Image source: Google.
Alphabet's (GOOGL 0.78%) (GOOG 0.87%) Google recently announced its first in-house smartphones, known as the Pixel and the Pixel XL (a larger variant of the Pixel). These are premium devices, starting at $649 for the base Pixel model (5 inches) and $769 for the base Pixel XL model (5.5 inches).
These phones currently use a Snapdragon 821 applications processor produced by wireless chip giant Qualcomm (QCOM 0.91%). However, according to Bloomberg, Google has expressed interest in building its own custom silicon out in time.
This leads to a very interesting question: Will Google develop its own applications processors for future Pixel smartphones?
Google could probably do it
Putting together a reasonably competent smartphone-oriented applications processor these days isn't too difficult, especially if one has the money to spend (and Google can clearly afford it). Many of the key intellectual properties that go into a mobile processor -- CPU cores, graphics, memory controllers, eMMC, and so on -- can be licensed from third parties such as ARM Holdings, Imagination Technologies (NASDAQOTH: IGNMF), and Synopsys (SNPS 1.04%).
One major hurdle that Google would undoubtedly face in trying to build its own custom chip is the fact that it could not easily license a cellular modem to integrate into such a chip. Google would either have to spend a lot of time, money, and effort to even hope to build a reasonably competitive cellular modem or it would need to use stand-alone modems from the likes of Qualcomm or Intel (INTC 5.41%), the two suppliers of modems into the latest iPhone.
The stand-alone solution is less than ideal because the current Pixel phone uses a chip that integrates the modem onto the same piece of silicon as the applications processor, which helps to reduce board size as well as power consumption. Apple (AAPL 0.96%) has been dealing with stand-alone modems for as long as the iPhone has been around, but for Google it would represent a step back from the solution that it's already using.
In order to build a differentiated solution, Google could choose to build unique intellectual properties to integrate alongside the core technologies licensed from the above vendors. Custom hardware to accelerate certain common smartphone functions (thereby improving performance and/or lowering power consumption), for example, could be worthwhile projects for a potential Google custom silicon team to pursue if it really wants to build its own chips.
Would it be worthwhile?
The impression that I get is that most of the companies that build their own applications processors don't actually produce differentiated (or, frankly, even worthwhile) solutions relative to what can be purchased from the likes of Qualcomm and MediaTek.
Apple's in-house chip efforts have been quite successful, and Samsung's (NASDAQOTH: SSNLF) have been reasonable, but efforts from the likes of Huawei's HiSilicon and LG have been mostly underwhelming.
If Google were a major smartphone vendor and its core revenue/profitability depended on selling large volumes of ultra-premium smartphones, then there might be a case to spend all of the time and effort to build in-house applications processors, though there's no guarantee that it'd be as good as, let alone better than, what Google could buy from Qualcomm.
However, given that the Google Pixel is unlikely to be a particularly high volume seller, especially when it's charging premium prices, and given the fierce competition in the Android landscape from companies that have been at this longer than Google has been, I doubt that there's much of a (rational) business justification for such an effort.
I'm not betting on it
My guess is that Google will build a couple more generations of these Pixel devices, but they ultimately won't sell all that well. At that point, I expect the effort around the Pixel to be scaled back, if not outright shuttered.
If the case for an in-house applications processor is tenuous at this point, I suspect that it'll seem downright silly if and when it becomes clear that the Pixel is a nice hobby, but not really a worthwhile product category for Google to pursue over the long haul.
