I think it's very interesting how prefab construction seems to pop up from time to time in the media. I grew up in a prefab house, trucked to the property in two halves and bolted together on a poured concrete foundation -- it's incredibly common. Apartment blocks however, they seem to be locked in a perpetual demonstrator state. One, maybe two prefab buildings will go up, and then no more, eliminating the point of t…
I doubt the logistics would make the task of prefabrication not worth doing, though.
The analogy in software that first came to mind was server vs. client-side rendering of websites. (I think much of it holds true for physical construction, too.) This topic seem to reappear every few years (2010 -> everything on the server; 2014 -> everything on the client; 2018 -> everything BACK on the server.) It will likely cycle for at least an iteration more (especially if someone finds a way to advance CPU's vertically, once again), but I believe in future, the consensus will settle to: if you can afford it -> do it on the server. (maybe it already has?)
In the web world, the "builder" would be paying a larger bill, if they would "prefabricate" what they serve to the world, but customers will be happier to get their product/service faster. On the other side - the builder will be able to get to a larger audience with smaller investment into infrastructure, if much of that computation is off-loaded to the clients.
But if we look at the overall cost of those computations - the servers should execute the same amount of computation cheaper (being a dedicated infrastructure for this specific purpose), (generally) it should be in everyone's best interest that computation happens on the server, because overall it costs less for everybody.
Which means that all that it comes down to, would be the question - how critical is this decision, in growing the service/product's customer/user-base and does the RoI make sense (i.e. would that company make more profit getting larger portion of the computation along with a larger userbase, or does it only make sense with lesser server bill)?
So while it would be cheaper, logistically, for facebook to serve their content in a completely raw format (to be fully rendered on the client), would they've still been able to grow their product to what it is right now, if they never made the switch. (I'm speculating about facebook, but I thought it was a safe bet that they do much of the rendering on the server, without confirming that first elsewhere)