There are a lot of calls for a native exporter in this thread, but it's still a very unlikely project IMO:
- renderer bottlenecked games will not improve performance with a native exporter. I keep saying this, but unless you can prove your game is logic-bottlenecked, it won't be faster with a native engine. It's not the holy grail of fast performance; mobile devices are still weak compared to desktops.
- we would have to effectively stop our weekly updates for 6-12 months while developing this
- HTML5/webview performance could improve so much in that time that it makes the whole project pointless (as @prinsukun said, the improvement from dec 2011 to dec 2012 is enormous, and maybe some newcomers don't realise the extent of that)
- it will cause ongoing painful compatibility problems with third-party plugins and behaviors, which most likely won't work without rewriting per platform
- it will cause a new class of bugs where one engine might work slightly differently to the other, breaking games in subtle ways
In short, it would take a long time, it probably won't be as good as you think it is, and could be pointless.