I work for a non-profit research group that is primarily supported by DARPA project funding. It's an interesting scenario, especially for a committed social liberal such as myself: my colleagues work very, very hard to find funding from DARPA, the National Science Foundation, and even the NSA, for their basic computer science research. The advantage is that, in the better projects, they are more or less left free to innovate, regardless of immediate commercial applications of their work. The downside, of course, is that we deal with the complications of trying to justify pure research in a paranoid, defense-focused environment.
My position is somewhat unique: I work primarily on website administration, building releases of our research tools, and general support and lightweight automation. Translated into English, rather than marketron, that means I'm a code grunt. I take care of all the stuff that's not worth writing papers about; I'm also one of the few technical staff to not hold at least a master's degree, and the only person at my office under the age of 25 who doesn't answer the phones.
I've helped to submit proposals, gone to DARPA-organized conferences, and sat in on countless planning meetings, and my current feeling is that DARPA is a titan looking for a cause. 9/11 impacted the short-term goals and outlook of project managers at DARPA to such an extent that entire research initiatives have been re-tasked with finding solutions to problems of national security and terrorism, instead of the core technology they were developing previously.
Of course, there's an ample supply of big-business capitalism, even in the government-funded research environment. Most major DARPA projects there days have one or more "OEP" (Original Experimentation Platform) providers -- big defense contractors like Boeing or Lockheed who basically farm out the hard problems of their pre-existing, nine-figure defense contractors to academic and non-profit research groups, who in exchange receive support for continued funds from DARPA and other government programs.
Really, it's like any other kind of work within a big system: there are those who learn to go where the money is, those who can survive on genius alone, and the lucky few who end up in the right place at the right time.