An alternative hypothesis that fits the data equally well is this:
As startups become more popular, the low hanging fruit is quickly picked over (social networks, mobile apps that require very little initial capital, etc.). Concurrently, startups have become more mainstream and seen as less risky (whether justified or not). As such, many "older" (by which I mean not in their 20s) people with narrow domain expertise have begun to start companies with the goal of addressing some narrow need that isn't addressable by the typical smart, CS-grad, startup founder with little to no real world experience outside of tech internships.