Microsoft (MSFT) really shook things up when it announced that its hotly anticipated next generation Windows 8 operating system would be available not only for traditional X86-64 chips from Intel (INTC) and Advanced Micro Devices (AMD) but it will also run on chips based on the ARM (ARMH) instruction set. While versions of Windows have traditionally been available for alternate architectures such as the DEC Alpha, the PowerPC, and the MIPS (MIPS) R4000, these have never really posed a significant threat to the Intel Architecture dominance in the space. With Windows RT, however, the ARM based chips now have a fighting chance.
What Makes This Time Different
The main reason that X86 and AMD's 64 bit extension were traditionally difficult to displace in the PC space came down to the fact that the instruction set was already firmly entrenched in the PC space and most Windows programs ran on x86 only. What incentive did users have to switch to another CPU architecture only to find all of their old programs incompatible?
The release of Windows 8/RT really changes things up, though. To understand why, I need to talk a little bit about two different types of software development:
The first way to develop software is to write "native" programs. This means that the software engineers write the programs in a human readable programming language (but is gibberish to a computer), typically C++, that is then compiled (translated) into something that the particular CPU can understand (its "instruction set). This type of code is much more efficient on RAM usage and CPU power and was traditionally favored due to hardware constraints in computers until very recently.
The second way is to write code that runs on a "virtual machine." The programs are still written in the human readable languages (a big language here is Oracle's (ORCL) Java), but instead of being translated directly into something that the CPU can understand, the code is translated into an intermediate form (in Java this is called "bytecode"). The code in this form is then read by what is called a "virtual machine" which then handles all of the details of running the program on the machine. This is nice because once a virtual machine for a language is written for a given platform (ARM, X86-64, MIPS, etc.), the code is portable across all of the platforms in exchange for a performance penalty.
Microsoft's Bold Move
While developers will still be able to write native applications on Windows 8/Windows RT and target a specific CPU instruction set, the push for "Metro" style applications will be towards cross-compatibility using Microsoft's .NET framework and C# language (this is essentially Microsoft's version of Java).
Make no mistake: Windows RT devices will be limited compared to Windows 8 counterparts at the outset in terms of program compatibility since most Windows programs will still require an Intel-compatible system to run. However, future software will likely be geared to run on both Intel compatible systems as well as ARM compatible systems which will eventually erode the compatibility advantage over time. Given how easily people have adopted Android and iOS, and given that software developers will be quick to want to make versions of their software available on all platforms, it seems that software compatibility really won't be too big an issue for very long.
The Battle In The Windows Tablet Space
So given that the playing field is a lot more "fair" with Windows 8, what will drive this battle? Well, it comes down to battery life and cost. The average consumer has proven time and time again that modern hardware is "good enough" and that what people really want is long battery life, snappy general usage, and cost-effectiveness. Differentiating features, like strong graphics performance, is also a plus.
There's going to be a lot of credible competition in the Windows 8 tablet space on the system-on-chip side. Nvidia's (NVDA) Tegra 3, Qualcomm's (QCOM) Snapdragon S4, Texas Instruments' (TXN) OMAP, Intel's Atom, and AMD's C-series APUs will all be vying for spots in these tablet designs. It is unclear who will "win" the battle here and emerge the market share leader, but this will be critical to watch. Why?
Well, if Intel or AMD ends up dominant here, then the "ARM" threat to Intel/AMD in the PC space essentially goes away and the Windows space stays normal. But if an ARM vendor steps up and dominates the space, the repercussions could be very serious to Intel and wonderful for the ARM space. Widespread Windows-On-ARM dominance will open the doors to ARM-based designs moving up the food chain into more traditional PC form factors.
Positioning For The Showdown
Right now, it's too early to tell who will be the winners and the losers in the space. Intel has the advantage of full vertical integration, giving it the ability to compete well on price as well as performance (more advanced process technology). Further, in terms of chips for higher end Windows tablets, Intel has no real competition.
On the more cost-effective side, all of the major players are vying for a slice of the pie. Intel's advantages on cost/margins remain intact here, too. However, Qualcomm's Snapdragon S4 is faster than any other available system-on-chip solutions, giving it an advantage at the lower end. Nvidia's strength lies in its background in graphics technology, although its Tegra 3 offering still does not quite compete with higher end PowerVR graphics solutions from Imagination Technologies (and used in TI and Apple (AAPL) solutions). Finally,Texas Instruments has the least of a "moat" compared to the other vendors as it uses off-the-shelf ARM designs for the CPU and graphics licensed from PowerVR.
The upcoming Windows 8/RT release will be very important in determining whether ARM-based chips are a credible threat to Intel/AMD in the PC space. It's too early to tell who will be victorious here, or whether Windows 8 tablets themselves will be popular at all, but the Windows space is no longer a two-horse race and it will be imperative for the investment community to keep a close eye on how this all plays out.
Additional disclosure: I may initiate a long position in QCOM over the next 72 hours