I don't think this is true, I went to a very small not well known college in Canada and a classmate of mine went straight to Google after finishing his CS degree.
Good companies generally hire off of provable skills not "how early" you started learning.
1) Majority of employees at Google FB etc are graduates of top tech schools.
2) Majority of employees started coding and had interest in CS before college.
How likely you are to be good at CS if you never had an interest in CS before college :)? I have a lot of friends from Easternn Europe now working at Google, Apple, FB, but this anecdotal evidence does not change the fact that majority of SEs there are graduates of top tech schools.
It might be true that the majority are coming out of top tech schools, I really have no way of knowing that. However, I'm skeptical about how important the school is to Google as an employer given their rigorous screening process. Coming from a top tech school certainly can't hurt but I don't believe the school you go to necessarily dictates what quality of engineer you are, though it can increase your opportunities for sure.
On the second point, I think interest prior to University is certainly helpful but saying it's a majority is just an anecdotal conjecture itself.
Do you happen to have any evidence for #2? I work at one of those companies and picked up coding in college, and so did at least a few of my coworkers. I really have no idea how common it is and would love to know.
#2 I have no evidence that time spent learning and practicing a particular discipline influences skill level, but I am guessing that there is a correlation :). For majority of top schools aside from Barkley it is not a prerequisite to enter into CS program. From people I personally know which is not a representative sample I'd say 50/50 but it skewed by good number of them having Stats./Math as original major and then CS Masters or Ph.D.
Good companies generally hire off of provable skills not "how early" you started learning.