it SHOULD handle a GPU + CPU any more than that tho would become problematic or so I would be willing to guess.
They like to show off in the NCIX video 2x GPUs + CPU sadly they ran 7970 cards but did so in window mode so only 1 GPU was being used
That said The H220 tends from other reviews to out perform all the other AIO kits or rival them, but at lower noise levels. For the most part most of the cooling seems to be about the same, but as you noticed with the H220 you can add to it.
Examples
As can be seen the cooler does well out of box it equals or rivals the other AIO kits but is quieter,
Add better fans its cooling performance improves of course.
From what most reviews show the H220 is the best choice overall of any of the kits. adding more fans drops temps further.
Still the H220 does perform better than the H100i
but what most people forget is the H100i's fans spin at 2600+ RPMs at max speed vs 1600-1700 in the H220, yet the H220 cools better and at about 10-12 dBA quieter. I reserve my full judgement untill I get one for testing.
Right now noise to cooling performance the H110 is hard to beat, sure an H100i might do it but 47 dBA vs 62 dBA in my own testing the H100 i expect to be fairly close noise wise to the H100i has the H110 being 15 dBA quieter. with fans that spin at 1500 rpm vs 2600 rpm. With dense radiators static pressure can improve performance 140mm fans tend to have less static pressure but more air flow. on a Dense aka 20 fpi rad the 140mm fans having less static pressure does come into play how much? not a whole lot. if you could crank the 140mm fans to higher speeds with the same sound level the H110 would outperform it. There are pros and cons to all these kits.
Hopefully I will get an H220 in for testing and I can give you a better answer currently I can only look and see what others have found performance wise but in general it seems the H220 is the better quieter kit but my testing methods may differ slightly so I can't even really compare noise levels across reviews either as my data is obtained with different tools or from different distances than other reviewers. So again I wont know unless I get one for testing.