Is it reasonable to target OpenGL 3.2 in modern games?

Question

I'm currently working on a game/engine that uses OpenGL for rendering, and recently have begun phasing out support for the fixed-function pipeline. I could target 2.1, however I would like to use tessellation or geometry shaders and various newer GLSL features that don't exist in 2.1.

Considering that the game will not be ready to be released for at least another year, would it be reasonable to target 3.2 as a minimum base?

Note that this game does take advantage of newer features such as MRT framebuffers for deferred lighting, etc, and I don't intend on supporting fixed-function hardware at all. So it's either 2.1 or 3.2, and I'm leaning towards 3.2. Does anyone know a very rough figure of how many people have 3.2 compatible cards and the rate at which they are being adopted?

Answer 1

I'd say that users of Valve's Steam covers a considerable percentage of the PC gaming population, in which case, the Steam Hardware and Software Survey would be a pretty decent source of information. I don't see any OpenGL support statistics readily available, but you might be able to infer that information from the graphics card statistics.