Wildly underestimates the cost.
Linux inherited a well-tested basic design (though no actual code) from Unix in 1991 and has accumulated 17 years worth of real world experience since then. X and the GNU tools have matured for over twenty years (I remember installing the GNU toolset on Sun boxes in 1990 because they were so much better, even then, than what Sun offered at the time with SunOS). There's no way to value that experience.
Plus, Linux vendors choose what they think is the optimal subset of the universe of free software to include in their distributions. Most software (whether Free or commercial) is crap. The fact that the code has been included in a distribution means that it's much better than average.
If you just gave a team of 2000 thousand developers the money and time to produce something like a Linux distribution (as suggested by these models) 1) the code wouldn't have the decades to mature and learn from mistakes, 2) it would be of merely average quality (mostly crap, some good stuff) and not comparable to a distribution that only chooses the best software to include.
To make it more concrete, there's a huge selection of Free window managers to choose from (http://xwinman.org/). They all work (they do what a window manager needs to do). A Linux distribution gets to choose the best from these pre-existing applications.
If you had to create a window manager from scratch you could give some money to developers and get something that worked. But if you wanted to duplicate what goes into a Linux distribution, you'd have to pay 20(ish) teams to develop window managers and then choose the best.