He still can't do math. Or pretends not to. The point is that at the end of his initial 5-7 year interval the competition will be even MORE intense. Especially for his category of "credible" science.
What he seems entirely clueless about is that the ability to gain research support early/easily does more to distinguish the PI's apparent ability based on track record than any other factor. With grant support, most scientists who have made it to the Assistant Professor level do a pretty decent job. So the assessment after the first 5-7 years will still be stochastic. The competition to continue will be intense.
His only solution is for population control* but you will notice he doesn't say much about how these now bodaciously funded Assistant Professors setting out to compete with their peers will do so without graduate students or postdocs. Back in the real world, this will just amp up the trainee arms race.
Germain is still not a serious thinker on this. Possibly because, as he admits in the preamble, he only ever had to win a single R01 in his entire career and otherwise spent it fully funded without having to lift much of a finger.
*Why, you might ask yourself, if this is the true problem does he not start here? With calls for all graduate training programs to immediately cut their enrollments by 50% or more?