Last week, Bloomberg published an article speculating about the big footprint Apple is building in Southern California. The article is light on detail, but seems to state that Apple is building Radio Frequency (RF) products that compete with Broadcom and Skyworks. The report knocked 10% off Skyworks stock price.
This type of article pops up regularly over the years, and they usually overstate the situation, making some case that 1 + 1 = 1,000. We viewed this on as an early Christmas (or belated Hanukah) gift, an easy topic for us to post on. And certainly, there are a lot of problems with this piece. So we went to the source, and looked up job listings on Apple’s website, and it turns out there is something going on here, albeit not exactly what Bloomberg implies. Apple is clearly building something there, but what? It is all so mysterious, it immediately reminded us this Tom Waits dirge.
Some context first. Apple has a huge office in San Diego. They
poached hired a 1,000 people from Qualcomm (and others) to help build their modem. But now, Apple is expanding 70 miles north in Irvine, home to both Skyworks and Broadcom, both big Apple suppliers. The problem with the article is that it is not entirely clear on what Apple is building. We are really talking about three different products – RF transceivers, RF front-end modules (i.e. filters) and connectivity (Wi-Fi, Bluetooth & GPS). The process behind building these are all very different and have very different implications for each one.
The most obvious next step for Apple’s path towards vertical integration is to build their own connectivity module. They tried this ten years ago, buying a team from Texas Instruments. This effort failed due to a combination of the incumbent supplier, Broadcom, cutting them a sweet deal and the Apple team’s inability to build a viable alternative. We often point to this failure when we caution that it may take longer than expected for Apple to complete its modem, networking is hard, and if they could not build Wi-Fi, cellular is much harder. But Apple Silicon has made immense strides since then, and so we tend to default to when not if. Apple will eventually build its own connectivity chip. So maybe the article is right, if a bit overstated.
Then there is that whole “RF” chip. What is that about? RF products are complicated – the RF chain that handles radio waves is comprised of dozens of chips and passive components. First, there is the transceiver. This is an obscure corner of a phone that once had 30 companies competing in it, but is now largely integrated into other chips in the phone, notably the modem or baseband. If Apple is building a modem, building a transceiver is a natural extension of that effort. This is essentially just another chip, closely tied to the modem, built alongside the modem. They get these today from Qualcomm, so when the modem goes internal it makes sense Apple would take this as well, to reduce their Qualcomm exposure to zero.
But then there is the other part of “RF” – the front-end which is comprised of power amplifiers, switches, envelope tracking, filters and a bunch of other things. Today, Apple gets most of these parts from Skyworks, Broadcom, Qorvo and Murata, in a dizzying array of combinations (e.g. Qorvo modules with components from Murata, or Broadcom modules with parts from Skyworks). Crucially, the big question we have been asking is does Apple plan to build its own filters? These are not semiconductors, and their design and manufacture is highly specialized, unlike anything Apple is building today. If we had to guess, we would say Apple is building some of its active semis components itself, but will still use its vendors to do the packaging and bundling into modules with passive components (i.e. filters).
However, there is another tantalizing possibility. We have written a lot about the “Holy Grail” of RF – the Integrated RF Front-End. This is an RF chain that is tightly coupled to a phone’s modem. This would entail a complete re-alignment of RF architectures, and offers significant performance and manufacturing improvements. It is always very hard to achieve. Crucially, the industry has realized that the only company that can accomplish such a module has to be the company building the modem. As such, today only Qualcomm can really deliver it, and we have written that if Qualcomm can deliver on this it would be an important reason for Apple to stick with Qualcomm. If Apple is building its own modem, they too could build their own Integrated RF Front-End, with a bit of help of someone like Broadcom or Skyworks. (So maybe not such bad news for Skyworks after all.)
So you can see why the Bloomberg article left us a bit cold. It entails a lot complexity built on top of the simple fact that Apple is hiring a lot of people in Irvine.
Which brings us to Apple’s job boards. A search on jobs.Apple.com for “RF” yields 694 results, which is a big number, even for Apple. There are two ways to read the results. The most likely assumption, and our default for Apple, is that they are hiring people to design specifications for systems that will be built by their suppliers. Recall our thesis that one of Apple’s core competencies is getting their vendors to do the labor-intensive parts of product development, part of the way they accomplish this is by hiring teams to dig into every detail of every component. Our first review of these job description aligns somewhat with that simple assumption. Apple products have RF systems in them, so they want to have some engineers to design those systems, that does not mean Apple is reducing their reliance on key suppliers.
That being said, there is an alternative explanation – that Apple would not need 600 people for this simple purpose outlined above. And as we read some of these jobs, there are more clues. For one thing, many of these jobs seem to be part of the Apple Silicon team, not the iPhone team. Some of the jobs explicitly mention SoC development (system on a chip), which is not something that they would necessarily need if they were just managing vendors (maybe, maybe not). So it looks like Apple is building some parts of the RF chain, maybe envelope tracking or even power amplifiers.
And then we came across this listing – RF System Architect. This is a pretty strongly worded job description
Join us to help deliver RF system design for devices operating in next generation groundbreaking radio access networks.
Our team has the mission to research emerging spectrum trends and RF architecture evolution leading to next generation radio access networks and use cases.
This job in itself does not answer the question as to whether Apple is going to drop any of its vendors. (In fact, this appears to be in the hardware group and not in Apple Silicon). But the scope of the role is immense and gives a lot of hints as to what Apple is planning.
First, just to get this out of the way, Apple is doing something with satellite communications, the listing explicitly mentions “non-terrestrial” networks. We wrote about this a while back, and we are now fairly convinced that Apple is working on some form of satellite based messaging system (not a replacement for cellular).
But the posting gets wilder. A big part of the job requirement is intimate understanding of wireless standards. When Apple launches its modem it is going to come with special communications features that tie Apple devices together in ways unavailable on other devices (think Blue Bubbles for iMessages, but on a far wider scale), the listing includes tantalizing hints of “sidelinks”.
Another of the requirements:
Drive practical RF innovation into 3GPP RAN4 specifications
Decoder ring: Apple is looking to build greater influence in the standards bodies that set the 5G and 6G cellular standards. Apple is not just building its own modem, they are looking to change the whole system (to their advantage).
But this job is about more than just cellular – there is Wi-Fi in there, some very high frequency work (THz), beamforming, and “wireless sensing”. The job has a huge range and if we were a senior RF engineer we would have already applied for this position.
To sum it all up. Apple is probably building its own transceiver (bad for Qualcomm, but not a surprise), it may be building its own Wi-Fi/Bluetooth/GPS chip (bad for Broadcom, but not certain), and is going to take a more active role in designing its RF systems (possibly bad for Skyworks, but maybe not), and almost certainly not building its own filters (Broadcom is unstoppable).
Beyond that, Apple has some very ambitious plans for communications and they are looking to drive the industry in their direction. They have set themselves some very interesting objectives, or at least left us with some big mysteries. What are they building in there?
Photo by Craig Whitehead on Unsplash
Pingback: What lies ahead – the loose wire blog·