Maybe it isn't the loot server's processor being too slow? What if it was the loot controlling program code being too "fat"?
While speculating, and especially based on experience (with EU), I'd say it's - in no particular order - an unfortunate combination of database choice, network congestion (!), lack of RAM, too slow disks, crap design, and incompetence.
Disk I/O isn't a problem nowadays for a system handling just a few tens of thousands of individual accounts (at best!). Slap on some mirrored SSD's, even if cheap 128GB, and you get half a gigabyte sustained both read and write (I have data to back this up, from "old" Intel 520 SSD drives). Keep in mind the "back-end" servers are never even told about how much money you have spent until the time comes to loot; the "area server" handles all your ammo-/decay-losses (VU10 proved this without a shadow of a doubt).
CPU? C'mon, if you have CPU issues nowadays for such
Embarrassingly parallel workloads you're doing something wrong.
Network bandwidth? We only know they co-locate (if even that; perhaps they just rent virtual servers) in Amsterdam. We don't know if they have the actual (bigger-)money-handling servers (such as loot-system) in Gothenburg or if it runs on the Port Atlantis server. Heck, we don't even know if they even can spell "10Gbps" yet. No matter, the amount of delay they repeatedly induce suggests a lack of competence.
RAM? Well, they
could theoretically try to run the servers on 512MB 32-bit Windows machines - even if virtual ones - and try to patch them up as we go.
Crap design? While it could explain the horrible downtimes to just roll out an update (that should take at most 30 seconds if properly tested), I can't believe even MA are that moronic (I admit, I had and still have some fears that I could come to regret that statement).
As I see it, the only remaining is and database choice and (potentially) connection Amsterdam<->Gothenburg. Anyone want to continue... Kim?