What’s the point of a Windows 7 ARM port?

Last week saw the emergence of speculation, instigated by none other than the CEO of ARM, Warren East. In a conference call with reporters, East suggested that Windows 7 support for ARM would make the netbook space "a very different marketplace," but he declined to elaborate, suggesting that there could eventually be ARM support, but it's up to Microsoft to comment on the possibility. (The EE Times was on the call and has the exact quotes used.)

I have no idea if Windows 7 will support ARM or not, and I won't speculate on whether this will happen. But I will suggest that East is wrong in his suggestion that a Windows 7 ARM port would shake up the netbook space. The reason is straightforward: an ARM port of Windows 7 would be a Windows that doesn't run very many Windows apps, and thus it would be pointless. Not only would it be pointless, but redundant, since Microsoft already has such a Windowless ARM-based "Windows" flavor in Windows Mobile.

ISA transitions are easy to propose, hard to pull off

As all of the Apple developers who lived through the PowerPC-to-Intel switch know, ISA transitions are painful, major events. Contrary to popular belief, and ISA port involves a lot more than just checking a different box before compiling your application. It was fairly long and labor-intensive process to get the entire Mac ecosystem transitioned to x86, and that ecosystem wasn't nearly as sprawling, unruly, and riddled with legacy issues as the Windows ecosystem.

The difficulty of getting anyone to port their Windows app to ARM on a Windows 7 ARM netbook would probably start life with a basic Windows 7 install, very limited driver support for peripherals, and a limited application lineup—probably something like a calculator, Solitaire, possibly Microsoft Office, and the handful of native .NET apps that are floating around out there. As for the rest of the Windows application base being ported, Peter Bright, our resident Windows developer, tells me that fat binaries (a la OS X) aren't feasible with Windows' current executable format. So developers would have to sell separate ARM and x86 versions of Windows apps like they did for NT in the Alpha days.

In sum, an ARM-based Windows 7 netbook just wouldn't run very many Windows applications, and if you can't run Windows apps on your netbook, then why not use Linux?

I suppose that Microsoft could put together a stripped-down Windows 7 "Netbook Edition" that's an ARM port that runs a limited suite of netbook-specific apps, but they already have a Windows Mobile, which is an ARM-based OS that runs a limited suite of mobile and .NET apps. Why not just promote Windows Mobile for ARM-based netbooks, instead?

So there really doesn't appear to be much of a point to a Windows 7 ARM port, which isn't to say that Microsoft won't try it. Companies in Microsoft's position—shrinking revenues from cash cow properties, layoffs, and under threat of losing market share to a more nimble competitor in a hot and growing sector—often make boneheaded, counterproductive moves.

If Microsoft really wanted to shake things up and take on Linux, the company would develop one single kernel and platform to run across desktops, servers, phones, and the Xbox. But even then, Linux would still retain one important advantage beyond its one-kernel-fits-all approach: Linux can never come under antitrust scrutiny for being too successful.