we should all take a deep breath and remember that any single one of us is simply talking for the most part. We have very few precious facts in regards to Star Citizens development despite its claim of being transparent.
We do have the evidence at hand, which is the current alpha. Spending some time in it is quite revealing... I try to keep to what i see. (edit) i tend to believe what i see is what we get and nothing more in the case of CiG, because CR is like a toddler and will never shut up if he's got something to show.
one virtual server with server meshing of physical servers
Dunno why they still insist on that wording while everyone else calls that "cloud". It's not rocket science, it's been there for a while, and certainly does not take years to set up.
Also if you are thinking of the nonsense spouted by CR, he's been out of touch with development for more than 20 years. If you know about game programming even just a bit, you should know that everything has to happen inside frame time, if you decide your server runs at 30fps then you have roughly 30ms to do everything and update the game state before you cycle over. Involving networking or i/o in that time span is impossible due to strong physical limitations, like the speed of light in fiber optics. You certainly can colocate your server racks and run a fiber channel and all, but your latency due to network overhead will quickly add up and eat away that precious frame time.
You certainly can implement asynchronous time warp and such, but that's really advanced techniques, and i've never seen it done on any MMO (it's for 1v1 fighting games - i have some ideas on how to expand the principle but the underlying code is really difficult).
So instead you would have to run everything in an async way then try and do data reconciliation when everything goes back and a few frames are gone by, changing the game state so you'll have to match the newer state, and solve all the conflicts that will emerge. Given the track record of CiG so far on various implementations (want me to talk about that barman "AI" ?), I highly doubt they would be able of such a feat. Add to that their engine of choice which has tier 0 networking and a completely synchronous game loop, with spaghetti code that ties everything up to it.
First order would be to call "cloud" what it is. Then to give an actual, realistic, implementation strategy, not CR waving hands and throwing unrelated buzzwords. For example that's what the devs Dual Universe are doing - they gave a few conferences at Devoxx for example where they expose their detailed architecture, with code, and it seemed like a really strong proposal (im not part of the closed alpha on DU, cant judge the results so far). The architecture and code to do all of this is really difficult to set up properly, with many pitfalls, and should have been the VERY FIRST thing to do at the beginning of the project, with just place holder grey boxes, until they got that perfect, because it's the very core of the whole game engine, and impacts absolutely everything.
