The way it seems to work is that CloudFront caches files at their edge servers, and then serves them using the CF pricing- The content is loaded FROM S3, howeve

The way it seems to work is that CloudFront caches files at their edge servers, and then serves them using the CF pricing- The content is loaded FROM S3, however. Amazon then requests it from S3, so it can cache it. This means you need to pay for that download from S3, just like you would if a user downloaded it from home. After this cached copy is stored, Amazon won't need to download it from S3 anymore, so you shouldn't get hit with that again, unless it expires from their servers somehow.