And honestly, I feel one of the single largest barriers to getting games onto Linux is support. Oh my lord supporting software on Linux is a logistical nightmare. There is no single base configuration you can count on. One can't assume that anyone will have X version of Y library, or that they haven't done something hacky to circumvent a problem on their system. The state of Linux video card drivers is also a shakey thing. They lag behind Windows drivers as a general rule. Same issue with sound. Toss in all the different distributions there are out there that behave very differently, and you have a test and support matrix that is crazy.
And what is keeping you from releasing it without any form of support? There are two very viable options:
1. Just get one Wine guy in your team. You will probably even find a volunteer and don't need to pay anyone. Or maybe you pay a little bit, one programmer isn't that costly. He will make sure the game runs fine with Wine. Then people can buy the Windows version and run it on Wine on Linux on their own. No additional Linux support needed.
2. Release Linux binaries with a big fat comment saying "hey, we do that because we are awesome and want to support Linux but these binaries are experimental, no support whatsoever. People on forums might help you but there is no official support". Linux guys are very used to helping themselves. Should really be no problem at all.
Another thing about the market size: Yes, the Linux gaming market is smaller then the Windows one. But on the other hand there is nearly no competition! You have to share the Windows gaming market with about 50 good games each year while there is only one of the high profile titles with native Linux support every few years.
So yes, the Linux market may only have 5% of the size of the Windows market. But it has only 1% of the competition.
From all the replies made here you can clearly see that there is a demand. The main reason I own all games from id Software is their native Linux client.
As for the "porting to OpenGL is costly": Making the D3D game work well with Wine is not costly. All it requires is a friendly e-mail to the Wine team and I am sure they will gladly help out as much as they can, doing basically all the work for you.
And is there any reason to develop a new game with D3D instead of OpenGL? I simply see none. OpenGL (with SDL) can do everything that DX can do and is portable.
I remember a time were every game was shipped with tons of DRM. Everybody thought DRM is the way to go and shipping games without DRM will ruin the sales. All big gaming companies followed that rule.
But a small company called StarDock thought differently. They went against the DRM, promoting their views of customer service and they had an enormous sucess with it. It was a real upset in the gaming market and proved the "big guys" wrong.
Right now, the same "big guys" are not thinking about Linux when developing their games. Developing a major game to run well on Windows and Linux is a leap of faith, yes. A leap of faith just like releasing games without DRM was. StarDock has been very successful with that leap of faith which should encourage more of them ...