Lasers work really well in space for secure sat-to-sat data links, but are a lot less viable on Earth’s surface due to diffraction and weather, nevermind the limits of the visible horizon for any height of a communications tower. For pretty much any scenario where laser comms would be considered, microwave RF links would likely be just as good, cheaper, and more commonly deployed and understood by telecom engineers. The only exception is when absurdly high bandwidths are needed, which is where lasers rule.

But using RF links across thousands of kilometers of oceanic waters? For that, you must construct additional pylons on floating islands to repeat the signal. Otherwise, the only RF signals that could reach land would be too low frequency to carry much bandwidth.

For reference, when the German Aerospace Center (DLR) set the world record in 2016 for free-space optical communications, they achieved 1.72 Tbits/sec over a distance of 10.45 km. Most optical systems observe a bandwidth/distance relationship, where at best, shooting the signal farther means less available bandwidth, or more bandwidth if brought closer. This is a related to the Shannon-Hartley theorem, since the limiting factor is optical noise.

So if 1.72 Tbits/sec at 10 km is the best they achieved in free air in 2016, then that pales in comparison to the undersea fibre cables of 2006, where a section of the SHEFA-2 Scottish-Faroese cable runs unamplified for 390 km and moves 570 Gbits/sec aggregate.

In short, free-space lasers are fast and long-distance. But lasers within fibre cables are much faster and cover even longer distances.