EOS's best bet is to capitalize on the parallel processing side of things. Vanilla Ethereum is designed more to be a basic foundation for Dapps (Golem/iExec) whereas EOS will be able to output a lot of compute power without the use of a Dapp like Golem. The challenge with all of this is successfully applying it to the consensus layer which I think EOS can do with enough RnD, testing, and time.
Ethereum isn't interested in attempting all of this on the consensus layer like EOS and instead relies on Dapps to handle those types of features. If EOS can pull it off it will be a valuable chain indeed.
RE: EOS Could DOMINATE: Don't Underestimate It's Future Potential! (Thoughts & Opinions)