What....like DX10.1 and multithredded games? if the take up of Px is twice as fast as they have been, that means it will be 10 times too slow for the consumer. Damn, DX10 cards have been around for 2 years now and there is still only a smallish proportion of all the games produced being DX10 enabled, for the sake of 5 games today personally I just dont see the point in buying an extra card, not when one can do the job if it's half decent. In GRAW2 I did see a nice effects difference with Px enabled, thing is, after playing it for a couple of hours.....you dont even notice the difference!
IMHO there's a big difference that could make PhysX faster to become widely accepted. Well there are many.
First of all, unlike the other two, physics do add something to the mix. What does DX10.1 (or DX10 for that matter) bring to the table? Nothing really, nothing special or new or even worthwhile to the eyes of the average Joe. Some improvements to AA and a lot of features to slightly improve performance (ejem in theory...) or that help developers a lot, making their lives easier, but don't offer anything tangible. Physics on the other hand, if properly implemented, will change how we play games.
Multithreaded games. That simply doesn't really exist, never did and will never do. Games are single threaded applications and there's nothing you can do to solve that. You can split the thread and run it on different cores as much as you can, one code line after the other whenever it's possible and aditionally you can run many of the secondary features, such as sound, physics and probably AI to some extent in other cores, but they are closely tied to the main program anyway, so you can't make totally independent threads. It's not either a matter of needing more research or effort, there's only one way to go from point A to point B, and it's been long since they learnt how to make a line. The only way of taking advantage of such single threaded applications into multithreaded CPUs is through virtualization or any other possible lower level parallelism method. There are many advancements in this area, but we'll have to wait for any mainstream solution, me thinks.
The second main reason (though probably the most important one) are consoles. Consoles are DX9 and until developers feel the average PC is already faster enough that it can run games with all the DX10 features at common PC higher resolutions, they won't bother making different versions. Instead they expect the PC gamer to use much higher resolutions and AA/AF levels to use that extra power of PCs, they also put higher resolution textures that don't really add detail, because sadly they made them interpolating the console version ones, etc.
In that sense PhysX is different. PhysX can run anywhere, CPU, GPU, PPU and it has not different features, just the same instructions that will run the same way in any platform, so it's much easier to scale it for different performance levels. Cell processor if not for anything else is wonderfully designed to run physics on it, and just happens PhysX is AFAIK the official physics solution of PS3 SDK. So Ageia and the PC do have an strong ally in this particular battle.