Suppose somebody does a brilliant Master's thesis and afterwards quickly writes two more good papers. He/She could get a PhD immediately at that point as far as research is concerned, but would benefit from waiting a bit longer and also getting mentorship etc. .
A person who worked and published for 10 years for all purposes already was doing PhD work for 10 years and thus can more or less staple some works together to get PhD whenever they want to.
It's possible to complete a PhD in 2-3 years, but this also typically leaves no room for exploration or any kinds of errors. At CMU, the only examples I know of were people who were on leave from the military and had a hard stop after 3 years to return.
My point is, and I'm four beers in, is that my head of data science is so good, he has been mentored by me for years (fwiw), and he has published a lot -- I feel he is the perfect candidate for completing a PhD asap. I think he needs two years for acculturation.
In that paradigm, smart folk in industry could publish papers and then take 2 years to get a doctorate --- and I think the world would be better for it.