Yes, lagging is such a big problem right now with HTML5 and there is much heated debates (most recent on on HN) going on in the tech community, the advocates say that the biggest advantage is to allow "one code base for the same damn thing on multiple devices" while others say that it is not ideal yet because of speed.
Yes, speed is a big problem, but as we follow Moore's Law (and hopefully Koomey's Law as well), speed problems will go away even if Apple/Google tries to retard the mobile browsers (which IMHO they won't).
The real problem is that the same damn codebase cannot be used on multiple platforms.
It just can't. Look at all the complaints Intagram got for its Android release because it didn't follow major Android guidelines. One of the major issues was the bottom tab bar. In iOS, that is totally ok, but for Android that is not an idea solution as the virtual/hardware keys are also right below. And this is just one of the major problems with fitting it.
When we started designing SleepBot for iOS, we looked at so many apps for both platforms, and there are just so many differences that we have to do redesign almost all the major interactions and icons. (Notice how icons have a "realistic feel" in iOS, a modern feel in Android, and a futuristic feel in Windows Phone)
That is Android, now let's take a step forward to Windows Phone 8. Can you even imagine directly porting an iOS app to it. It would look like SHIT and usability would be terrible because that is simply not how it works.
What's gonna end up happening is that developers will start writing native bridges for performance tweaks and UI tweaks. These developers are usually not experienced in doing those as they usually have a web app background, so it ends up either wasting more time and more resources, for something worse.
Someone may argue that there will be "plugins" that will provide awesome native compatibility. Yea, I'd give those a shot, but good luck customizing your app. :)
If you want the best user experience for your users, at least within the next 5 years, go native.
(note: this is not about having a mobile version of your website and very simple one page apps that does not require complex navigation etc )
P.S. If there is really someone that may disrupt everything, look at Parse's new feature: