Technically AMD is still using an old architecture for DX11, while NVIDIA is using a new architecture. So until AMD launches Northern Islands, I consider them to be behind.
Here is the problem with your arguement for Nvidia; you are assuming new means that it is better.
What are the Fermi architecture pros?
Great tessellation performance.
I haven't seen the numbers, but I assume it'll be a great performer in the HPC world.
Good DX11 performance?
But, at the same time AMD's older architecture has some pros of it's own that overshadow much of Fermi.
AMD's old architecture gives much better performance per mm2.
AMD's old architecture gives much better performance per watt.
AMD's old architecture appears to be the correct configuration for 40mm, by that I mean AMD is able to get their parts to market much quicker than Nvidia.
AMD's much smaller more efficient GPU appears to be more scalable, they've had a dual GPU out part (that some people seem to forget is the fastest card) months before Nvidia had anything.
In this case new does not mean necessarily better. Fermi may very well be a great architecture and process technology moves forward, but it appears to have a number of issues today.