Sure enough, right there in the middle of the top row are some numbers that strongly support your claim: 56% of the Steam respondents are using DX10 (which also implies Win7/Vista).
Uh... 56% is a very bad number. You're talking about throwing away half your potential market by developing for DX10 exclusively. You'd have to double your sales with the remaining market in order for this to pay off, which is highly unlikely.
This is really straightforward: in order for it to be worthwhile developing a DX10 game, the increase in sales from the higher-quality product must exceed the lost sales from people whose systems cannot run the game. For most games, that probably means DX10 won't be viable until market penetration gets up into the high 80's.
When it happens, it's going to be all at once, as the remaining DX9 holdout users upgrade en-mass. Until developers see it as a worthwhile investment, however, it's really easy to remain a DX9 holdout and so the transition is going to be delayed.
You're not getting the entirety of Steam Users in that survey, and less than half the people I know who frequently play PC games want anything to do with Steam.
I doubt these are significant sampling biases. I see no reason to believe that the group of people who use Steam and respond to surveys have substantially different system specs than the general gaming population.
Anyways, as has already been mentioned the consoles are what form the upper bounds of graphical support. If DX10 really was enough of a breakthrough to be worth it, we'd be seeing console gamers converting en-mass to the superior PC market. This isn't happening; the consoles are "good enough" for most gamers, and developers are happy to oblige with graphcis that are "good enough".
