Skip navigation
Marea submarine cable running into the ocean in Virginia Beach, Virginia Run Studios/Microsoft
Marea submarine cable, funded by Facebook, Microsoft, and Telxius, running into the ocean in Virginia Beach, Virginia.

How Hyperscale Cloud Platforms are Reshaping the Submarine Cable Industry

They have changed how cables are funded, where they land, and how they’re designed.

A couple weeks ago a new submarine cable linking US and Europe came online. The system, called Dunant, set a new record for data transmission capacity on a subsea cable, and that record has everything to do with one of its primary investors and users, Google, and a trend the company started more than a decade ago, which has since reshaped the submarine cable industry.

The tech giant was first of the big four US-based hyperscale cloud platforms to start investing in its own slices of these transcontinental data highways. The group, which also includes Facebook, Microsoft, and Amazon Web Services, has invested roughly $20 billion in new cables all over the world, Alan Mauldin, research director at TeleGeography, told us in an interview for The Data Center Podcast. (Listen below or anywhere you get your podcasts.)

“That’s probably 20-30 cables that they’ve invested in,” he said. And more cables are in the works, due to light up in the next two years or so. TeleGeography estimates that the hyperscalers are contributing about $8 billion to those upcoming systems.

TeleGeography is a telecommunications market research firm, but to the broader internet-using audience it’s best known for its Submarine Cable Map.

New Money

Traditionally, submarine cables had been funded by consortia of big carriers. Each system costs hundreds of millions of dollars to build. Carriers pitch in to build a cable and then sell capacity to their customers.

Another model, which emerged later, was the privately funded model, where a company like Global Crossing would fund an entire cable project and then lease capacity on the cable wholesale to carriers, Mauldin explained.

Both models still exist, but carriers in the consortia are increasingly joined by hyperscalers, who often put up the bulk of the needed investment. This change in the balance of power has affected everything from how the cables are designed to where they land, he told us.

The Unity cable, which came online in 2010, landing in Los Angeles and Chikura, Japan, took an initial investment of $300 million, according to Submarine Cable Networks. The consortium behind it consisted of Google and five Asian carriers.

Google’s trans-Pacific connectivity needs had reached a scale at which it made more sense to own a part of a cable than to continue buying capacity from carriers. Years later, other hyperscalers saw the value in this approach for their platforms too.

“At some point they were growing so fast and becoming so large, it made sense to actually move to a different layer and opt to invest directly in submarine cable systems themselves,” Mauldin said.

Hyperscaler Innovation

The Unity cable, with eight optical fiber pairs, has a design capacity of 7.68 Tbps. The Dunant cable that was launched earlier this month has 12 fiber pairs, providing record-breaking 250 Tbps of capacity. As Google put it in a blog post, that’s “enough to transmit the entire digitized Library of Congress three times every second.”

According to Mauldin, that massive increase in how much data can be pushed through a cable is a result of the innovation hyperscalers brought to the submarine cable industry.

“Cables are a big cost to their business,” he said. They have every incentive to lower that cost on a unit basis, and the best way to do that is “to increase the capacity of the cable as much as you possibly can.”

Eight fiber pairs was state-of-the-art when Unity came online. Soon we’re going to see 16 pairs per cable, Mauldin said, and work is well underway to get to 24 and beyond.

“Those are going to allow the operators to have a massive amount of capacity on a single cable but also to massively reduce the cost per bit of that capacity as well.”

Globe-Spanning Data Center Networks

The sole reason hyperscalers need so much capacity for shuttling data across the ocean is interconnection between the massive data centers they’ve been building around the world. As a result, many of the newer cables that get built land in places they wouldn’t have landed in the old days, when carriers made these decisions based solely on their own needs.

The cloud platforms are “having a large influence in determining where cables land,” Mauldin said. New trans-Atlantic cables now land in places like the west coast of Ireland or Denmark, close to where hyperscale data center campuses have been springing up in recent years.

For a lot more on how hyperscalers are changing the game in the world of submarine cables, listen to our interview with TeleGeography’s Alan Mauldin in full on The Data Center Podcast. Listen below, on Apple Podcasts, Spotify, Google Podcasts, Stitcher, or anywhere podcasts live.

