There is a difference between "block" and "no longer support." Anyone can use older drivers to make PhysX work under Nv+ATI config..I don't care of nVidia doesn't allow PhysX in AMD cards, I do care if they block PhysX from their own cards when the primary card is AMD.
Lets be fair for a second, if you ain't buying newer product from Nvidia, than why shall you benefit from newer codes from Nvidia via its drivers?
In high-level speech, Nvidia video cards are designed can handle both physX, cuda, and generate graphic contents. Yes, it can also serves as a off-load unit when the primary display card is also from Nvidia. Otherwise, Nvidia must extend their testing consisting all video cards from other companies. That is an expense Nvidia is not willing to pay.
Again, newer video cards all follow the WDDM standard, so in theory, Nvidia driver can work on ATI cards and vice versa, but they don't. If someone ask "Why doesn't ATI driver works for my Nvidia card?" and the answer will probably be "are you flicking dumb?" Which is perfectly logical right? You may say "But it works without any problem whatoever!" Well that is simply coincident. Gratz if you have tried and succeeded. Again, the real dirty part about Nvidia is that they will actually allow a driver get out of their doors that does not prevent a Nvidia card to act as a PPU. I mean seriously? It looks like a deliberate act than an accident to me so people will be able to taste what PhysX is while using their ATI card as the primary display card.
Note that neither Nvidia or AMD support cross vendor setup. You can complain to AMD when your Nvidia card stops working after you put in your new AMD card. To be a bit technical, each video card has an ID which allows drivers to detect its presents. If Nvidia can disable its card as soon as it detects the existence of an ATI card, AMD card also disguise itself to fool Nvidia's detecting algorithms.
Of course, little red riding hoods won't tell you that. Instead, they will actually give you a little hacking script/method to enable it. That way, neither Nvidia or AMD are responsible. And the best part is, if it fails, user will believe that it is because they tries to hack it. Those are the cool tricks these companies' use.