Seeking Alpha
What is your profession? ×
Long/short equity, contrarian, deep value, event-driven
Profile| Send Message|
( followers)


My suggestion that Qualcomm's income would be threatened if Apple took baseband in house was hotly debated.

The lack of detailed analysis in my article may have hampered the discussion.

This article provides the detail.

Analysis of the degree of Qualcomm's (NASDAQ:QCOM) dependence on Apple (NASDAQ:AAPL) is complex. My last article engendered a terrific discussion among SA readers with expertise in the area of Qualcomm's intellectual property licensing and component supply. That discussion made an enormous contribution to my understanding of the linkage between these two key players in mobile.

I thought it might be useful to provide readers with the details underlying the analysis I did to support the theme of the article - that loss of Apple as a customer would cause considerable harm to Qualcomm.

First, I prepared a model of Apple's fiscal 2013 income statement built up from each model of its iPhone and iPad using IHS estimates for the bill of materials for each device. A bit tedious but worthwhile.


Apple shipped 151 million iPhones in its fiscal 2013. Making some assumptions about the mix of these devices I pulled together the following model of iPhone revenues and costs.

(click to enlarge)

Sources: IHS bill of materials estimates; Apple 10-K; Blair analysis

As I understand the IHS bill of materials data, Qualcomm supplies the Wireless components for prices between $23.50 and $34.00 depending on which device, and the Power Management components for between $6.90 and $8.50 per unit again depending on which device.

While I recognize that Qualcomm in all likelihood sources some of the sub-components of the parts it supplies to Apple, I have assumed Qualcomm books all the revenues from these components and the sub-component purchases are part of Qualcomm's cost of goods sold.

I have also assumed Qualcomm receives a royalty of 5% of the total cost of each device; that being 5% of the amount Apple pays to Foxconn or Pegatron for the completed units they assemble for Apple.

If that is so, the Qualcomm content in iPhones in 2013 amounted to $5.9 billion for components and $1.6 billion for royalties.

(click to enlarge)

Source: Blair arithmetic


I did a similar analysis for iPads.

(click to enlarge)

Sources: IHS, Apple 10-K, Blair analysis

In this case, I understood Qualcomm to be the supplier of the Wireless components for the 4G models at prices from $32 to $41.50 per unit depending on the model, and the Power Management components at prices from $7.25 to $10.00 per unit again differing by model.

Making the same assumption about the treatment of sub-components I calculated Qualcomm content in iPads during Apple's fiscal 2013 to amount to $1.7 billion for components and $470 million for royalties.

I understand from many of the comments on my last article there is a general belief that even if Apple develops its own baseband or sources from another supplier, Qualcomm will retain substantially all of its royalty income if not from Apple then from Apple's supplier as the case may be.

Retaining the royalty stream more or less intact does not speak to the damage from the loss of what seems to be $7.6 billion of component sales. Qualcomm reports that its gross margins from its QCT segment was 42% in the fourth quarter of 2013. Qualcomm says its technology licensing revenues are reported in its QTL segment but it is not clear to me that QCT does not include some royalties from CDMA. Perhaps one of the better informed SA readers could advise.

In any event, I believe it likely that Qualcomm earns 35% to 40% margin on its component supply, which would be consistent with the high margins typically earned by semiconductor suppliers somewhat reduced by lower margins, if any, on the pass through of sub-components supplied by others.

If I am wrong that Qualcomm buys the sub-components and sells them together with its own parts, the revenues Qualcomm earns would be less but the margins would be quite a bit higher since the Qualcomm content is largely semiconductors. I think there would be very little difference in the outcome in either event.

I have taken some liberty in estimating the mix of Apple's iPhone and iPad units across its various models but I believe the mix is at least representative and after a bit of sensitivity testing found that broad changes in that mix had little effect on the Qualcomm calculation.


I reiterate the view I expressed in my earlier article, that is, that should Apple take its baseband components in house, perhaps in an integrated SoC together with its forthcoming A8 or A9 processor, the impact on Qualcomm would be severe. I could see a scenario where Intel (NASDAQ:INTC) builds an integrated SoC for Apple as a remote possibility that could benefit both companies longer term.

The conclusion that Apple might or might not move to an integrated SoC is for each investor to draw, but my view is that it is more likely than not over time. By putting more capabilities on individual chips SoC's tend to yield lower risk of failure, lower costs and less power consumption while taking up less space in the device. Discrete baseband does not make sense to me in a smartphone where baseband is necessary in every device (unlike tablets where some include 4G and others only WiFi) and the knock on Intel's efforts in modems has long been its inability until recently to offer an integrated SoC. Sauce for the goose is sauce for the gander.

I remain short QCOM, not particularly because of this issue but because there are a lot of headwinds affecting Qualcomm from fast growing competitors like Mediatek and Allwinner and a determined effort by Intel and in my experience increased competitive pressure is ultimately felt in margins by even the strongest suppliers.

Disclosure: I am short QCOM, AAPL. I wrote this article myself, and it expresses my own opinions. I am not receiving compensation for it (other than from Seeking Alpha). I have no business relationship with any company whose stock is mentioned in this article.