Most engineering classes include projects, but they're presented as homework and/or "something you can show an employer."
And "building things" isn't enough. Many projects have zero commercial potential.
I don't think PG is wrong, but I also don't think he's right. I think if universities wanted to do something with startup culture they could filter the more motivated and talented students into incubators of their own - not through some kind of overly academic post-grad degree, but literally a startup school where the university provides resources, subsistence, and access to funding, in return for a share of project equity. The "teaching" would be more about finding customers and handling finances and less about managerial MBA nonsense.
The big win for students would be "You don't have to worry about food, rent, or basic running costs."
Post-grad research overlaps with this, but is more fundamental. The emphasis here would be more on projects with immediate income potential, not on technology that could be commercialised in the future, possibly.
The problem is academic culture is very different to startup culture, so the startup school would need a much looser relationship to its parent than a typical academic department.