A single Windows platform for phones and PCs: Can it be done?

Some see Android as the biggest looming competitor to Microsoft's Windows Mobile. In fact, Android is more than that: It is a potential threat to both the Windows Mobile and Windows client platforms.

So does it stand to reason that Microsoft is planning a single line of defense against the Android platform? Or will it continue to field different operating systems, optimized for different devices and processors, as opposed to a single Windows variant, to compete with Android?

Currently, Microsoft advertises that Windows runs on everything from phones, to netbooks (at least when it comes to XP and Windows 7), to souped-up laptops and desktops. But this is an oversimplification. Windows Mobile is not Windows. Even the next two releases, Windows Mobile 6.5 and Windows Mobile 7.0 are built on top of the Windows Embedded CE, not the Windows client, kernel.

This distinction gets more interesting when ARM and other low-power processors are added to the mix. Will Microsoft be able to get Windows 7 (8, 9, etc.) to run natively on ARM? Or will Microsoft only be able to get some Windows CE/Windows Mobile variant to run on the ARM-based PDAs and netbooks waiting in the wings? (Update: To make sure this is clear -- I know Windows Mobile already runs on ARM phones. But so far, Windows does not run on ARM, at least not "officially.")

If Microsoft can't and/or won't port plain old Windows to ARM, it would stand to reason that existing Windows apps won't be able to run without substantial tweaking on these new ARM-based netbooks and PDAs  the way they can run on x86-based netbooks.

One theoretical way that Microsoft could head off this problem is to make Windows, not Windows CE, the core upon which future Windows Mobile releases are based. Could it be done? Should it be?

Matt Rosoff, a research analyst with Directions on Microsoft, said delivering a Windows-based mobile operating system might be feasible but doesn't seem to be where Microsoft is focusing its attention:

"I think it's technically possible (power consumption might be an issue), but I don't think Microsoft's moving in that direction in the Mobile 7 time frame, and I doubt they have concrete plans beyond that.  Right now, Microsoft's mobile strategy is focused on moving 'up the stack' with hardware reference designs and attached services. They view Apple and RIM as the competitors to beat. So they're less interested in being a commodity OS provider, which is where Android seems to be playing. (Although Google expects to make money from attached services as well.)"

Michael Cherry, another analyst with Directions on Microsoft, agreed that Microsoft probably could port Windows to other processors, but wondered whether such a move would make business sense:

"Regarding the porting of the Windows OS (say Windows 7) to the ARM processor, in my opinion, it could be done. Consider that Windows 7 is a relative of Windows NT, which has at its heart an architecture that has supported a bunch of processors in the past (Alpha and MIPs come tomind), and currently supports a bunch of similar and yet disparate processors such as x86, x64 (AMD and Intel) and Itanium.

"But the subsequent testing to make sure everything still works is a lot of work. People have expectations about what they would be able to do, for example, run Office and other Windows applications. Connect all the peripherals they own. This spawns a significant cycle of test, fix, re-test again (or if you prefer: Lather, rinse, repeat)."

One developer close to Microsoft, who requested anonymity, agreed that moving Windows to a new architecture like ARM would be quite the task. The developer explained:

"Microsoft ships NT based solutions as binary compiled images, even for embedded scenarios. They provide a kit (Windows Embedded) to allow you to build a custom componentized image, but what they do with CE is far different -- they actually ship the source to IHVs (independent hardware providers) and they build the system complete with customizations for their hardware. Doing this for an NT derivative would be a pretty difficult proposition, as the NT build process today is highly structured and controlled. Productizing that for outside use is a non-trivial obstacle to having people source build their own NT OS images (not to mention, the NT source is one of the company's crown jewels).

"Lastly, there are other obstacles as well, namely some smartphone devices likely need real-time (hard interrupt service routine latency) support, something NT does not have (nor likely ever will, as it would require a substantial and undesirable change the driver architecture) and CE has a lot of infrastructure around smartphone hardware that would have to be ported."

Microsoft, for its part, isn't talking about its porting plans for Windows. Officials declined (again) to comment on reports that Microsoft has Windows running on ARM systems in backrooms somewhere. And trying to get the Softies to talk about Windows Mobile (beyond which designer wallpapers are on tap for Win Mobile 6.5) is like trying to get Chairman Bill Gates to allow his family to bring an iPod into the home compound.

What's your take on Android vs. Windows and/or Windows Mobile? What do you expect Microsoft will do here to try to head off Google's OS before it gains steam on multiple mobile platforms?