Take my answer with a grain of salt because I am not really a graphics programmer but I say "Nope, not outdated"

My reasons are this:

Technical papers written by Nvidia actually tend to be more useful after they write them as apposed to game developer's papers that are more useful immidiately. Nvidia is commonly trying to take advantage of new features of their video cards in their demos which you may not be able to assume exists for a couple years to come. If anything being a couple years later may just be helping the "usefulness" of the paper.

That said, shadows needs change a lot game to game and the algorithm may not be appropriate depending on what your needs are. From number of lights to visual style and available memory different algorithms may be more appropriate. I would fully understand the strengths and weaknesses of an algorithm along with my exact needs.

Haven't read that particular NVidia write up, but shadow-volume techniques are still being used in a lot of games.

The main alternative is shadow mapping.

If you're talking about your average game for the PC/Xbox 360/PS3, then shadow maps have become much more popular than shadow volumes. This is mainly because shadow maps are more straightforward and flexible algorithm, and also scale much better with geometric complexity. There's also much more promising research into new methods for improving the quality of shadow maps, like all of the recent variance shadow map stuff.

Shadow volumes were pretty popular on the previous generation of consoles (and the Wii) because they made sense for that hardware: you didn't really have flexible pixel shaders to work with, so creating shadows by doing some vertex work on the CPU or in the vector units of the PS2 made sense.

Some games do actually mix both...for instance shadow volumes handle point lights better since they don't require rendering depth to 2 or more buffers for a single light source.