What is the forecast sales of Steam Machines, and Witcher 3 on Steam Machine, over the commercial life of the game? It's not a matter of being more fair or open, or even technically superior; it's a matter of can you make this at a profit that justifies the expense, and can you deliver this on a schedule that does not delay revenue or shorten the life of the product?
Solutions you can get and use now is where this competition has been, is being, and will be played out. Nobody who is a newcomer to the console business is fool enough to favor open source that adds months or years to development cycles because the product you need right now has not yet been written, nobody is being paid to write it, and your competitor already bought and integrated it.
Open-sourcing drivers is a Good Thing but really doesn't figure into that calculation. If nVidia wanted to squash AMD's or Intel's potential advantage in open-source drivers, all they would have to do is open-source theirs. But it doesn't change the real engineering problem and the market for solutions, which is this: You write the game to use a toolkit and an API that are at a much higher level and much less effort than writing to use the drivers; by doing this, you get your game to market before your competitors at less cost, and you make more sales over a longer product life.
I really hope AMD gets a big win with Mantle and a lot of adoptions, and a lot of toolkits and APIs come out for it. Because if they don't, this game is just going to get more and more one-sided.