Demonstrating skill through building a product is also hard, and also helps build value.
I think of the math like this: are the next 1000 engineers at Facebook worth 1% of their market cap? Yes, the math works.
In the kind of world where the government can't get the federal Obamacare exchange working, and people pretend that Waterfall is a competitor to Agile while really, the most common alternative is no process at all. Most companies are highly effective at demotivating software developers, and if you get hired through the usual process you'll start out on the bottom of the pyramid and behind the 8 ball.
If it costs them $1M+ to hire you there's just a chance they'll behave as if your time is worthwhile.
I like your last sentence.
There is another side here too: companies often overestimate their ability to manage acquisitions. A price floor means that those more likely to be poor integrations will be filtered out.
It might be easier to just "take a job" (debatable) but it is probably not as lucrative. As an engineer, your choice is:
1. Get aqua-hired for $1MM right away (well, maybe after a few years of working in the start-up) plus whatever your new salary will be vs.
2. Slowly earn that $1MM spread over the course of 10 years or so as a regular employee.
I know which one I'd rather do.
Why not just skip all pretenses of "doing the startup" and "developing the product" and go right for the aqui-hire. Get a well-gelled team together (demonstrate the well-gelled-ness with a demo of some sort) and then shop the whole team around to a big company looking to acquire talent. If they're willing to pay $1MM a head for a team working on a start-up and then shut the start-up's project down, why wouldn't they simply pay $1MM a head for the team?