Far from it...
As the manufacturing process shrinks, and transistor density increases, GPUs become increasingly constrained by power draw. This trend has been growing for the past decade with GPUs, and even longer with CPUs. Now that we are approaching the limits of what can be cooled comfortably with air in a dual-slot config (around 300-400W maximum realistically), the key to unlocking performance increasingly becomes efficient power management.
This is no great secret... I've been discussing it on these forums for a couple of years now (
see here for example), and no doubt the engineers at AMD and Nvidia have been investigating it for far longer.
The first foray into power management was AMDs Powertune, introduced with Cayman, which approximates power draw based on GPU activity, and places a cap on the maximum power draw. This allowed them to ensure that the GPU will fit into a particular power envelope, even on high-power 'outlier' applications (like furmark for example).
The "dynamic clocking" present in the 680 is just an evolution of this concept; allowing the GPU clock to be adjusted in order to maintain a more consistent power draw over a wider range of applications (again, if you check the thread I linked to you will see discussion on the potential for such a mechanism). I have no doubt that we will see something similar from AMD in the future. AMD were first to the table with power containment, and Nvidia have improved on the concept. It's the nature of the business that the two companies continually leapfrog each other in this fashion, and I expect AMDs answer to be even more advanced, when it arrives.
All this is quite separate to the issue of overclocking headroom on the cards. The 7970 clearly has a lot more overclocking headroom than the 680, which is clocked much closer to its limit. The 7970 is something of a special case when it comes to overclocking headroom... In the past, very few high-end cards have been able to offer more than ~10% overclock comfortably.
