Apple: Becoming A Semiconductor Powerhouse

Summary

• Apple's iOS devices all use Apple's custom processors.
• Because of the explosive growth of its mobile devices, Apple has become one of the world's largest semiconductor companies.
• Apple's semiconductor prowess will make it possible to build ARM-based Macs at lower cost and higher margin than with Intel Inside.
• This economic imperative will mandate a Mac transition to ARM beginning in 2016.

Often overlooked by Apple (NASDAQ:AAPL) bears is the fact that Apple has become one of the largest semiconductor makers in the world. Apple's custom systems on chip (SOCs) power all its iOS devices and represent a $6 billion business on its own. More importantly, Apple has created a technical discriminator that devices with commodity ARM (ARMH) processors have had difficulty matching. The future of Apple's semiconductor business is bright, as it forges SOCs to rival Intel's best.

Commodity Volume

The volume of Apple's chip business is one that all but a few chipmakers would envy. Based on only iOS devices (iPhone, iPad, iPod Touch), Apple is selling 100s of millions of SOCs a year. This doesn't include other Apple products such as iPod, Apple TV or the many smaller supplemental chips that carry the Apple brand that can be found in its mobile devices. The chart below shows the unit volume growth over the past five fiscal years.

Note that Apple's fiscal year starts in October, so 2014 doesn't include the explosive growth in iPhones of the fiscal 2015 Q1 December quarter.

It's not easy assigning a dollar value to Apple's chip business, because Apple doesn't provide a cost breakdown for its devices, of course. There's wide divergence in the cost estimate for Apple's latest SOCs. IHS provides an estimate of $22 for the A8X and M8 motion coprocessor in the iPad Air 2. While Techinsights' estimate of the cost for the smaller and less capable A8 SOC in the iPhone 6 is $37.

For purposes of getting a ballpark value of equivalent revenue from Apple's SOC business, I settled on an ASP of $25. This is consistent with the estimated cost for typical ARM SOCs. In the chart below, I rank semiconductor companies by revenue in fiscal 2014, including Apple. (Here, I used MarketWatch as the source of revenue data for the companies listed.)

The list of companies in the chart is by no means an exhaustive list of chipmakers, but includes the major companies involved with making SOCs or embedded processors. Apple's $6.1 billion equivalent revenue is just the $25 ASP multiplied by the 244 million unit shipments of fiscal 2014.

More important than the size of Apple's chip business is its rate of growth, which far exceeds the industry average. In the chart below, I plot the percent change of Apple's equivalent SOC revenue compared to an "Industry Average" of the companies listed in the previous chart excluding Apple. In addition, I include the annual revenue of Qualcomm (QCOM) and Intel (INTC). Qualcomm had the fastest growing revenue of the chipmakers I listed, while Intel paced the "Industry Average." (Once again, data is from MarketWatch.)

Challenging Intel

Apple's growth in SOCs for its iOS devices has allowed it to invest in design innovations at the SOC level. Apple introduced the first 64-bit processor for smartphones and tablets (A7) in 2013. These are only now being matched by Qualcomm's 810 series and the Samsung (OTCPK:SSNLF) Exynos 7xxx series being introduced in the latest flagship Android smartphones.

Apple's SOCs (and other 64-bit ARM SOCs) have not merely become the best performing ARM SOCs, but can challenge Intel's best mobile SOCs as well. This was brought home to me by the recent release of the MacBook with Retina Display, which uses an Intel Core-M-5Y70 processor. The Core M is fabricated on Intel's most advanced 14 nm node and runs without a cooling fan in the new MacBook.

Although the MacBook hasn't been tested yet, its Core-M processor has been benchmarked in a Windows 64 device, the Lenovo Yoga 3 Pro. What's interesting is that it has about the same performance on Geekbench as the A8X in the iPad Air 2.

Think about what this means. The A8X is fabricated on a less advanced 22 nm process (non-FinFET). Despite the process difference, the area of the A8X of 128 mm^2 is smaller than the area of the silicon (2 separate chips on a circuit card package) of the Core M at 139 mm^2.

Although Intel doesn't reveal pricing information on the Core M, Apple's SOC is almost certainly less expensive, by virtue of the smaller chip area, the simpler packaging and the less advanced process. So Apple is now able to equal the performance of an Intel Core architecture chip at lower cost. It's a testimony to the power of the ARM 64-bit architecture, as well as Apple's custom processor design.

The performance of the A8X also suggests that the next series of Apple SOCs, which are expected to be fabricated on Samsung's 14 nm process will exceed the performance of Intel. These next generation SOCs will be significantly smaller and lower power, but more capable than Intel Core M. They will probably be less expensive as well.

The ARM Future

Apple could have easily based the new MacBook on its own SOCs rather than the Core M, if not for other considerations such as app compatibility. If it had, the MacBook would have been even lighter and thinner, since the logic board would have been smaller (about the size of what's in the iPad Air 2). The ARM-based MacBook would probably have been less expensive as well.

Apple's control of the Mac OS and iOS ecosystems, its growing prowess as a semiconductor maker and its Xcode developer tools give it the ability to port Mac OS to ARM with minimal disruption. Apple's ability to build lower-cost ARM SOCs with equal or better performance than Intel makes the economic imperative clear: Apple will be able to build Mac OS devices at lower cost and better performance with its own SOCs inside. These ARM-based Macs will also be lower retail cost higher margins: a win-win for Apple and consumers.

Up until now, I've been cautious about this issue, saying that I don't think Apple has made a decision yet. I believed that the key milestone would be successful production of the next Apple SOC series at 14 nm. That production has almost certainly begun, based on Samsung's impending release of the Galaxy S6, which uses its 14 nm Exynos SOC. With the passing of this critical gate, Apple's decision to move on from Intel is straightforward, even obvious. I expect the first ARM-based Macs to appear in 2016.

Analyst’s Disclosure: The author is long AAPL. The author wrote this article themselves, and it expresses their own opinions. The author is not receiving compensation for it (other than from Seeking Alpha). The author has no business relationship with any company whose stock is mentioned in this article.