AT&T often gives $250 - $380 off smartphones, which fits with the $325 figure. So that's not unusually high at all.
I'm not sure why an analyst would think that Apple would get an extra $100 for signing up in an Apple store. Not believable.
Why would it be not believable when this is a renegotiation of a contract which gave Apple a monthly cut of subscription revenue - an exception to the industry? Was that not unbelievable, but true?
And when you're checking the phone prices, don't count the $180 refurb discounts or manufacturer rebates as AT&T subsidies, they're not. The maximum subsidy I saw was $250 with the average around $175. Apple's still getting paid for exclusivity.
AT&T doesn't care who/what they sell, as long as it gets old/new customers. The exclusivity only works one way.
My difficulty here is understanding why people think that Apple, the company that (over)charges for everything including their new cloud computing initiative MobileMe which can be pieced together for free elsewhere, would say "let's sign a five year contract that screws us."
More subscribers means more need for voicemail servers, backhaul bandwidth, and so forth. So no, the cost isn't anywhere near to being constant.
As I've said, I work for a Fortune 500 full service broadband provider (telephone, internet, tv, voicemail, email, etc.). I didn't say the cost was constant. I said, "The operational cost remains relatively the same whether there are 10 data subscribers or 10,000 data subscribers." The infrastructure cost for 10 users is about the same as for 10,000. So reaching your 10,001st subscriber, you may as well reach your 20,000th subscriber. Reaching your 20,001st subscriber, you may as well reach your 30,000th subscriber. Actually, the capacity is much higher than 10,000 per cost layout. And the profits increase exponentially the higher you go since the majority of wireless cost is building and maintaining external sites, not the back end where storage and processing is comparatively inexpensive.