Isn't it still a pretty slow process? Efficiency doesn't matter if it generates a trickle of H, and you need a large fixed volume.
Splitting at point of use is silly anyway- why use power to split water into hydrogen you can use for a fuel cell to generate power, at the point of use?