With Apple now somewhere in the process of building its own modem, we have had a few people recently ask us if Apple is next going to build its own radio frequency (RF) chips? Or put another way, how vertically integrated will Apple go?
Smpartphones have eight main systems – the applications processor (AP), the modem, the screen, the battery, the RF chain, cameras, audio, and the package they all go in.
For most of these there is little point Apple doing any more than they are already doing – they design the package, the audio and camera systems, and battery – they likely already get all that those vendors have to give and it is hard to see how much advantage Apple could gain from getting in any deeper. Maybe they add a chip for audio (if they don’t already) – but this is already a highly abstracted supply chain. The situation with screens is similar – Apple provides a lot of upfront funding to the companies that manufacture the screens, and as a result get a close-to-custom product. The story is similar with memory – manufacturing memory is capital intensive with massive scale economies. Apple has multiple vendors to play off against each other (the addition of YMTC will just reinforce that point), and so again there is little left to gain from them getting into this field. Similarly, we see no reason why Apple would ever build its own semiconductor fab, not when they can pay TSMC to do that for them. File all of these under “Why buy the cow when you already get the milk for $22 billion in R&D and $11 billion capex?”
This leaves three systems – AP, modem and RF. Here there is a strong argument that Apple can generate real strategic advantage from designing their own chips. This is clearly true with the A-Series AP, a chip that Apple has been designing for over a decade and leveraging to differentiate the iPhone and iPad from competing devices to great effect. The same could be true about the modem as well. There are all kinds of interesting things that Apple could do with its own modem, not least tie it closely to the operating system and drive significant battery life improvements and likely faster data rates. We also suspect they would implement clever private communications protocols – think Blue Bubbles for iMessages.
There is a similar argument to be made that RF could offer similar tangible, strategic improvements – the kind that make the iPhone stand out even more from the competition in consumer comparisons. But that is likely as far as it goes – an interesting an idea on paper, but much harder to put in place.
RF is not like the other chips in the phone. For starters, it is largely handling analog signals (i.e. radio waves). Designing these is unlike anything else Apple has designed so far, and would require a new dedicated team. Given Apple’s resources this is certainly achievable, but there is a bigger problem – RF is not a chip. RF systems are really a chain of modules or packages which contain dozens, even hundreds of components. Some of these can be categorized as semi-conductors, but many are not. Notably the filters, on which Broadcom has built an immense business, are ceramic. Simpler to design but incredibly hard to manufacture. For Apple to have any real impact on these systems they would likely have to go much deeper into manufacturing of those RF modules.
In theory, Apple could design its own modules and let others manufacture them. But this is not too far from where we are today. Apple could theoretically benefit from taking a stronger hand in design, but they are also really good at getting their vendors to do that kind of work for them. Regardless of who really authors the design Apple is almost certainly going to have to continue to rely on vendors to supply some of the components that go into those modules. This is good news for the likes of Broadcom, Qorvo, Skyworks and Murata who supply most of the content and manufacturing of RF modules for Apple today.
There are a couple of situations in which this might change. First, Apple could easily design some form of ancillary chip that acts as support or ‘controller’ for the RF systems. For instance, when Qualcomm was first dipping its toe in the RF world made a big deal about its “envelope tracking” chip. The advantage of this approach is that this chip looks other chips with which Apple is familiar, largely digital, made of standard materials. The disadvantage of this approach is that such a chip can have only minimal impact on the RF system. Nice to have, not need to have.
The other wrinkle is whatever Qualcomm is working on. We have written repeatedly about their ambitions to build an RF system that is integrates more tightly with the other digital systems on the phone. Such a system could deliver a meaningful improvement in RF performance and thus overall phone battery life and call quality. The key to such a design is that it requires very low level access to the modem. This is a big problem for the incumbents mentioned above as Qualcomm is not likely to give them that access. And the performance gains here are meaningful enough that we think the iPhone engineers at Apple would have to revaluate their plan to give up on using their own modem, not when the Qualcomm integrated modem + RF combo is delivering tangibly better performance. If Qualcomm could deliver such an RF system, Apple would likely be faced with a choice between sticking with the Qualcomm modem, or going down the path of designing their own integrated RF front-end. If true, the timing is highly uncertain. Qualcomm has been working on this Holy Grail for ten years, and it is unlikely Apple could turn out their own version any time soon.
Admittedly, there are an awful lot of ‘Ifs’ in that speculation above. Nonetheless, we think it is unlikely that Apple is going to start designing its own RF chips, at least any more than they already do. That being said, notice that nowhere in this piece do we say it is impossible. We are just scratching the surface on this topic here, there are many more layers we could drill down, but we think Apple is already getting much of it wants from its current vendors, and any change to that would likely only come after a major shake-up of the RF industry.