I am actually somewhat puzzled at the relative dearth of analyst coverage pertaining to Intel's (INTC) mobile strategy, and in particular the roadmap revealed at the 2012 Investor Meeting. A lot of folks are still unclear over both the significance of "Medfield" and the mobile products on the roadmap. The purpose of this article will be to give investors that clarity.
Medfield: Competitive x86 In A Phone
Okay, so first off, with "Medfield," Intel was finally able to get an x86-based processor to comfortably fit inside a system-on-chip that is suitable for phone usage. The commonly held belief was that the x86 instruction set prevented Intel from being able to provide a competitive solution in phones due to concerns over performance per watt as well as compatibility with the pre-existing software written for ARM (ARMH) processors. The existence of "Medfield" based phones in the commercial market today debunked both of those myths.
Busted Myth #1: "Atom Is Too Slow And Power Hungry"
Well, that turned out to be completely wrong. In fact, when Anandtech tested the Lava Xolo 900, the single core "Medfield" running at 1.6GHz was able to, with competitive power consumption, significantly outperform dual core ARM Cortex A9 implementations in tasks that only utilize a single core, and was even able to edge out the dual core ARM-based chips in situations that utilize more than one core.
The only competitor on a core-for-core basis was Qualcomm's (QCOM) Snapdragon S4, which proved to be roughly 13% on a core-for-core basis. But note that the S4 is a brand new micro-architecture that significantly resembles ARM's next generation "Cortex A15."
In addition, while the current Atom Z2460 "Medfield" is running a single core, Intel has a higher end dual core Z2580 (with LTE) that will be available in the first half of 2013, which should close the performance gap with quad core Cortex A9 and dual core Cortex A15 class chips while maintaining competitive power consumption characteristics.
Busted Myth #2: "Software Compatibility Will Be A Problem"
While it seemed unreasonable to doubt Intel's ability to deliver proper hardware, what I was most surprised by was how well Intel handled the software side of things. Intel's (considerable) software team partnered up with Google (GOOG) to port the latter's popular "Android" operating system to the x86 instruction set while maintaining compatibility with existing applications. While a detailed explanation of the technical details can be found here, the upshot is that Intel chips should be just as compatible with Android and its applications as any ARM chip is thanks to work that both Google's and Intel's software teams have put in.
Tick-Tock: A One-Two Punch
Interestingly, one of ARM Holdings' greatest weaknesses is that the firm's business model introduces substantial latency in getting designs out on the market. The Cortex A9 design was released in 2007, but we didn't see phones with chips built on that core design until 2010. Similarly, the design for the Cortex A15 was released in 2010, but system-on-chip designs from the major vendors did not show up until 2012.
Intel's advantage here is actually staggering. First, the company develops its instruction set and core architecture internally, so it is not dependent on a third party before it can design its own system-on-chip solution. Once that system-on-chip is designed, Intel is also fully capable of building it without worrying about trying to secure manufacturing capacity at a foundry such as TSMC (TSM), Global Foundries, or UMC (UMC). Oh, and Intel's manufacturing processes are significantly more sophisticated than the competition.
To complement this time-to-market advantage, Intel has now initiated what they call "tick-tock," which means the firm releases a new micro-architecture on a given process in one year and then in the next year, moves that micro-architecture to a new process to allow for more cores, lower power consumption, and small design tweaks. This makes Intel a very dangerous competitor, and it remains to be seen how ARM and its legion of licensees will combat this fully vertically integrated chip provider.
It's Not All Rosy: The Risks
While I do believe that Intel will see significant success in this space, there are also some very real and significant risks that those considering investing in Intel for the mobile strategy should be aware of.
First, while Intel has managed to overcome the obstacles to penetrating the Android ecosystem, it's not at all clear that Intel will make headway in the Microsoft (MSFT) "Windows Phone" devices. The apparent strength of the relationships that the software giant has with both Nvidia (NVDA) and Qualcomm via the "Surface" and reference Windows Phone 8 designs, respectively, provide reasonable evidence that Microsoft may want to stick with ARM-based chips for its phones.
While I am sure that Intel would be more than happy to help Microsoft get current and future iterations of Windows Phone up and running, it is becoming more apparent that Microsoft wants to distance itself from Intel to embrace the "new" technology. While it is not at all guaranteed that Windows Phone 8 will make a meaningful dent in the smartphone space, Nokia (NOK) Lumia sales are quite encouraging for the new ecosystem.
Another risk is Qualcomm. Unlike Nvidia or Texas Instruments (TXN), Qualcomm licenses the ARM instruction set then designs its own CPU cores for their SoC solutions. This allows the firm to get a "head start" on the other SoC vendors, as it does not need to wait for ARM's next design. In addition, the company is firmly entrenched in the smartphone ecosystem, including Nokia's Lumia, Research In Motion's (RIMM) next BlackBerry, a number of HTC phones, and even some versions of the Samsung Galaxy S III.
Finally, while Intel currently has the most sophisticated transistor technology, it still needs to bring its "Atom" lineup into sync with its notebook/desktop chips in terms of process tech. Intel's "Ivy Bridge" for PCs has been available since April 2012, built on the 22nm process, but we will not see 22nm smartphone/tablet chips from Intel until late 2013. Intel has openly stated that it is aggressively working to bring these two into lock-step, but a risk is that they are not as successful as expected in doing so.
In short, it is no longer correct to assume Intel is not going to be competitive in the smartphone space. "Medfield" is a very capable chip against today's ARM Cortex A9 designs and even the Cortex A15-class Snapdragon S4. With Android support in place, Intel's TAM is fairly large (but it could be larger if the firm could secure a place in Windows Phones; a Nokia win would be incredibly good news). Further, the company has been scoring a number of design wins with Lava, Orange, Lenovo, ZTE, MegaFon, and Motorola Mobility.
It's not clear that Intel will be the "winner" of the smartphone SoC space, but it would be irresponsible to count them out of the race, especially as there is an entire world outside of the USA that will be moving to smartphones.