Transcript: How Hyperscale Cloud Platforms are Reshaping the Submarine Cable Industry

An interview with Alan Mauldin, Director of Research, TeleGeography, on The Data Center Podcast

Yevgeniy Sverdlik, DCK:

Hey everybody. Welcome to the Data Center Podcast. My name is Yevgeniy. I'm editor in chief at Data Center Knowledge. Today with us, we have Alan Mauldin. He is research director at TeleGeography. TeleGeography is a research company that tracks the global telecommunications market. Alan has spent the last two decades studying the global communications infrastructure, and that includes terrestrial networks and submarine cables. Alan, thank you for agreeing to talk with me today.

Alan Mauldin, TeleGeography:

Hey, very happy to be here.

Yevgeniy Sverdlik, DCK:

If you've ever searched for any kind of information about submarine cables on the internet, one of the first things you came across was the submarine cable map. It's literally just that, a map of the world's existing submarine cables and the ones in the works. Beautifully put together and available absolutely free online. I've spent many, many hours with the map over the years. It's answered many questions in my reporting work. Alan, maybe we can start with you telling us how the map came into being.

Alan Mauldin, TeleGeography:

Yeah, so we are definitely very well known for our map of the world's subsea cables. We started this map about over 20 years ago, which was a paper map. We sold copies of it on our website. We now have a interactive version we've had for over a decade now, which is available at submarinecablemap.com. There you can see, as you mentioned, all the active cables, the planned cables, who owns the cables, and other facts about the cables. So that's definitely a very popular thing that we offer, and it leads into more of what we offer at the company, which is more of a market research focused on subsea cables, but also data centers, corporate networks, IP backbones, and many other areas.

Yevgeniy Sverdlik, DCK:

I'm sure other similar maps exists, but they probably cost a lot of money to access. Why'd you guys offer it for free? So is sort of like content marketing?

Alan Mauldin, TeleGeography:

So by having the map for free, it's a service that allows us to present data that we've been gathering to the public. It's just freely available for anybody to see. If you want to get the exact locations of cables, the physical routes they take, you're definitely correct. There are paid services where you can get the depth of the cable, the type of cable, all these different metrics, but that is a very expensive thing to offer. So we're offering the basic which countries connect which ones, who owns it, the year it's going to be in service, things like this.

Yevgeniy Sverdlik, DCK:

Yeah, which is probably enough for a casual submarine cable enthusiast. The world of submarine cables has been going through a lot of changes in recent years, thanks to hyperscale platforms getting into the space. Let's start with Google. They were the first to get in in this way.

Alan Mauldin, TeleGeography:

That's right.

Yevgeniy Sverdlik, DCK:

Tell us some and give us some background. How did they get involved, and why?

Alan Mauldin, TeleGeography:

Sure. So many of these hyperscale companies or content providers, whatever you want to call them, they have very large demands for international capacity. So for years, they were buying capacity in the market from the true traditional carriers, right? But at some point they were growing so fast and becoming so large, it made sense to actually move to a different layer and opt to invest directly in submarine cable systems themselves. So Google was the first one, as you mentioned, with their investment in the Unity cable, which entered service in 2010. Since then, you've seen them invest in many other cables around the world. One was launched just this last week, the Dunant cable from France to the United States, and they are involved in many other planned cables. So besides Google, Facebook also is a very heavy investor in new cables. Also, Amazon and Microsoft to a smaller extent also are investing directly in owning submarine cables.

Yevgeniy Sverdlik, DCK:

And these big cloud platforms, they sort of changed the old model of how these cables were funded, how these projects that are very expensive, they cost hundreds of millions of dollars, how they came together. Explain how the old model of consortia building cables worked.

Alan Mauldin, TeleGeography:

Sure. So in the old days, cables were originally just built by a consortium of the largest carriers in each country, and they would invest on the basis of how much capacity they wanted to have for their own internal use. That was for voice calls, right? The voice traffic was very easy to predict in the old days, those type of things. In the late '90s, you saw the introduction of more of a private cable model. Companies like Global Crossing, an infamous company perhaps, but one of the companies that got started in building privately owned cables and then offering the capacity to carriers and others on a wholesale basis.

Alan Mauldin, TeleGeography:

After the telecom bubble exploded in the early part of the 2000s, we have seen really a continuation of both models. You do see private cables being built on some routes, where one party is the sole investor, but you also see consortiums, but often smaller consortiums, or smaller consortia, rather. The difference is now, those consortia don't just include carriers, they also include increasingly content providers. These content providers are often consuming now the majority of the investment in the cable. So clearly for new cables to be built on many routes, they are relying heavily on having the backing and the financing of content providers.

Yevgeniy Sverdlik, DCK:

And then weren't in the old model, and maybe this is still happening, weren't some public money also involved in these older model consortia? Did the governments invest as well together with the telcos, or did I get that completely wrong?

