Microsoft and Facebook to lay Atlantic cable

Please consider donating to Behind the Black, by giving either a one-time contribution or a regular subscription, as outlined in the tip jar to the right. Your support will allow me to continue covering science and culture as I have for the past twenty years, independent and free from any outside influence.

Microsoft and Facebook have announced plans to lay a trans-Atlantic communications cable from Virginia to Spain.

Running from Virginia Beach, Virgina to Bilbao, Spain, MAREA (which is Spanish for “tide”), it will be the first cable to connect the US to southern Europe, over a distance of 6,600 km (4,100 miles). From Blibao, it will connect to network hubs in Europe, Africa, the Middle East and Asia, with the goal of improving speed and reliability. For Microsoft this means improvements for users of its cloud services, such as Bing, Office 365, Skype, Xbox Live, and Microsoft Azure, while for Facebook it means improvements for users of its eponymous social network.

I find this story very puzzling. The whole reason communication satellites exist is because they have historically been far cheaper to build and launch with far greater capacity than ocean cables. Thus, the decision of these companies to go with an undersea cable instead of satellites suggests that something has changed in that equation, though I can’t see what. Have undersea cables improved so much that they have a bigger capacity than satellites, so much bigger that it compensates for the higher cost of installation and maintenance?


  • Wayne

    I have no clue what capacity satellite can handle, but according to the article the cable will carry 160 terabits per second.
    Engineering aside, MS & FB might be looking to avoid having certain segments of their traffic be carried on a regulated-basis. (The difference between a “private” cable & being designated as a Common-Carrier.)
    –This is however, way above my paygrade & I would like to hear some input from anyone who knows more.

  • pzatchok

    It all about control.

    The internet is controlled and charged per KB by the companies who own the backbone telecommunication lines including wireless systems.
    Is it cheaper to install and operate a cable for 50 years or a satellite?

    MS and FB have billions of dollars sitting around doing nothing so why not get into a new business?

  • Matt in AZ

    The data will have a much shorter path through the cables than having to travel far out to one or more satellites before coming back to Earth. Reducing latency this way would definitely be a benefit to Microsoft’s Skype, and their Xbox’s online gaming, which has become more and more international. I’m not sure what Facebook properties would benefit from low latency, but they own so much there has to be at least a few big ones.

  • Cotour

    While I am not an expert I think what is the point of the cable are two things.

    1. Massive capacity.


    2. And this may be primary, its fiber optic and it can not be tapped as I understand fiber optics.

  • Wayne

    Matt in AZ:
    Interesting point about latency/jitter & gaming.
    Interesting point about control. (That’s why I went with the “common-carrier” thought.)
    Above my paygrade as well, but I don’t think “security” is their prime concern, as-in tapping the cable. (But I do not know one way or the other.)
    (Intercepting the data is easier to accomplish on-land & doing so at the major routing/translation-points. Truly seamless, 100% end-to-end fiber-optic, I do not believe, is all that common outside of highly specific applications/customers. (but again, I do not know.)

    Definitely want to hear from someone who knows more.
    –What sort of capacity does a modern satellite handle??

  • Tom Billings

    “(Intercepting the data is easier to accomplish on-land & doing so at the major routing/translation-points. Truly seamless, 100% end-to-end fiber-optic, I do not believe, is all that common outside of highly specific applications/customers. (but again, I do not know.”

    And yet, the greatest intelligence coup of WW1, the Zimmerman Telegram, was from the British Navy’s fishing up a cable carrying German traffic, and then breaking their codes. The fact is that however small it is, a lightbeam in a fiber optic cable is still electromagnetic, and *will* still have an electromagnetic effect outside the fiber, ….it’s just incredibly tiny compared to an electrical cable at longer wavelengths.

  • Cotour

    I did not think it was possible tap an undersea fiber optic cable, but apparently it is possible.

    “l WSJ – May 2001
    “…former intelligence officials confirmed that NSA technicians used a special
    submarine to tap into a fiber-optic cable on the seafloor in the mid-1990s,
    around the same time that fiber amplifiers began displacing electro-optic
    amplifiers. The sub supposedly had a special compartment into which the
    cable could be hauled, enabling technicians to install the tap.

    IEEE – June 2003
    “Further evidence of the NSA’s ability to tap undersea fiber-optic cables – and its
    intention to go on doing it – is a $1B project at Electric Boat in Groton,
    Connecticut, to outfit a new Navy submarine, the USS Jimmy Carter, with a
    special 45-meter-long section. The Navy has never disclosed the exact
    purpose of the expensive addition to the $2.4B sub, but most
    observers…believe it is to tap undersea fiber-optic cables.”

