Earlier quoted context omitted.
That's the point , though: usage driven fees. It's absolutely not the most economical offering, but the freedom and ability to either quadruple your infrastructure or shut it down entirely, no strings attached, in an instant is what you're paying for.
With most of dedicated server providers now, you can get any server with a 1 month contract, no strings attached, for a fraction of the price of an equivalent ec2 instance, and good IO as a bonus. And good providers usually delivers within 24h. I love aws services and use them myself, but for ec2, but I doubt most users have such volatile needs. The main advantage I see, for aws users (non ec2) is to have everything…
Still not really the point. You're going to have to buy enough servers in advance to handle your peak load every month. And then you're going to have to forecast ahead at the end of the month and cancel all the ones you don't need. It might be theoretically possible but in practice it is ridiculous. EC2 brings the contract interval to 1 hour and lets you control it with API calls.