Clearly you should do more reading. Part of the reason that Fermi was delayed was because nVidia was having a hard time working out all the kinks in TSMC's 40nm process (they are not as experienced at moving to brand new processes as ATi). As for architecture, of course it won't. nVidia and ATi have always had differing architectures, even more so now. ATi prefers smaller, scalable cores whereas nVidia prefers to make a single monolithic core and scale it down for lower-level graphics points.

But yes, Fermi does indeed run hot and is very power hungry compared to ATi's offerings.