I think the difference here is that Amazon is taking in massive revenue and investing it in future infrastructure. Jason Fried advises against taking no revenue now while expecting to "figure out" some way to make some revenue in the future. Both have little or profit now, but Amazon can reasonably expect to make profit in the future, while that SuperCoolFreeApp can't.

>I'm not buying the "we'll grow really fast and then profit later" - didn't we learn about how that worked out already with the dotbomb?

Actually, Amazon did spectacularly well during the internet bubble with just that strategy. This is a return to the company's roots, in a way.