Considering that this is Chrome, Google is probably looking for a logical way to introduce optimizations into their browser by allowing developers to write in an unmanaged language. This seems like a step backwards in some ways because a managed language more simple syntactically, naturally allows for less bugs (particularly in the dreaded memory realm) and is all around far easier to write than the lower level counterpart. One glaring problem with browser apps, however, is it can never compete with an application in performance. An app created for a browser has an associated overhead because it, by being a script parsed by an application, is necessarily several more steps removed from the processor than a program compiled to native calls in the first place. This feature would erode the performance disadvantage that browser developers occasionally have to struggle with. Naturally, it makes sense that Google would have an interest in introducing a way to have high-performance applications on a browser because, if it works, this would be a feature foundational to Chrome OS, their browser-based operating system.
As far as this taking over conventional browser development, like javascript, there is no chance of that. This would only be a useful tool for a developer trying to get more performance than the traditional script can offer. In any other case, it would not be worth the effort.
The thing I don't understand however, is unless these programs are wrapped in some sort of local API or library suite, aren't they just introducing a huge target for virus exploit?