Apple makes the best chips in the industry. It is hard to escape that conclusion. The company that makes iPhones and was resurrected on the back of neon-colored pod computers, makes better chips than most companies who only make semiconductors.
Yesterday, we posted our initial impressions on the design choices Apple made in launching their laptop CPU. Here we want to think through what this means for others in the industry.
In some senses, it is unfair to compare Apple to stand-alone chip makers. Apple has the benefit of entirely controlling the software that runs on top of its silicon. This gives it a big advantage. Companies like Qualcomm and Intel, who make merchant silicon for others, need to constantly adjust their products to make them usable by a large pool of diverse customers. There is a little bit of lowest common denominator strategy tax here. That being said, the performance gap that Apple now boasts should raise some warning flags in San Diego and Santa Clara.
All of this is possible because Apple has sufficient scale to be able to afford a massive chip design team. Hundreds of millions of dollars a year in headcount alone. But as we have noted in the past, designing a chip has in many ways gotten easier over the past decade.
So an obvious question is “Will other PC makers now design their own CPUs?” The answer to this question is almost certainly no, with one big exception.
The world’s leading PC makers – Lenovo, HP, Dell, Asus and Acer – are constrained by a number of factors. First, they do not control their own software which limits the advantage they would get from their own design. Our guess is that Microsoft may open a program to work with the PC makers on some sort of joint product, but given Microsoft’s track record on merely porting their operating system to generic Arm processors, we give any such effort low odds at success. (Note the 2006 date on that linked Microsoft press release.) A second concern is that these companies are very dependent on Intel, and are unlikely willing to risk that relationship over the years it would take for them to build their own CPU. They would risk being at the end of the line on product allocation from Intel and losing out significant marketing subsidies.
Further, the PC makers generally lack internal chip design teams. To be precise, they lack them now. Lenovo used to be part of IBM, but gave up making CPUs, 30 years ago. HP used to have a chip design team, but they spun that off into what is today the colossus that is Broadcom. Dell, which is part of the Dell/EMC/Vmware conglomerate does have internal chip design teams, but spinning up a CPU effort would take them years.
The one exception, as with everything lately, is China’s Huawei. Huawei already has a server CPU. They too lack control of their own operating system for PCs, but are working on that too. Huawei’s OS is still years away, but Apple has likely given them a good case study on what they should build.
There is another glaring and more important problem for the industry with Apple’s chips. Apple not only makes better chips, they seem to be managing chip design better than anyone else. Sifting through our notes from Apple’s presentation, we repeatedly came back to the flexibility of their approach. As we noted in our initial piece, one of the advantages of using Arm processors under the hood of a chip is that it provides a high level of flexibility. Apple, as with all chip designers, mix and match the individual blocks of real estate on the chip to do specific tasks – these blocks for video processing, those blocks for high performance, these for power saving. But Apple appears to have taken this a step further.
Apple now has three families of chips the A for phones, the A-X for iPads and the A-Z for laptops. We strongly suspect that these chips all share a lot more than just a common naming scheme. We know that the A-X for iPads is a derivative of the phone version. And all those video processing cores we highlighted from their presentation are almost certainly common across all three device types. It looks a lot like Apple has created a set of Lego bricks for its chips. This was likely motivated by their goal of providing a common environment for all their developers across both iOS and MacOS, but it is hard to achieve. Most importantly, if true, going forward this means Apple will have a massive productivity advantage to leverage the efficiency of their chip design team. This is a form of compounding advantage from specialization, and it will be very hard for others to replicate.
In all fairness, it can also go horribly wrong. What happens when the audio processing team misses a deadline, risking the ship date for the entire chip? We have seen that too. That being said, Apple is still an incredibly well run company, so this is unlikely a problem any time soon.
For all the industry complains about Apple post-Steve Jobs – bad keyboards, never-updated software, Apple TV, etc. – it should be clear that on this very important front, Apple has built an incredibly powerful technology platform. This will make competing with them in laptops or phones much, much harder. And for companies building chips for those devices, they should take honest stock of how they can improve their own productivity.