Perfect occasion to point to the absolutely wonderful article "Mother Earth Mother Board" in which 'the hacker tourist ventures forth across the wide and wondrous meatspace of three continents, chronicling the laying of the longest wire on Earth' - Neal Stephenson.
My partner used to work there and did research in their archives as part of her PhD research. It's well worth a trip since you have the astonishing Minack Theatre [1] and a very beautiful beach within walking distance [2].
It's very impractical to put anything in Antarctica. The continent is very far from any place where substantial numbers of people live. The northern hemisphere has much more habitation in the arctic and sub-arctic areas. Moreover, the northern hemisphere in general is wealthier and more populous than southern, so it makes sense to put data centers to the North.
Tierra del Fuego, the southernmost part of South America, lies at 55°S. There's nobody living further to the south than that. Very few people live even at 50°S.
55°N lies Copenhagen, Newcastle upon Tyne, Moscow, Kazan and Chelyabinsk. Cities like Glasgow, Oslo, Gothenburg, Stockholm, Kaliningrad, Helsinki, Tallinn and St. Petersburg are further to the north. Population gets more sparse north of 60°N (I think about 20-25 million people live north of that, a quarter of them Finns) but even when you are looking for cool surroundings, it makes sense to put data centers in places where you have people and infrastructure like stable energy and connectivity.
Ars Technica did a good article on this very thing. While the cooling is nice in some aspects, down there it's a bit too much, as components become brittle and easily damaged at temperatures that low. Static electricity is a larger concern with air that dry. (moisture is frozen out, and Antarctica is literally a desert.) Not the place to simply set your PC outside and go for the overclocking record.
Are you referring to the Gray Cable with all the landing points in Northern Canada?
Firstly, this is not the North Pole. There is no land at the North Pole, which is why it doesn't appear on the map.
What you're looking at is infact the Ivaluk Network, designed to connect up remote Arctic communities (and I'm sure some Oil and Gas interests).
The other cable system going to the Arctic Circle is the Svalbard network connecting mainland Norway with this remote island. Why? Well it's the location of a ground station for Low Earth Orbit satellites (including imaging/spy satellites).
I skimmed through it, looks like that it's an upcoming cable network to diminish the latency on Asia-Europe connections. Unfeasible before global warming and melting icecaps.
Or actually aren't there giant icebergs that break off and float away each year? Maybe there's way to have that pull a cable wrapped around around a very geared up generator. .. not sure how to explain this idea ...
Supplying any hardware to Antarctica is hugely expensive and makes no sense whatsoever, except for the purpose of exploring Antarctica itself. Maintaining a submarine cable there would also be extremely expensive. And you just don't need to go to such extreme conditions to cool down a data center; it's cool enough in Northern Europe. And you can even utilize the heat (here, we use it to warm up houses).
In other words, it's likely that the cables are laid all within a few feet of each other in some places, or bundled together. To show this as it appears would make the map look confusing and would not convey what it wants to show.
This is more of an infographic, or a network diagram.
Perhaps in Russia or China.
In the UK and Europe there is a big effort to prevent accidental 'sabotage', usually from fishermen.
KIS-ORCA[1] project has a very detailed GIS map for undersea infrastructure.
I'm curious why none of the cables go through the Gulf of Mexico into Texas (or somewhere around there)? There are many datacenters around Texas, I'd think it would make sense to have some cables going there to speed things up, or maybe I'm wrong?
seems like the design minimizes the runs that are underwater. The gulf is VERY deep at the center, and the distance saved is probably negligible compared to the easier to maintain overland routes. Just my guess though.
What costs less, laying and operating cable underwater or over mountains? That is, what are the relative costs of land and submarine cables for long distances?
Looking at the runs that follow the coasts of South America, SE Asia, and northern Canada, I wonder why some of them weren't run over land. (Obviously very many cables do run over land; this map only shows submarine cables.)
In my imagination, it seems easier to drop a cable from big spool on a ship than to run it over mountains, for example, but I really have no idea.
Is there a way to sort this by capacity, completion date, and other metrics? I'd love to see a timeline view of how submarine cable capacity has changed over the last decade.
Uses Google data to show the locations, along with the lat/long (using the textual lat longs should be fine from a copyright/data license perspective, though I think lots of OSM people would hem and haw about it being in the same document as Google data).
A lot of the filings are ownership updates to existing cables and don't contain any information about the locations of the cables.
http://archive.wired.com/wired/archive/4.12/ffglass.html
(Quite a long read though: over 40k words.)