A myth has made the rounds for several years that open source is only good for commodifying tired old markets. Bob Bickel brings it up on the Ringside Networks blog today, and suggests a few examples of open-source innovation (like, err, the entire Internet :-).
Having been a part of the JBoss success story, I can speak personally about the thousands of applications that would never have used enterprise Java had not JBoss existed. Certainly JBoss did not get paid by 95% of the user base, but the market that was created by JBoss was what supported the eventual financial success of JBoss.
Putting technology aside, unless you're an obese incumbent, why would you ever try to open a new market with proprietary software?
Adoption is a new entrant's biggest challenge, not intellectual-property protection. Open source is an effective means to cast a wide net on distribution.
On the technology, too, it's true that a new market requires tight cohesion between a small core group of developers as they iterate toward a solution, but this is descriptive of open-source communities, too. There's simply no reason that open source can't innovate new markets and, indeed, everything to suggest that it can, and may well be the best means to do so.