That FB/Meta is buying into cable systems does not surprise me, considering how the submarine cable industry is structured and the massive scale of FB/Meta. Cables have massive massive capex to build and relatively tiny opex. The cost structure is completely front-loaded and if you want to be paying the lowest cost prices and you have a large enough requirement, the best way to do this is to be part-owner on cable systems (or an IRU https://en.wikipedia.org/wiki/Indefeasible_rights_of_use).
Edit a bit because I see this line here: partial ownership over 13% of the world's total length of backhaul infrastructure
Basically they are adding up the length of all the cables where FB/Meta have any ownership interest (or perhaps just IRUs) and dividing by the total number of submarine cable length. This is wrong for a number of reasons. These cables are shared in a number of ways. Firstly, each cable contains multiple fiber pairs, in some of these FB/Meta is using only 1 of 6 pairs. Secondly, each fiber is further subdivided by extremely precise DWDM technology (https://en.wikipedia.org/wiki/Wavelength-division_multiplexi...) which allows multiple signals to travel along the same fiber. Finally (and less commonly) fibers are divided into segments. This is more common in SE Asia and Africa where a single cable is really a system which may land at 10++ locations, each segment can be a unique route, or sometimes some fibers in the system bypass some locations, sometimes some fiber pairs completely bypass some landing points. Sometimes just some optical channels within a single fiber are diverted to a landing point! The total system design is incredibly complicated and reducing it to this 13% figure is poor. It is completely plausible that on some of these systems that FB/Meta are using <1% of the design capacity.