I'm betting on a massive improvement on the MMORPG genre in general and korean MMORPGs in particular. Those run badly whatever hardware you have and Mantle should alleviate the need of raw power. Too bad almost all of them rock an Unreal Engine and Mantle is likely to get there rather late.
Yes PC RTS and MMOs will benefit the most, problem is GCN market share is just a fraction of the overall market.
So if you build an MMO around Mantle, you have a very tiny user base to extract money from. If you build it around DX9 (a joke, but that is where we are with DX9 hardware holding everything back), obtain poor performance, then also offer the same version, but with Mantle support and thus more performance what benefit is that to the game studio?
Mantle does what we need, but what we need is Mantle for all hardware, not just one fraction of the market. But that goes against what Mantle is, which is a hardware specific API. The reason it works so well for GCN is because it doesn't work for Intel or Nvidia, or 6xxx series or later. It is meant only for but a fraction of an overall market, what kind of penetration can we actually expect with such a grossly limited scope of market this benefits?