the data really isn't that large and the algo isn't difficult, most people who do this work don't need to compare proteins across every organism in the world. most people just compare to a few organisms that they work within the context of for clinical trials or data.

if you're comparing DNA and looking at every organism ever you might need such power, but if you do i'd hope you work at some institute that makes your own algo in house because blast isn't the end all be all