This is one of those times I wish HN wasn't so strict about the title of the submission matching the title of the article--this title is extremely misleading. (As a guest post, I'm guessing the author didn't get to write the title, so I'm not blaming him). 4ormat didn't "drop" IE, they never supported it. As a result, their calculation of how much they "saved" appears to be based on some back-of-the-envelope calculation of how much more expensive it would have been to develop for IE over a period of years.
tl;dr: if you build a tool which solves a significant and recurring pain point for creative professionals, you can get away with not supporting IE.
The article does not provide much insight into building a consumer-facing application--this is the equivalent of a company using an intranet app which is IE6-specific.