I've been using a Droid Incredible for 2 years. It was released 1.5 years before that. It handles all of my mobile computing needs just fine, CM7.2 is sturdy as hell (just passed 30 days uptime) and I haven't spent money on mobile hardware for 2 years.
I've gone wifi-only since buying this phone. I have wifi everywhere I need it except grocery store checkout lines. The phone lasts 2 days on a charge in perpetual airplane mode too.
the definition of "enough" will change in 2013
This thing will sit in tablets, nettops, maybe even consoles.
I know the steambox is likely to be x86, but with the power on offer here you could certainly make a living room media-box and games-machine in one.
So in so much as we can do more faster with our phones I think it is not a bad thing to add computational power into our phones. One day we will all have one handheld to rule it all - phone, tablet, desktop with external display, gaming console etc.
Imagine that you just dock your phone into a screen when you want to work and that you don't have to haul a laptop around with you anymore. To get there, we need the machine-power-per-square-inch to shrink further and this is what drives that.
You also have to remember that these clock numbers don't really mean anything. The processors are designed for low power consumption, and frankly, it shows.
I do.
Higher resolutions, especially for tablets. For example iPad 4 has only now gotten enough GPU performance to be as fast as iPad 2 running at its old native resolution. So iPad 4 at 2048x1536 is as fast as iPad 2 at 1024x768.
So basically the iPad has stagnated for a year and a half in performance, just to get the benefit of that crisper display. And only with new upcoming chips, the iPad will start to get higher performance for native-resolution apps.
However, about speech recognition for instance, it is for now done "in the cloud": your recorded voice is sent to a server which does the intensive computations and sends the corresponding text back to your phone. Why would that change? It even allows Google or Apple to train their speech recognition system so I think they will want to keep it that way.
More generally, more and more computation are done in the cloud, even on the desktop (Chrome OS).
That said, I agree about high-res gaming and augmented reality, etc.
Also, hopefully, the more power for the same TDP, the sooner we can get phones which much better TDP for the same power /hopes
However this also may be because when they make the chip twice as fast - for the same content as before - and using the same energy, that means that once content catches up with the chip, and is 2x more bloated/advanced, then the chip will consume twice as much energy processing it. I'm making it sound a lot more simplistic than it probably happens, though.
We don't only use the phone for calls, but also for:
1) high performance 3D games 2) demanding web use, with Canvas, SVG and other technologies 3) image editing 4) movie editing 5) audio editing 6) hi-res video capture and playback
And at some point we will do several of those at once.