The latest whispers from the far east suggest that Apple's (AAPL) oft-talked about "cheaper" iPhone will be powered by Qualcomm's (QCOM) Snapdragon line of processors. At first, the rumors seemed to make very little sense, especially as Apple has its own in-house processor design team. However, upon deeper reflection, I believe that there may be a very legitimate business case for Apple to use Qualcomm's chip - rather than older editions of its own A-series chips - for such a device.
Apple Designs Its Own ... Right?
The most obvious reason to be skeptical is that Apple spends a great deal of money designing its own system-on-chip products. Each generation, Apple has bought and built more chip design expertise to the point where the most recently announced A6 processor actually comes packed with a custom-designed processor core. Given the volume of product that Apple ships, the R&D costs of its own system-on-chip and processor IP are likely to be easily made up, and paying another company's design margin doesn't make sense. Unfortunately, Apple still needs to pay Samsung's (SSNLF.PK) foundry margin, but unless Apple wants to gamble on building and maintaining its own factory network, this is unavoidable.
So, given that Apple designs its own chips, why would it use a solution from Qualcomm and pay the latter firm's design margin, as well as TSMC's (TSM) foundry margin? It is purely a question of the cost of connectivity. While Apple designs its own apps processor, it sources anything and everything that has to do with connectivity from other vendors. More expensive/premium phones can afford to use separate connectivity chips, but when a company is trying to aim for the very low end while keeping margins acceptable, an integrated part makes a lot more sense.
Let's first take a look at the BOM costs for the iPhone 5 to get a sense of the economic factors in play here.
The iPhone 5 BOM -- Chip Content Dominated By Qualcomm
Thanks to the good folks at IHS iSuppli, investors have a fairly good itemized picture of the iPhone 5's BOM costs:
The total BOM for the iPhone 5 is $199, which makes sense if Apple is selling a $650+ device, but is absolutely unacceptable for the $100 - $300 device space that seems to be growing at an outsized rate. This means that Apple needs to start cutting costs wherever it can.
The largest part of the BOM outside of the screen is the "wireless section," which is a Qualcomm modem and RF front end at $34/chip. This, coupled with a $17.50 A6 apps processor, eats up a full $51.50 of the BOM for this potentially cheaper phone -- yikes! Considering that Intel (INTC) is shipping chips (modem + apps processor) into phones that retail for $125, the maximum BOM for such a cheaper iPhone can't really be more than $100.
An obvious place, then, to cut costs would be to buy a single, cheaper integrated apps processor + modem. Further, a lower end apps processor can also be had for much cheaper, since the die sizes are smaller. Further R&D is already paid for since the processor/graphics IP can be very cheaply licensed from the likes of ARM Holdings (ARMH) or Imagination Technologies. I suspect that a low end, highly integrated apps processor from Qualcomm would cost perhaps $15 - $20 in sufficient quantities, which would give Apple the ability to more effectively compete in the low end segments.
This all ties back into my previous article in which I note that, going forward, Apple may need to acquire a baseband vendor if it continues on the path of wanting to have its own in-house IP. My bet is on the ST-Ericsson JV, since both ST Micro (STM) and Ericsson (ERIC) can't seem to get it to be profitable in the merchant vendor model. It could also try to staff up an internal team, but that would make it exceptionally difficult to ever make it to the leading edge.
However, it may be the case that Apple eventually puts a stop to its internal chip design business and shifts chip supply to a vendor such as Qualcomm (the market leader), especially as mix shifts more and more towards lower end, lower cost phones. Besides, as Apple loses market share, and as chip R&D gets costlier each day, it seems unlikely that the vertically integrated model for most phone vendors is really going to work out. The merchant vendor model for processors won out in the PCs, and it is my view that it will win out in smartphones and tablets unless your name is Samsung (and even they use Qualcomm's integrated apps + modem parts for their most successful designs).
Additional disclosure: I am short ARMH.