But when you structure your business model such that you don't need to sell millions of consoles to make a return, it doesn't really matter. Like if I make $200 per console and my initial R&D was $250k, all I'd have to sell is 1250 consoles in total to make my investment back. I'll be more conservative at say $1 million initial R&D investment, which can easily pay for like 7-8 full time engineers working a whole year on the project, and you'd still make your money back with selling 5000 consoles.
Why would anyone pay you $1000 for the hardware, +$100 for assembly, +$200 for the retail cut + $200 for tax + $200 for your profit = $1700 for a device that has no exclusive games, and can't be used to do anything other than play games you can already play the games you can already play on your Windows desktop? Why wouldn't you just pay $1000 for a big Windows box and put Steam on it, and you can use the $700 bucks that you have left over to buy yourself a new Xbox and a PS3.
At the moment I'm not convinced that they'll sell 5000 boxes. Consoles are not an easy market to get into, and I think SteamBox doesn't really have any features that anyone wants. It's going to be expensive for the hardware compared with other consoles (because of lack of subsidy), it'll have no exclusive titles, and even if it plays all Steam games, there's a whole load of major titles it'll be missing out on from Windows/PC, Xbox and PS3.