The applications shown in games are anything but exciting so far, and so is the concept of buying yet another multi-hundred dollar piece of hardware. I'd like to see what I have actually used before I invest in yet more "exciting" hardware. Physics cards have started off as basically gimmicks if what has arrived is all Ageia can pull off for an all-important first impression.
I'm just not convinced that game devs can come up with compelling uses for physics. It all feels very businessy right now, as in lots of big ideas with very little concrete results. They want us to buy based on ideas. And, since I've read and have been told quite enough about Reality Coprocessors and Emotion Synthesizers, I'm more than slightly skeptical.
I'm also not convinced anymore by the idea that physics can enhance gaming. Half Life 2 did some cool stuff, but now we seem to be at a total stand still. BTW, I've seen flight models on 486s that are entirely convincing (Flight Unlimited). I think it would be cool to see a "game" simulate an alternate reality, where newtonian physics isn't happening, but that seems like asking too much from game devs as it's more than a bit complex (and assumes that game devs can even simulate what we have right here on earth lol).
These announcements by ATI and NV, with their GPUs being said to be physics capable, feel an awful lot like bandwagon marketing. Buzz. Burning up GPU cycles on physics also seems like a wonderful way to make us all need to upgrade quicker (gee, in whose best interests is that?) My next graphics card will have its work cut out trying to render 1920x1200. I don't want it working on other nonsense too (especially effect physics with zippo interaction).