It's no secret that Surface RT and Windows RT, along with Windows RT on other platforms, didn't do as well as Microsoft hoped.
Surface/Windows RT had no shortage of skeptics even when it launched, but it's possible things could change over time. I'm not saying that this *will* happen, but that there's a reasonable scenario for it. Here's how it works:
First, some Microsoft business plan forensics: Microsoft wants developers to write apps for the new, Modern UI (a.k.a. Metro). Releasing Windows 8 only for Intel architecture, they must have believed, would have made it too easy for developers to bypass Metro because conventional Windows programs would already run on it (and on Windows 7 and other versions). But if Surface RT were a success, developers would want to be on it, and would choose to write Metro apps in order to be on both platforms.
Well, that didn't work. In fact, color me surprised at the degree of reticence of developers to write Metro apps, as the sheer number of users who can run them will undoubtedly be very large, even if it's small enough to be considered a failure for Microsoft. Remember, any other company in the world would love to have a disaster like Windows Vista, hundreds of millions of copies of which were sold. Such is the worst you can expect from Windows 8.
This holiday season you can expect to see touch-enabled Windows systems heavily promoted and Microsoft will try other promotions to get people buying apps from the store. In fact, the failure to get developers writing apps for the store is the single biggest problem they have. With good apps users will undoubtedly come, and with users good apps will come.
And if the apps do come, then the decision to buy an RT device could become much more reasonable. There needs to be a cost advantage compared to x86 because the RT will still be less capable, or it will have to demonstrate far better battery life or something to give people a reason to buy it, as opposed to an Intel-based system.
Depending on the performance and power consumption of the latest chips from NVidia and Intel, all of this is possible. It's also possible that Intel will narrow the cost and performance consumption gaps, and RT will lose all its raison d'etre.
But if, come holiday time or later, the Windows app selection is respectable and RT systems are less expensive than Intel-based ones, it might be perfectly reasonable to buy one. If they get cheap enough, people may get casual about buying them.
My money's against it, but it could happen. There have been attempts in the past to put Windows on other architectures, but they have all failed because the Intel has always improved their chip performance enough to make the cost of incompatibility too high relative to the benefits.
You could make a case that Microsoft should have pursued it this way to begin with: x86 first, other architectures once the app market was solidly established. It looks like that will be the Plan B for Microsoft and NVidia, and perhaps it was built-in from the beginning.