, but couldn’t make up for the losses in other areas. Heck, who isn’t to say that Intel didn’t show Apple their next version of the same thing, and it isn’t better? Who is to say that Apple wouldn’t use that as a support processor? And just because the specs look good on paper, doesn’t mean they work when the rubber hits the road. Either you trust Apple to know engineering better than you do, with detailed information or you don’t. Personally, for me it is mixed; but until I definitively know better, I give them the benefit of the doubt. The Cell and others ideas look good on paper; but they also had limitations — I/O or memory addressing, or require major changes to the way we write software or the OS, and so on. Could just be too risky, or too short-term thinking.
But what about all the issues
Look, it isn’t that big a deal. Apple has moved platforms before; this will be the last one. Some pundits are saying, “but Apple lost customers or ISV’s in the PowerPC transition”. The PowerPC transition was huge, and it was at a horrible time for Apple. Apple was down, their OS was outdated, the technical press was on crack, too much was riding on the iron (tied to implementation of hardware), and Microsoft was releasing Windows95 just as people thought it was “as good”, Apple’s fundamentals weren’t that good. Compare that to this transition.
Apple’s fundamentals are great, they have other products helping and money in the bank. They may have deals with Intel to help them through. The press isn’t nearly as anti-Apple or Anti-Steve. Apple lost ISV’s because they were losing marketshare, hype and the transition was painful. This transition is less major to programmers than OS 9 to OS 10 was and they’ve been gaining in software developers. In some ways this one is less painful than going from 10.3 to 10.4 (and taking advantage of new features). Seriously. And in the past, people were thinking, “Microsoft is ‘as good'” — today people realize that is only true if you ignore interface, style, and SECURITY. Before Apple was moving from one proprietary processor to what turned out to basically be another. Now Apple is joining the rest. This is about Apple being able to get into all the contracts that say “x86 required”, it is about using more of the same hardware and being more open. Frankly, this is a major transition, but much more minor than the last few.
Before when Apple moved, you had slow emulators, low compatibility and required major rewrites. Today, you have faster emulators, better compatibility, and it has all been done before. Push-button recompiles is there. Apple didn’t show what they can do in 2 years, they pretty much showed what they could do tomorrow, but they are mostly waiting for their developers and the next generation of processors. Now don’t oversimplify it; sure Java Apps, Scripts, Apple’s Apps, and many Cocoa Applications go over fairly easily. Still the lower level Carbon Apps are going to have some work to do. But for most, the transition is going to be easier than past ones; especially if they had done their jobs of keeping clean/well written/supported code. And by the end of summer, customers will already have far more of the OS and Apps waiting for the hardware and software update of next year than they had when the first PowerPC’s or even OS X were released. So my confidence factor is pretty high.
This is far from flawless, and far from perfect. Apple announcing they were “transitioning” platforms instead of adding one was a shock to me. I figured at most they’d lie, say they were “adding a platform”, and then pull the rug out of PowerPC people later. They’re going to have some sales difficulties for then next 6-18 months. The truth is it shouldn’t matter much; that machine you buy today or tomorrow’s life-span was not hugely altered. But psychologically, it got the scarlet letter added for many people; it is a harder sale. On the other hand, there are lots of leases, and future technology refreshes that could help Apple, as well as new markets.
Different alone wasn’t working, they need to add value. Different has to be better to justify not being the same. Facts are PowerPC just finally wasn’t considered better enough in the future, to justify still being different. x86 will give Apple more opportunities to be different/better in many other ways; and maybe break down further barriers to entry. Having Macs be able to run Windows will make it an easier choice for many. Period. Think about VirtualPC apps being able to run the same speed as real PC apps? Think about if Apple wanted to move into becoming an OS company, and licensing their OS to others; not saying they are going to — in fact, Apple has said they are not at this time, but the option is there for them in the future. This is about choice, options, growth and freedom.
Really smart people had good reasons why Apple shouldn’t do it, like; “Apple doesn’t save money by making the move”, “PC makers have thinner margins that Apple, and this commoditizes them”, “PowerPC’s are better”, “too big a move, too soon” and so on. All valid arguments and good points. But in the end, Apple felt there were more reasons to make the move than not to. Apple will become a PC manufacturer, and I suspect soon after that, try to become the biggest one. Some will say it is about ego; the Mac market wasn’t big enough to contain Steve’s ambitions — but isn’t that what you want in a CEO? Someone that will take the risks for the long term and big picture? Someone that’s trying to dent the Universe, and not just sit there, fat rich and happy in their small corner of it? Like it or not, and there are certainly reasons for both, Steve is making a play for the big leagues, and you can either get on board or watch the train go by.