  • Max

    Due to chromatic dispersion, most undersea cable lengths need a relay every 80 to 100 km. Some cables have longer lengths, but they pay for it in signal degradation. Also every large ship can only hold so much cable so splices that attach one cable to the next result in some signal loss. New technologies are constantly upgrading the system to a higher level. Perhaps they have a ship that houses the cable making factory to reduce signal loss and they lay it as they make it? The best part about cables is the resistant to EMP (from a nuclear device or the sun). Cable is tamper resistant and that if anyone moves the cable they know it. (fiber optic cable’s are used to measure land movement)

    Satellites, on the other hand, are also improving their technology and their abilities. The downside is there signals can be interrupted, captured, or jammed. They’re hard to maintain, and if A nation wants to blind or disrupt another, one nuclear device in the ionosphere would take out most of the satellites except for the Harden ones which will be crippled and blind for a while. Including the ground base connections.

    Most cables laid recently have been for faster trade connections. A millionth of a second can make a difference against a competitor of millions of dollars per day. The entities doing it have strong crony connections to the US gov. In short, hedge fund’s will pay handsome fees for the faster connection, the government will spy on the data and know how much is being transacted and with whom.
    If something happens to the satellites, they will have a near monopoly on information to Europe. (depending on if the other cables have been sabotaged or compromised)
    That cover story is nice, but this is about the control of information and maintaining one’s power while making large sums of money.
    yes, I let my conspiracy side of my brain Runamuck…

  • TL

    Matt has the idea.

    Bouncing off satellites can give data transfers all the bandwidth they need, but there are HUGE latency issues. You can increase bandwidth all you want, but the data still can’t move from point A to point B faster than the speed of light. To bounce that data off a geosynchronous satellite it has to travel ~36,000km up and than another ~36,000km back down for a total of ~72,000km of travel. That travel time at the speed of light is ~.024 seconds. Doing the same trip on an undersea cable can drop that same journey from point A to point B down to 7000km or less, dropping the travel time to ~.002 seconds. So each chunk of data gets to it’s destination in roughly 1/10th the time.

  • Edward

    I’m going with Matt and TL on this one (although TL is off by an order of magnitude; it is 1/4 second for each up and down to a satellite, but the cable does get the data there in 1/10th the time). Latency has long been a problem, and is a big reason that cables are still in operation. Satellites are excellent when you are not concerned about two way communication, such as broadcasting TV, but the more bounces you need to talk around the world, the more latency becomes a problem.

    As individuals, we are impatient. Even a 10-second wait for a web page gets to be annoying, and we often will move on to another page when we have to wait that long. When having a conversation, or playing an internet action-game, the half-second delay of a round-trip via GEO satellite gets annoying, over time. Keeping customers happy may be worth any extra expense Microsoft and Facebook may encounter.

  • Wayne

    TL & Matt are on the right track.

    >There is an article in the WSJ today on this MS/FB cable, which is longer than the Gizmag article.
    [Section B, page 4, (Midwest print-edition)]
    Cost is “around $200 million,” it’s route covers an area that has few if any existing cables, it’s not the first cable MS or FB have funded in-whole or in-part, and for MS the article states their primary concerns are increased bandwidth & a greater control over their skype & gaming traffic.

    Tom & Cotour: interesting stuff.

    Tangentially– anyone who uses ATT; you can check their global-network latency & performance at:

    Gives a good perspective on global speed, point-to-point.

  • mkent

    It’s not a matter of choosing fiber over satellite. It’s a matter of choosing fiber and satellite. And high-altitude planes. And balloons, etc.

    Facebook, Google, and Microsoft are all trying to expand their reach into under-served regions of the world and are using a variety of means to do it. Facebook, for example, has leased substantial capacity on three of SES’s Astra satellites and just recently partnered with Eutelsat to sign a 5-year lease for the entire Ka-band spectrum of Spacecom’s Amos 6 satellite due to launch this summer. These leases are all aimed at expanding their reach into sub-Saharan Africa. The partnership is expected to then transition to Eutelsat’s 150 gigabit / sec high-throughput satellite scheduled to launch in 2019.

    O3b, OneWeb, and SpaceX’s planned LEO constellations are also targeting these markets (in different ways).

    Having said all of that, 160 terabits / sec is about 1,000 times greater than even the new high-throughput satellites. The thirst for bandwidth remains forever unquenched.

  • Stevem


    Satellites have not been competitive from a capacity or cost standpoint for many years for high volume routes such as this new transatlantic route. The capacity of optical technology is astounding and it is improving constantly. A typical transatlantic cable consists of 8 optical fibers; two fibers are reserved as spares and the remaining six provide three two-way optical systems. Each fiber can carry up to sixteen channels (slightly different colors of light) each having a bandwidth of 100 Gb/s. Total capacity for a fiber pair is 1.6 Tb/s and for the entire cable is 4.8 Tb/s. Capital costs for a fiber cable can be hundreds of millions but cost per bit is cheap. Satellites are not cheap either – given launch costs, satellite and infrastructure, and insurance we are looking at what, $200 – $300M?

