The quoted complaint that the CS curriculum didn't assist one with their startup is an odd one. Startups are meant as vehicles for rapid growth with an end goal of a high exit, and usually based on top of some form of end-user application that can be tiered, sold per unit or commissioned to advertisers.

In contrast, doing CS research (such as research operating systems), despite its high technical importance, frequently has no opportunity for direct end user monetization and is totally inappropriate for a startup structure. It can yield concepts that may lead to huge cost-saving benefits once implemented in the wild, but in of itself it has no market value, only possible utility.

In all honesty, I think the communication failure mostly falls on the industry's shoulders, not on academia. The industry has time and time demonstrated it values short-term conceptual complacency and having more of what they know, as opposed to any truly large paradigm shifts (in software, at least).