Continuing a trend for this show, there was much discussion about whether garage gamers could emerge once again on the scene. The general concensus is "no, but we sure hope they do." Basically, the folks in the industry have gotten to the point where they feel it has no reached a point of maturity where the barriers to entry (market and content, but not necessarilly technology) preclude the ability for a few people in a garage to come out with the next great title.
It remains to be seen if they are right, with the continuing stabilization of APIs (DirectX, OpenGL) and the growth of the middleware market (RenderMan, NDL's NetImmerse and Havok's physics products), MMORPG hosting systems such as Australia's BigWorld, and other licensable tools, there are opportunities to get a leg up on the technology without spending a fortune in time or money.
However, the requirements for quality production values are the most likely aspect to stymie the game industry. As games have approached "movie quality", the need to create horrificly expensive art has become a tacit requirement.
Some (like John Carmack) predict that we'll move beyond photorealistic to non- photorealistic art, which may help the situation. Others think that we're stuck in this rut of trying to simulate nearly everything.
I think that there's still room (especially on hand-helds and lower resolution devices) for games that have a simple kind of play and a much lower graphics budget.
However, as with game engines, I think we'll start to see less expensive content development going on, especially in the 3D modeling space, as more of the technology moves into the lesser-developed (and less-costly) parts of the world. The Internet is our opportunity for some of these games to be put together on smaller budgets with similar quality concerns.