Last week I wrote an article wondering whether x86-based tablets running "real" Windows will ever be as thin, light, and long-lasting as ARM-based tablets like iPads and Androids, because based on a quick glance of the current x86 tablets, they're either (1) really underpowered or (2) have short battery life. To be clear, I don't necessarily blame Intel for this. Instead it's that the Windows OS was designed for a different world with different devices than we're using now.
This leads to an interesting thought exercise: if the whole "Windows-versus-whatever's-next" conversation in enterprise IT is really about Windows desktop applications, won't we ultimately get to the point when a $500 iPad has enough power to run any Windows desktop app we need in a VM?
For all this talk about Windows apps and how much the enterprise needs Windows at the desktop, most of it is about legacy Windows desktop applications. In fact that's what VDI and RDS are all about. I mean if you're going to use new applications, they're either web-based / HTML5 / SaaS apps that already run everywhere, or they're apps that are written for the "post Windows" world of iOS, Android, or Windows TileWorld.
That said, if Windows is all about the legacy desktop applications, then at some point we stop getting new Windows desktop apps. Again, I'm not saying the desktop, laptop, keyboard, or big screen is dying or going away, rather I'm just suggesting that if we need a desktop app in 2020, it will not be written as a legacy Windows desktop app.
If that's true, that means that at some point, new Windows desktop applications stop getting created. Whenever that happens we'll essentially have a "freeze" for the hardware requirements for those apps.
You can kind of see this today. Whatever a Windows desktop application needs in terms of hardware doesn't change after the app is released. Office 2000 still works fine on a 600MHz single core system with 20GB of storage and 64MB of memory.
Now compare those 2000-era desktop PC specs to the power and capabilities of an iPad. The current iPad has a dual core modern architecture 1.4GHz processor, 1GB of memory, and 64GB of storage. It ought to be able to run the real Office 2000 suite no prob. Sure, you'd also have to get Windows XP loaded on it, but that shouldn't be too difficult to make work in a VM. And sure, iPads are ARM-based and Windows XP and Office 2000 are x86, but how hard would it be to emulate that in the VM? Yeah it will take some overhead, but remember today's iPad has orders of magnitude the horsepower of a 2000-era desktop, so again, we're well within the relm of possibility. (There are several existing open source projects, such as Bochs, which can do that today on a jailbroken phone.)
Of course even if this Windows-VM-on-an-ARM-tablet technology was ready to go today, you might be thinking, "But it's 2013! I don't want to run the 13-year-old version of Office. I want the latest stuff, and the latest stuff needs way more horsepower than what an iPad can provide!" This is true. But remember my whole theory is based around Windows desktop applications eventually stopping development, replaced by HTML5 apps (or whatever). So imagine for a moment that Office 2013 is the last version of Microsoft Office that runs as a Windows desktop application. Fine. If it needs a 2.0GHz dual core x86 CPU with 4GB RAM and 10GB disk space, don't you think that will be no problem for the iPad 11 in the year 2020?
By the way, I'm not suggesting that this is going to replace VDI as the go-to solution for legacy Windows desktop applications. I'm just pointing out that when the world moves off of them, we'll essentially have a snapshot-in-time hardware requirement for the last generation of Windows apps, and if we wait long enough after that, even our phones will have enough horsepower to emulate them properly.
We're already seeing this today. Remember the sickest, hottest arcade games that you couldn't even run on the best home system a few years ago? The original arcade machines didn't run on ARM, yet all those apps are in the iOS app store today and they all run perfectly on iPads. Windows apps are next.
Throughout this article I've used the iPad as an example, but it could just as easily be Android or anything else. And why wouldn't this be possible? There are YouTube videos of people getting Windows 95 to boot as a VM on Android. And you know VMware's probably got versions of this running in some basement somewhere.
So even if the performance sucks, give it a few years and throw more hardware at it. Done.
By the way, I recognized that running a Windows VM on some future ARM-based tablet will still require a Windows license. Fine. But it won't require and x86.
What's the point of all this? Just that the "we need x86 to run Windows desktop applications" argument isn't going to last forever, and that if ARM turns out to be the right processing architecture for the future, so be it.
Interesting to think about. What's this mean for Intel? What about Microsoft?