Advanced Micro Devices (NASDAQ:AMD) offered investors reasons to be hopeful at its latest APU13 developers conference. There was a flurry of announcements and plenty of buzz, but when everything is distilled, the question on investors' minds is simply whether the company can successfully grow its top and bottom lines in a sustainable, healthy way. While AMD has many challenges ahead (both secular and structural), the company made a number of announcements at APU13 worth examining more closely.
Say hello to Mullins and Beema
Earlier this year, AMD launched system-on-chip parts targeted toward low-cost notebooks, convertibles, and -- to some degree -- fanless tablets. The one aimed at the notebook portion of the spectrum was called Kabini, and those targeted more toward convertibles and fanless systems went by Temash. AMD investors and management alike were hopeful that these would finally enable AMD to take share against Intel (NASDAQ:INTC) in notebooks, while at the same time penetrating the high-growth tablet market.
However, while Kabini was solid for low-end, low-cost notebooks, Temash utterly failed to impress. It was too power-hungry and lacked the necessary deep, low-power states to successfully power fanless, thin tablets in the same way that Bay Trail from Intel and numerous chips from ARM (NASDAQ:ARMH) have been able to do for quite some time now. While Temash saw some design wins in low-end notebooks, and even some bulky 13" detachable notebooks, it was out of its league in the low-power space.
AMD claims, however, that it has been hard at work revising its Jaguar processor core and reworking critical elements around its system-on-chip solutions. The fruits of this labor yielded the successors to Kabini and Temash, known as Beema and Mullins, respectively. AMD claims that performance per watt has doubled over its previous parts, although it isn't using actual power measurements for the "performance per watt" comparison, but instead the rated thermal design power numbers. If this is true, AMD will have a better shot at finally playing in the low-power tablet space.
Is AMD's biggest advantage moot?
AMD's biggest advantage in this space is that it's the only player (other than Intel) that is legally able to produce an X86-compatible design which, in light of the recent OEM abandonment of Windows RT, seems to be the only instruction set architecture that can successfully play in the Windows space. Additionally, AMD has significant graphics/gaming expertise that could help target OEM designs that are more focused on gaming.
That being said, it's difficult to ignore that Intel has been improving its graphics engines at an incredible rate and that it will -- when Beema/Mullins launch -- be shipping low-power Atom parts built on the company's 14-nanometer FinFET process. Given that AMD will be on TSMC's 28nm high-K metal gate process, it will be very difficult (speaking strictly based on the laws of physics) for AMD to have a competitive design on a performance-per-watt basis.
Even if AMD had designs that were more clever than Intel's, these designs would be fundamentally limited by the performance and power characteristics of the transistors from which the design is built. In other words, all of the problems that keep AMD from meaningfully competing against Intel in the high-end PC space are also present in the low-power/low-cost space.
There's still hope
In scenarios where power consumption is absolutely critical, AMD will have a tough time winning designs against Intel. However, in the low-end notebook space, as long as AMD can provide a "good enough" part at a price that significantly undercuts Intel (since AMD is willing to live with lower gross margins), it can probably do OK. In fact, even with Kabini, AMD was able to score the Samsung Ativ Book 9 Plus.
However, the real story for AMD is its semi-custom business. All of the experience it is building with these chips, and all of the IP that it is building in these products, can be reused for a number of interesting purposes. For instance, the Jaguar cores that didn't do well in tablets found a wonderful home in both the Xbox One and the PlayStation 4. Future AMD IP can be recycled in a similar fashion to meet different end market requirements.
Foolish bottom line
AMD is a risky bet and the company faces a number of competitive headwinds, but as long as the company continues on its semi-custom path, the IP it develops for the PC space won't be for naught, even if share gains prove elusive.