quoted from that site:

"Cisco estimates that total internet traffic currently averages 167 terabits per second. FedEx has a fleet of 654 aircraft with a lift capacity of 26.5 million pounds daily. A solid-state laptop drive weighs about 78 grams and can hold up to a terabyte.

"That means FedEx is capable of transferring 150 exabytes of data per day, or 14 petabits per second—almost a hundred times the current throughput of the internet."

i imagine most with mathematical instinct would think something 'real-world' is missing in this approximation, in that you could not realistically use-up all FedEx resources for such a data sharing project without likely taking down the system or being denied access- and thus existing traffic and congestion are not included in this ideal naive model.

what would actually occur if you tried to use all FedEx shipping resources on a single day, and then after considering that, doing so repeatedly day after day. not only is I/O data transfer omitted (time needed to access/store/exchange data between platforms), though also hardware failures which, perhaps i am wrong- packet technology succeeds in transferring via multiple attempts and thus the 'internet traffic' of 167 terabits could potentially includes delivery failures and successful resends, and routing around congestion- whereas it is completely unrealistic to assume you could decide to ship such material and use up all "network" resources of FedEx without considering its tolerances for additional bandwidth to cover this parallelism, and also limits of local delivery or to various locations in an overnight scenario-- so delays would likely be involved, and if any data on laptop drives were to fail it would seemingly require reshipment to compare to the internet data transmission approach. or not.

it is not to lose grasp of the notion, only to consider it in its depth, and knowing what may feasibly and more actually occur in such a scenario could provide a more accurate understanding of the limits of analogies without corresponding matched dimensions. such that maybe it would not be as efficient or as easy or even possible, as believed. though the simple mathematical comparison makes it appear so. and most comparisons of this nature often have similar 'modeling errors' where approximations function in an ideological realm as a result, though likewise makes for interesting considerations in considering where the gap is between what is said and what exists as a situation.