    The first transatlantic fiber cable was TAT-8 in 1988 with an installed capacity of 280 Mb/s. Optical technology advances have driven transmission costs down dramatically over last thirty years and should continue to do so. As mentioned satellites are great for broadcast and for reaching remote locations. For bulk point to point traffic satellites are limited to the bandwidth available in the microwave radio spectrum. Satellites can use higher and higher frequencies but as anyone who has DirecTV or Dish knows, higher frequencies are attenuated by rain, fog, and even snow. Not so good for reliability. Satellites can use spot beams and reuse frequencies but total capacity of even new satellites is a small fraction of a fiber system.

    By the way domestic fiber systems running between cities may have hundreds of fibers, most of which are “dark” and have “lit” fibers with capacities similar or better than described above. I wish we were in such great shape for the “last mile” to the home.

  • Wayne

    Interesting stuff, thanks!
    –I have Dish for TV, pretty much any substantial rain or snow-storm will cause a temporary “complete signal loss.” HD channels always go first.
    >Its better than weather-radar! (he wrote sarcastically)
    –Talk about that “last mile!” I have ‘basic’ DSL over 20 year old twisted-copper for internet access & I’m apparently on the edge of distance-limit from a central-switch. Despite that, I can (generally) stream live TV w/o inordinate buffering.

  • Wodun

    Facebook’s business is data. They collect data and sell it. On Facebook, you see this in the form of ads but this is only one way they sell data.

    Controlling the tubes means they get to collect more data and not just from people who sign up on Facebook, or who can be tracked through other means.

    The advantages discussed above are all great but the big one is data.

  • D K Rögnvald Williams

    Might be a cover story for bringing up an enemy submarine that was lost.

  • John

    A cable has far, far more capacity than a satellite, especially now that multi-fiber cables are normal. We aren’t talking copper co-axial wires, this is fiber.

  • Edward

    D K,
    Unfortunately, the Glomar Explorer is scheduled to be scrapped. Unless, that is also another cover story…

    Mkent’s report of 150-ish gigabits per second sounds right to me as a modern satellite’s maximum capacity.

    Satellite data capacity depends upon a few things, but my recollection is that 1 gigabit per channel is about as high as they go. The number of channels depends upon the number of Travelling Wave Tube Amplifiers (TWTAs or TWTs), the number of channels per TWTA, the size of the antenna’s reflector, and the broadcast frequency. Distance can also be a factor, as the power flux drops with distance, reducing signal quality, but we aren’t so concerned about that with GEO satellites — they are built with plenty of broadcast power.

    The higher the frequency, the more data capacity. Essentially, with more waves per second, you can get more data per second. Ka-band is the highest frequency in use, right now, and several modern satellites are using that band for all or some of their capacity.

    The larger the reflector, the more the capacity. It’s a physics thing, but larger antennas allow for better signal quality and tighter beams, so you can pack more information without fear of dropping bits. BTW, this is why New Origins streams its data at a relatively low rate; they saved weight on the antenna and radio in order to add more science. The decision seems to be the right one.

    I’m not sure how many channels can be put on each TWTA, but a GEO communication satellite can have from 10 to 50 or so TWTAs. Fewer than 10, and it is hardly worth putting up a satellite, and the upper number is limited by available frequencies at the GEO location, power availability from the solar panels, and heat dissipation capability of the satellite. TWTAs generate a lot of heat, as there is a limit to their efficiency.

  • Wayne

    Tom Billings:
    Serendipity– John Batchelor Show has an interesting re-run tonight >> author-discussion on the Zimmerman Telegram. (by Thomas Boghardt)
    [JB does THE absolute best author-discussions!]

    D K Rögnvald Williams/ Edward:
    “Glomar Explorer,” –cool! I could not remember that name.

  • Edward

    Oops. Bad proofreading on my part. I apparently changed the sentence mid-thought. It should read:

    “The *data capacity* depends upon the number of Travelling Wave Tube Amplifiers (TWTAs or TWTs), the number of channels per TWTA, the size of the antenna’s reflector, and the broadcast frequency. Distance can also be a factor, as the power flux drops with distance, reducing signal quality, but we aren’t so concerned about that with GEO satellites — they are built with plenty of broadcast power.”

    Sorry for any confusion.

  • Wayne

    Thought you were referring to the interocitor control-circuits. ?

    “This Island Earth”
    (Just goofing you!)

    Lost my Dish-TV signal this morning briefly & it promptly down-poured 5 minutes later. (>Better than weather-radar! points south-west-ish. )

Leave a Reply

Your email address will not be published. Required fields are marked *