So instead, release a steambox version which is only going to be popular with "enthusiasts", bug/beta test the hell out of them and then more aggressively market steamboxes in 2-5 years time when the current console generation is starting to look old.
Valve are playing a long game here I think, the value proposition of a steambox isn't going to make sense until we're about half way through the current console cycle.
Companies like Microsoft, Nintendo and Sony make the really big bucks from the publishing fees of the games, the hardware is sold basically at cost (or at loss if we consider production, design and software costs).
Third party Steam Machines sellers can save on software, but they need to have some margin on the hardware itself, unless Valve cuts them a percentage on the actual sales.
I don´t think price will ever be the selling point of Steam Machines.
Unless you are buying a PC for a company, buyers fall into two categories today. Those that buy a $300 "Office PC" from the shelf, and those that build their own. High-end preassembled PCs are basically gone (save for "all-in-ones").
Since the PC market is generally shrinking (or stagnating), it makes a lot of sense to join forces and collect the big gaming PC assemblers under a united label. It's a little bit like people don't want a Sony or an Acer or an Asus anymore, they want an "Ultrabook".