At Advanced Micro Devices' (AMD -0.67%) recent Core Innovation Update presentation, the company announced that it will take ARM Holdings (ARMH)-based intellectual property much more broadly into its product portfolio. In particular, the company indicated that it will attack the client space with both X86-64 compatible CPUs as well as ARM-compatible processors. Believe it or not, this strategy actually makes a whole lot of sense.
Project Skybridge is an easy and cheap way onto Android
Right now, Intel (INTC -2.85%) is fighting rather mightily to make X86-64 a first-class citizen on the Android platform. To date, the company has done a really good job, but there is still work left to be done -- although that work should accelerate as its tablet penetration increases. While AMD could rely on the trailblazing that Intel is doing for X86-64 on Android, it can fairly cheaply buy its way onto Android without needing to piggy-back on Intel's enablement efforts -- ARM.
Source: AMD.
What AMD is doing with Project Skybridge is building a common system-on-a-chip with two flavors available to the customer: an X86 variant, based on AMD-designed low-power X86 cores, and an ARM-based variant, which uses off-the-shelf ARM IP. Indeed, AMD executive Lisa Su indicated at this event that the client-based ARM products would be based on ARM's own IP core roadmap. (AMD's custom K12 ARM core is aimed at different performance/power segments.)
This, in a nutshell, allows AMD, once all of the driver work is done for Android, to play in the Android space with an ARM part differentiated from whatever other IP blocks the company does in-house -- and in this case, graphics could potentially be that big selling point. It's an interesting strategy in theory, but of course it comes with caveats.
Jack of all trades, master of none?
If you look at the trend in AMD's R&D spending, you'll see that it has been down quite significantly over the past five years. In turn, AMD's competitors in both the X86 and ARM space, although it's all really the microprocessor space, have been ramping up R&D significantly.
AMD Research and Development Expense (Quarterly) data by YCharts.
Indeed, AMD's R&D budget is about a tenth of Intel's, yet it is going to attempt to compete with Intel head-on in the development of an X86-64 core as well as a "big" ARM core, along with a bevy of SoCs around those cores and IP targeted at a number of different segments. With fingers in this many pies -- tablets, PCs, servers, graphics, and such -- the risk here is that AMD does play in all of these segments and has a theoretically large total addressable market, but ultimately fails to capture a meaningful portion of it.
Foolish bottom line
AMD is smart to offer essentially the same SoCs with both ARM and X86 varieties, particularly if it wants to have a viable tablet story. While it really does look like Intel is opening the door wide open for X86-64 on Android with an aggressive enablement effort, AMD appears to be unwilling to bet its tablet strategy on that.
Further, while AMD's "ambidextrous" strategy does indeed open up a large total addressable market of essentially all performance-focused microprocessors, the distinction between "ARM" and "X86" is really quite arbitrary, as there are very few markets -- for example, Apple's iOS devices, relevant to AMD where a merchant ARM chip can go where a merchant X86-64 chip can't.
