As far as my opinions go, I think that the biggest advantage to a power API might be that energy becomes unified without being made the same, if that makes any sense. Consider a hypothetical Forge hook for registering energy points. Regardless of what energy you would be using, you would be moving X amounts of forge energy. And since forge energy = forge energy, this would mean that all energy could be used anywhere. But does that have to infringe on the modders' creativity? Not at all! Indeed I would hope it would increase the diversity of power systems that we have. Because you see, even if the energy is all the same you can still make your power system the way it fits your design goals. With any luck, what this would result in is energy systems where what produces power and what uses them is irrelevant; how the energy is being transported and routed is important.
I find that even now, with the myriad of different things available, I only ever use two things: IC2 power and TE power. Regardless of what the advantages of other systems might be, without built-in compatibility you would have to make a point of, say, setting up an MFR system even if you basically have no machines to power with it, and I think it is a shame. A fully modded-out Minecraft has so many things to use and do, and I would much prefer to use each of them than not to; after all, I could get much the same experience with less mods, and have less CPU strain to boot. For instance, I drown in lead, but really the only thing I am using it for is making hardened glass. Likewise, if I had Metallurgy installed, I would basically do nothing at all with a lot of the things I pulled out of the ground because the option to use them is not there. In much the same way I see the current state of power systems. Better compatibility could open a whole new range of things for the player to do! I envision a scenario where choosing between leaky machines and distance loss is only a small subset of the options available for every undertaking that requires power, and indeed, instead of having a monolithic power system of one type with maybe a few converters at the ends for different machines you could more or less freely mix and match to fit your needs exactly.
Now I realise that I am blowing fairy bubbles all over the place, but in my opinion KL's current undertaking to make a simple, extensible power system wrapper is a large step towards the remote possibility of maybe someday seeing that happening.