As for texts, as far as storage and all that, there's probably something to some additional infrastructure there. But as far as sending and receiving them, they were basically part of the same packets that were already being sent to and from phones, so not much in the way of additional bandwidth was really being involved.
Yes, the characters are indeed eventually embedded inside a cell-to-phone packet that exists during a call or call setup. However, that oft-repeated explanation is far too simplistic, as that is the very smallest part of its journey. It's like thinking that the only cost to sending a letter cross country is the carriage from your postman's mail truck up your walkway to your mailbox.
First off, a text message must use carrier resources to find and be sent to a particular phone somewhere in the entire world. It can easily trigger cross carrier fees.
The text arrives as a call page (aka a ring for voice calls or a pager message). This has used up extra resources already, as a page is always sent to multiple towers in the last radio cell network you connected with.
When the target phone receives the page, the message is still not there yet. The phone must first authenticate itself to the network and get its own control channel. Then it can finally receive the control packet with the SMS stuck in it. All of this is 90% of a full voice call, which is why texts cost similar to a one minute voice call.
But wait. Not done yet. Then the phone has to acknowledge receiving the text before giving up the channel. If the sender has requested a receipt, this whole process had to now be duplicated in reverse back to the sender.
If the recipient is offline, the text must be stored and tried later. If the recipient is an email address, the text must be resent over the internet, and vice versa. If it's an MMS message, there's even more overhead. This all requires carrier resources and computing centers.
And this is all repeated for each text!
In fact, texts use up enough resources that, before safeguards were put in place, you could cause a call denial of service attack in an area simply by sending a bunch of texts to phones targeted in that area.
During busy hours, people on one US carrier can be sending more than a
million texts a _minute_, with the target phones having to be found all across the country or further. Because they have to use carrier control channels to locate the recipients, that's the control load equivalent of initiating 2+ billion phone calls a day, with all the associated extra control server costs.
Said text servers have to be distributed safely in multiple locations around the country, with associated building, physical and logical maintenance, electricity, cooling, security and backup power costs. In short, texts aren't a free ride at all.
Whew, that was hard to do on an iPad!
