Often times I am faced with the situation where a new client comes to me with an application that has literally 100s of unnecessary features and it is quite clear that things need to be drastically simplified for the project to have any chance of succeeding. How do you convince the client to take a more MVP approach and simplify?

edit:

So the current top answer is to provide the client with a time/cost estimate for the huge application. I'm not too fond of this answer because it doesn't address the real problem with this situation. And that is - it's a bad practice to spec out a massive application and then try and build it from the get go. I feel much more comfortable initially building a small, simple MVP foundation. And then adding small features to that foundation one by one. So how do I convince the client to approach building software in this way?