Well ... UNIX back in the old days was something of a moonlight or stealth project. Then they got money for an original PDP-11 (later the 11/20) for a specific project or task, serious text processing as I recall, and that lead to the money for their PDP-11/45 (significant since its split I and D allowed for 64KB of code, 56KB of data and 8KB of stack (the latter split due to the 45's MMU), significant because that allowed much larger programs to be written (as I recall the 11/20's architecture left the last 8KB for devices)).
That's why nroff (from a Multics program, as I recall) and troff were such a big thing back then, troff would output to a professional phototypesetter. And they had upstream piped software that would DTRT with e.g. typesetting math. And this was practical, all parts of AT&T had a lot of manuals and papers to produce.
At some later point, AT&T/Bell Labs realized they had something seriously useful here, and the research people got funding for OS work while Western Electric forked UNIX for AT&T production work. (You can find this all in the appropriate histories, I know it from them and from starting with V6 in the summer of 1978 and writing my final project in nroff for a XEROX Daisywheel printer.)
Anyway, my point here, going back to the original point, was that Bell Labs per se was much more of a pure research lab, and it was the job of Western Electric to turn some of their output into stuff they'd use in the system.
And Bell Labs was a tacit part of the monopoly agreement that gave AT&T all the telephony business in the US that they wanted, it was "a price of doing business". That's why they couldn't sell research UNIX (Vx) for real money, things like the transistor weren't locked down with evil patent terms, etc. etc.
In other words, not a situation we expect to see repeated after the '80s breakup.
Right now Go is in an alpha stage (significant parts are missing and known to be needed, code you write will get broken) ... we should perhaps judge it more by guessing whether Google will eventually use it if and when it's of production quality. In the meanwhile, we can look at them as a fairly unique company that really understands that technology is a competitive advantage and that having some of that be open source is good for them.
See Joel S. on how companies desire to commoditize their supporting technologies. Google's open source browser and mobile phone and netbook OS projects are examples them doing very directed projects that support their real business. Indirectly they do things like hire Mr. Python.
