That's a bit like trying to use Apple ][ machines for cloud business.

Although your simile is reductio ad absurdum, a version of this idea does, occasionally get surfaced, in the form of using (modern) small, lower-power processors (e.g. Atom, mobile ARM) in very large numbers in the datacenter.

While touting the purchase cost or energy benefits, these ideas routinely ignore the overhead cost inherent in a distributed system, let alone the Fallacies [1], which is the GP's and OC's (and possibly your) point, I believe.

[1] https://en.wikipedia.org/wiki/Fallacies_of_distributed_compu...