After all of the talk about leveraging the AMD successes with the game consoles -- and about smoothing ports -- surprised to see nvidia only.
Prototype hardware always costs a gzillion dollars.
Another post mentions this as a way of getting SteamOS out there, but really -- as this is just standard PC hardware -- there are already hundreds of millions of candidate devices.
The only real way I can see SteamOS having a chance is if it offers a console-like value proposition (because the next generation of very powerful consoles are just coming out, and even "hard core" gamers are looking at them longingly), which valve needed to make happen through tight integration, optimization, and mass production. Another go at HTPCs == almost certain commercial failure.
But, back to Steam Machines. I don't think it would be a good idea for Valve to try to get Steam Machines down to console like costs. They're not in a position to sell hardware at a loss like console manufacturers have traditionally done in the beginning of a cycle. The Steam Machines should offer quality.
http://www.forbes.com/sites/davidewalt/2012/03/07/valve-gabe...
Look at the specs we're discussing. Even the low-end version is decidedly non-casual.
Those specs are high end but that also means the device will be compatible with games for the rest of the decade.
At this point if I have to be skeptical about something is about how well the hardware will work with the OS considering the reputation nvidia has with Linux...
The point of SteamOS is that it's open, and anyone can make them (and people other than Valve can sell them). If they become insanely popular, I'm sure that someone will jump on integration to sell a cheaper console.
Maybe (probably) I'm just biased, but I spent a few years trying to get MythTV, Satellite PCs, Silverstone chassis, Mediacenter, XBMC on a modded 1st gen XBox, etc etc working smoothly.
It was a nightmare. I don't ever ever never want to have to think about overscan again as long as I live.
For me to be interested it needs to be flawlessly plug & play.
This was a few years before HDMI output was available on videocards, so maybe that has already solved the problem.
But as a ~8 year Mac user who converted from Windows XP for Ruby I will never, ever buy another Windows PC. Ever. Never. I mean, my wife still has to occasionally open network settings to get her work laptop onto the network. And because it's an Airport Extreme, that never has to be rebooted (unlike the dozen other WAPs I had previously), it's never the network. Just her laptop. That's crazy. No way I'm letting that into my living room.
Pitch me a box and out of the box experience that is guaranteed to work as long as I keep it vanilla (which is all I want) and is otherwise a PS4 I have the ability to upgrade piecemeal... Yeah. I'm totally on board with that.
And if the price of entry is pushing $600+, I'm probably OK with that too. That's what I paid for an 80GB PS3 on release day after all. I don't think I could make the mental leap past $999 though. ;-)
Price isn't my primary concern. Plug & Play access to a top-notch gaming environment is.
Honestly the PS4 may be a better fit, but the ability to swap out circuit boards on a SteamMachine is appealing. And the game library. If I could play DOOM and System Shock on my TV without spending as much time getting it working as playing I'd be pretty stoked.
The plug-and-play thing has never really worked. However not having to wait the normal generational time between consoles would be an advantage -- it was always baffling that each iteration of the 360 and PS3 (there were many of each) couldn't improve it in some way beyond making it smaller or lower power. GTA IV on the 360 had draw distance issues and would stutter mid-town -- it would have been nice if some GPU improvements in the 2nd and then 3rd edition eliminated that, etc, but the nature of that platform makes it so they can't.
nvidia (closed source) > AMD > nvidia (open source)