Blog Post

4 Actions Microsoft Can Take With an ARM License

ARM (s armh) and Microsoft today announced a new licensing partnership that will give Microsoft (s msft) more research and development opportunities with chips that run many of today’s consumer electronics devices such as mobile phones. Prior to today’s agreement, the two companies had a more distant relationship: Microsoft-powered handheld devices such as Windows Mobile handsets and Zune media players run on chips built by other ARM licensees. In contrast, Windows software runs on chips from Intel and AMD (s amd) which are based on the x86 architecture.

To be sure, Microsoft is certainly familiar with ARM processors — it has made its smartphone software work with ARM chips for years — but the new licensing agreement provides far more access than a CPU spec sheet. As Kerry McGuire, director of strategic marketing for ARM, explained on a conference call with GigaOM today:

This is about getting a hands on feel for the ARM architecture, a richer dialogue around what the tech is capable of and a way to experience the limits and potential of what the tech can do in a way that is different than reading a high-level specification.

Indeed, even though Microsoft is a software company, the license it now has with ARM allows for custom chip design and creation, which offers a number of possible scenarios, both in terms of hardware and software.

Emulate Apple and design a chip

Microsoft may never actually take the step into chip design, but it has to realize that a major competitor has done just that: Apple (s aapl) licenses the ARM architecture to design the custom A4 chip that powers both its iPhone 4 and iPad. And with Apple selling almost as many ARM-based iPads (3.270 million) as x86-powered Mac computers (3.472 million) last quarter, Microsoft might view this as a huge paradigm shift in computing for which it currently has no answer. Creating a chip, and a new device around it that doesn’t compete with hardware from smartphone partners would allow Microsoft to face Apple head-on in the mobile computing space. Perhaps this is the missing ingredient to the Windows Tablet PC recipe that Microsoft has been simmering since 2004.

Windows “Lite” on Portables

If Microsoft would rather continue focusing solely on software, the new ARM license could accelerate porting part or all of Microsoft Windows to work on the ARM platform. This would open up an entirely new opportunity for Microsoft at a time when sales of ARM devices are quickly growing. Indeed, ABI Research suggests that ARM-based chips will power more ultra-mobile devices than x86 chips will by 2013. That’s advance warning to Microsoft that a whole class of Windowless devices is forecast, and a deeper understanding of ARM architecture could help offset such a situation.

Less Power-Hungry Servers

Perhaps Microsoft is prepping for a shift to ARM-based chips in servers and experimenting with ARM-powered servers for is online services like Sharepoint and Bing. It may also need to adapt its server software to the architecture as well. For as much progress that Intel (s intc) has made with power efficiency in its Atom line of CPUs, ARM chips are still process more data per watt of power — and the largest cost to any data center is the electricity needed to run the servers. Marvell, another ARM-licensee says an ARM chip can deliver 1 GHz of performance while consuming 700 milliwatts and up to 2 GHz while still consuming less than a watt. The x86 chips perform better, but require greater power consumption: they can deliver as much as 3.6 GHz while consuming up to 130 watts or as little as 1.8 GHz at 40 watts, according to Simon Milner, VP of Marvell’s enterprise group.

Handheld Gaming

Another possibility arose a few hours ago in our weekly live mobile technology podcast. Folks in the chat room suggested a viable option for portable gaming — Microsoft has a large Xbox Live community but hasn’t leveraged it in the mobile space. While the upcoming Microsoft Windows Phone 7 phones will have some type of Xbox Live integration, they won’t be poised as true handheld gaming machines. Even with the Xbox Live effort generating $1 billion a year, Microsoft currently has no portable gaming initiative to compete with Apple (s AAPL), Nintendo and Sony (a SNE).

Related content from GigaOM Pro (sub req’d):

Infrastructure Overview: 2Q, 2010

Mobile Predictions for 2010: A Mobile Gaming Phone

10 Responses to “4 Actions Microsoft Can Take With an ARM License”

  1. Devices are critical and distributed computing are critical to the future of Microsoft. Perhaps the largest problem with the company is the reliance on current business models centered around the PC. What about televisions, mobiles, and cars?

    Being able to move across these boundaries — viewing office documents on a television — is the next step. Cloud and embedded are the future for most functions. My corporate laptop is essentially an Exchange brick, actually it’s Lotus Notes, cough.

    All the focus on mobile or actual devices takes away from the portability required. A Windows television or an XBox portable that opens Word documents is what we will see to address these needs.

  2. I still prefer 1Ghz x86 performance over 1Ghz ARM. I mean a Mhz or Ghz isn’t a dimensional unit to compare different architectures cause i am sure, that an underclocked 1Ghz Core i3 would still outperform any 2Ghz ARM Cortex-A8.
    Marvell’s VP should know that because this comparison makes them look kinda stupid.

  3. Here are the less interesting things I think Microsoft is much more likely to be doing with this license:

    Designing a custom chip… but at the low end: The PC peripherals (keyboards and mice) division produces large enough volume that a custom Cortex-M0 or M3 based microcontroller could yield good cost savings. Input peripherals for game consoles also represents a huge but cost sensitive market. And who knows how many microcontrollers there are in a game console these days? Everything from storage controllers to power-on systems has grown a processor in recent years. Designing a custom microcontroller for these applications could save Microsoft millions of dollars, more than enough to justify development cost.

    Compiler optimization: Microsoft makes compilers for ARM devices in order to support Windows Embedded. Detailed understanding of processor instruction scheduling is required in order to produce optimizing compilers.

    Research: Microsoft Research does some very advanced work, and I can easily imagine that they’d be interested in doing custom chip design for multicore, network, or other applications using ARM cores.

    None of these possibilities is as flashy as the ones you mentioned, but they’re much more plausible applications.

  4. Microsoft designing a processor chip is pure fantasy, at least on their own. Apple has been a hardware company since birth, yet even it decided it was worth buying in processor design boutiques. I’m not sure that Microsoft would really know what to do. This seems more like a defensive strategy, perhaps to spur Intel to get their low-power game going (1-2 W vs. 40-130 W is not in the same continent, much less ballpark). Porting a long-in-the-tooth x86 os to ARM is not trivial, and probably not achievable. Multi-architecture support needs to be designed in from the start.

  5. Karen Tenterfield

    Microsoft’s mobile strategy has been a mess. A debacle which is just getting worse by the month.

    Microsoft does not have the economies-of-scale in the mobile market to make it worthwhile creating its own mobile CPU.

    Regarding a port of Windows 7/8 to ARM, that doesn’t make sense either. It will lose most of its x86 apps, which couldn’t be ported, removing any advantage of the Windows platform.

    Microsoft is between a rock and a hard place. Shoehorning Windows 7 into portable slates is also a failed strategy, as it is uncompetitive against Android and iOS which were designed from scratch for touch interface, and so were their apps.

    Then again, Microsoft’s actions in the mobile market so far have been illogical, and made zero sense, so I guess the company could do anything.