A look at Apple’s history with CPUs; and how its vertical integration helped beat Intel, Qualcomm, and others

4/ An unlikely partnership was formed between IBM, Apple, Microsoft. This followed the breakup of the alliance known as AIM between Apple, IBM, and Motorola. Recall, Moto was the provider for all Apple’s chips going back to the start.

5/ Apple relied on Motorola including the launch of the PowerBook. But even in the latest 680030 was woefully far behind Intel. Many at Apple, such as @gassee firmly believed Apple needed to gain full control of computing, but becoming a Chip maker cost billions.

6/ This is really key in terms of understanding the present. Apple was essentially left hanging by a partner for chips, when their core deliverable to customers was a computer. That seemed an impossible situation.

7/ Over in x86 land, compute speeds and megahertz kept rising with Moore’s Law. Intel’s investment and excellence in manufacturing made it all but impossible for anyone to catch up, especially making a similar CISC product.

8/ Apple moving to x86 would put them in direct competition with everyone else. That seemed the opposite of the right way to go, much as licensing Macintosh System seems to be wrong.

10/ The Mac System needed to be ported to the PowerPC chipset, including all the hardware and peripherals. It also meant the whole ecosystem needed to move as well, but that ecosystem wasn’t healthy and had started to slow down Mac and favor Windows.

13/ There was little Apple could..for the next few years. Essentially the price/performance gap would continue to become even more bleak. Then Steve Jobs returned.

15/ In the meantime, Apple began to use ARM processors for the iPod. ARM was a RISC design, but had only achieved success in small devices and peripherals. Intel and many other companies were all ARM architecture licencees. This is how Apple got into chips. No one worried.

16/ The end-to-end control of iPod was exactly what was needed to bring together hardware, software, and internet services. First some custom chip features, then more, then more…Of course as we know now, that also marked the resurgence of Apple. F

17/ Rebuilding Macintosh began as well. By this time NeXTStep was already ported to Intel. The work to get the rest of Mac working involved a lot of knowledge about chipset transitions from NeXT and Apple history.

18/ There were no choices for chipsets for “computers” at the time, but of course Apple was clearly going to use ARM for the iPhone (and iPad) project which itself has a long and interesting story (out of context here).

19/ At the 2005 WWDC, jobs announced that Apple would be transitioning to Intel. Many said this is something that should have happened long ago. It really was the only option (IBM was not going to invest “Intel money”)

22/ A huge part of the Intel transition was bringing up hardware, seeding developers, universal binary, and emulating PPC instructions. If this all sounds familiar it is because it is 2.0 of a process that was done in the 68k->x86 transition. It was much smoother.

23/ As Apple progressed with ARM it became clear that the breadth of the ecosystem was a benefit for ARM but was holding Apple back in terms of innovating. iPhone success gave Apple the resources to realize its chipset dream. Thus the System on Chip from Apple began.

24/ The M1 chip was a realization of all the iPad and iPhone work (sensors, OS port, security, power management, graphics, and more). The M1 not only aimed at fixing what ailed Intel, but also PPC. It was learning from the past decade+.

26/ Here’s a thing about “Laptops versus Phones”. The Phone is the computer for everyone around the world now. Laptops (and desktops) are specialized devices for work. About 400M people really use/need laptops for work. That’s what M1 is for and why the prices.

27/ The number of laptops won’t grow, but it is likely Apple will continue to take share from Windows (as will Chromebooks). At 275M units a year, laptops are big but serving this base of 400M. Phones serve everyone including them. That’s where software innovation is for masses.

28/ That’s why this transition to M1 is so fascinating. Back when Apple went on its own with the A-series chips, one could easily be concerned that they would end up in the same place as PPC–not enough volume to win against Qualcomm/Samsung doing their ARM designs.

29/ Apple, by virtue of being vertically integrated, raced ahead. Because of their units and revenue/R&D investment they are in a globally unique perspective. Today, Qualcomm is closer to Intel than it is to Apple Si/Ax/Mx.

30/ Even though Intel serves Windows, ARM/QC/Sam serve Android, they all must serve myriad of OEMs. It appears as though that point of openness (v say at the s/w and service level) is a real constraint on innovation. Partners don’t all want to make the same device, for example.

31/ At the same time, Apple itself has developed a whole family of operating systems. These are not just related, but vastly similar enabling a huge ecosystem opportunity. THIS was Microsoft’s vision going back to early 1990s. Slide from 1992 PDC.

33/ Some might say only 1% of people (of the 400M) even need these. 2 things. First, everything today for super advanced becomes commonplace down the road. Second, never before has one desireable brand made the ubiquitous device AND the super high end one.

34/ Across phone, tablet, tv, watch, speaker, laptop, desktop, there’s a platform of capabilities are unmatched even in one device group, let alone the whole spectrum. It would be like if Ferrari also made mass market passenger cars and electric bikes. Unprecedented situation.

35/ I’d be remiss if I didn’t say these weren’t perfect devices. Why no ethernet in the giant power brick like iMac? Why no cellular option? Yeah FaceID doesn’t fit but soon? Oh and people harping about the notch, STOP!

18x/ Was asked: Apple using Intel for phones, how Intel famously declined. I’ve heard first-hand from both sides and honestly not sure how serious Apple was or how uncommitted Intel was. Also, within Apple it was a debate of sorts, which seems crazy given what I saw on Windows.

32/ The M1x capabilities of shared memory, SoC that isn’t just smaller but has so many aux functions, prores, super fast SSD, even multiple TB ports–all these things require deeply integrated software (from the chipset to the experience). (Out of order, sorry manual process)

Follow @stevesi

Get in