The observation is not "famous people did things"
It is structural, and it is James L. Pelkey's rather than mine: his history of computer communications treats the field as the behaviour of an entrepreneurial population rather than as a sequence of firms. The people recur; the companies are what they were doing at the time.
Once you look for it the pattern is hard to miss. Ralph Ungermann left Intel to co-found Zilog, then left Zilog to found Ungermann-Bass — the first company whose entire business was networking. Robert Metcalfe left , where was invented, to found 3Com. The Estrins appear repeatedly across the period in different combinations.
What the pattern actually explains
Why clusters of companies appear together. A market window opening is a necessary condition; a population of people who have already done it once and can raise money on that record is what turns the opening into a dozen firms within three years.
Why the same architectural instincts show up in competitors. People carry their priors. An engineer who spent five years watching one approach fail does not arrive at the next company neutral, and rival products from firms founded by the same cohort often share assumptions that look, decades later, like a house style nobody chose.
Why acquisition is not the ending it appears to be. Being bought is what success looked like for most firms in this period. It also released the founders, with money and reputation, into the next window — which is why the acquirer's name on a timeline is frequently the beginning of somebody else's second company.
What it does not explain, and the trap in it
It does not explain who won. The same people founded firms that succeeded and firms that did not, sometimes in the same decade. A population theory tells you where the energy goes; it does not tell you which bet pays.
And the reading has a failure mode worth naming: a field described through its recurring founders becomes a field where only the recurring founders are visible. The engineers who stayed, who made one company work for fifteen years and never founded anything, do not appear in a story shaped this way — and much of what got built was built by them.
That is the same criticism this site's Practice corpus makes of the industry generally: the record over-selects for the people who moved, because moving generates events and staying does not.
Why it matters to a reader now
The population is different and the shape is not. Whenever a cluster of companies appears in a short window — virtualisation, then software-defined networking, then cloud security, then whatever is next — the founders are substantially people who were inside the previous wave and left.
That is a useful thing to know when a technology arrives claiming to be unprecedented. It rarely is. It is usually the same population, applying what the last window taught them, to a problem the last window created.