So
OpenGL 3.0 is out now, and it doesn't contain what they've been promising for a long time now: the rewrite of the API. Instead, it's basically an OpenGL 2.5. Game developers are in rage and feel cheated. I've never read so many words like 'unbelievable letdown', 'epic fail', 'fiasco' or 'disaster' in a single thread like in the one of the
gamedev.net forum.
What happened is that it seems the
khronos group wasn't capable to agree on the technical details of the redesign of the API and decided to release an update of OpenGL 2.x as OpenGL 3.0 now. And together with the non-existing communication of the group with the developers during the last year, game developers really have a reason to be angry. Lets face it: OpenGL today still is a collection of partially outdated, crappy documented extensions, not a solid, clean API like Direct3D. OpenGL 3.0 was supposed to fix this and because they failed, a lot of developers now will begin to support Direct3D instead of OpenGL. Simple as that.
OpenGL won't die though, if you are forced to write 3D applications on non-Windows machines, there is no alternative to OpenGL, unfortunately. But as PC game developer you can savely ignore this - 98% of all PC-gamers are using Windows, so why going through the pain of using this antiqued API just for maybe getting 2% more users instead of simply using a nice, clean Microsoft written API like D3D?
Nope, OpenGL won't die. But this just was a step into its grave at least.
Adding it all together, I think it's muuch larger than 2%. In addition to this, windows gaming is kind of saturated. I saw some interesting statistics from a smaller gaming company delivering cross platform games to windows, mac and linux (java based). I can't recall its name, so no sources. But the percentage of mac/linux buyers (especially mac) was exceptionally high, despite only having like 2% of the market.