I went on to Telegeography's site to find something worthy of noting about Internet Traffic and discovered that we are not yet done building the network of the future.  In effect where traffic is local is dropping in price as density creates competition, while where the backhaul is long the prices remain stable and in some cases monopolistic.

While in the major cities that have been hubs for traffics forever, such as NYC, LA, London, etc. A great majority of the ROW is still sending traffic to major centers that should remain local.

While it is great to talk about the cloud if the cloud ends up in a focal point like NAP of the Americas or some other IP Transit hub, the reality is that the Internet starts to loose it's redundancy.

We are already faced with service providers that have a tendency to want to "own" the traffic and keep on net what should be offloaded.

The funny thing is that wireless operators have come to begrudgingly accepted this fact and use Wi-Fi to maintain performance.

For IoT the use of MVNO's may be a beneficiary of these cost anomalies.  If you think about Kore's acquisition of Wyless and the specific mention of Sao Paulo, Brazil. This is a case in point where IP Transit on traditional systems is costly and it may be that the Kore/Wyless infrastructure provides a local offload.