Alan Mauldin, TeleGeography:

So governments do play a role really in helping more remote communities to build cables. So if it's an island in the South Pacific or some remote communities in the Arctic right now, there is some government involvement there, but largely it's private funding that is being used to build and fund submarine cables around the world.

Yevgeniy Sverdlik, DCK:

So with all these hyperscalers now investing in their own slices of these new cables, right, Facebook or Microsoft will not fund the entire cable on their own and just have it to themselves. They invest, and as you said, they invest the bulk of the money oftentimes, and then they get us a slice, some bandwidth on that cable, some capacity. So with this money, with this new influx of money into the market, how many cables approximately have been built in the last 10 years with hyperscaler backing?

Alan Mauldin, TeleGeography:

Yeah. So in the last decade, content providers have invested roughly about $20 billion in new cables really all over the world. That's probably about 20, 30 cables that they've invested in, and there's many more planned for the coming years. I think looking at from what's planned to be deployed this year, in the next two years, we could see another maybe $8 billion worth of investment from content providers coming in. Now, it's important to remember, they aren't the only ones investing in new cables. So overall, there's going to be an even larger amount of investment coming. The content providers share of investment, let's say for the next couple of years, it's going to be about 30 to 50% of the overall total. But on certain routes such as the Atlantic or the Pacific, there's a much higher concentration of content provider investment, and that's really due to where they are trying to link, which is their largest data centers in Europe and Asia back to the United States.

Yevgeniy Sverdlik, DCK:

So let's talk about maybe the effects of these players coming into the market, other than in just pure amount of investment, amount of activity. What other changes have they caused in the submarine cable market?

Alan Mauldin, TeleGeography:

Yeah, that's a great question. I think one of the most important things we're seeing right now, really, is that they're having a large influence in determining where cables land. So they are helping to shift cables to sometimes locations where cables previously had not gone. So we've seen new cables between the west coast of Ireland and the United States, from Denmark to the U.S., which is heavily focused again on where they have their data centers and how they want to configure the cables to most directly connect those data centers to each other. The other thing I wanted to say about their impact is in technology, and how they are really pushing the envelope in terms of how much capacity we can fit down a single cable.

Yevgeniy Sverdlik, DCK:

I want to come back to the technology part a bit later. How has the nature of the cable landing itself changed, the place where a submarine cable comes ashore and upstream from there? How has that changed under the influence of the hyperscalers?

Alan Mauldin, TeleGeography:

Sure. Yeah. So there's been a shift, I guess, over the last decade or so to not just have a cable come ashore and basically stop in a cable station on a beach and have the terminal equipment there. There's really been a push to extend the cable inland directly to a data center or multiple data centers in a country where the cable lands. So in this way, you do have a cable station where there's maybe a power feed, but the terminal equipment is actually in a data center where customers can most directly access capacity on that cable. This was driven not just purely by content providers, but by carriers as well, who were very keen to make the service as easy as possible to interact with others on land and with other cables as well.

Yevgeniy Sverdlik, DCK:

Since there are new places where cables land, and oftentimes cable landing plays a role in data center site selection by hyperscalers and others too, by call providers, by carriers. So has this hyperscaler trend put new places on the map in the global communications slash IT infrastructure?

Alan Mauldin, TeleGeography:

You know, part of where they're putting their data centers are also where there is also a lot of users, as well. So places like Singapore, for example, is a major hub for data centers, also large users. A lot of cables land there. The cables have been there for a long time, so nothing new about that being a hub for submarine cables. In terms of newer locations, new emerging hubs, I can't think of any place where a content provider has singlehandedly created a new hub for cables by themselves, but we do see the development of new hubs throughout the world. It truly has a synergy of the content provider element, [IX's 00:11:47] data centers, and cables.

Alan Mauldin, TeleGeography:

So one good example of this is Marseille in France, where years ago, that was just a transit point. Cables would just go within Marseille, and it would just shoot straight up to Paris or the rest of Europe. But now Marseille has tons of data centers there, lots of cables land there. Lots of content is actually in Marseille. It's not just in other parts of Europe. So that's where you've seen the creation of a new hub, which is very clearly focused on the role of cables to help to drive and develop that location.

Yevgeniy Sverdlik, DCK:

That is a very interesting story, Marseille, and it was Interxion, now owned by Digital Realty. They were early on in getting data centers there, and they became kind of the gatekeeper to those cables. And they built out, right? They converted this old World War II German submarine base, submarine bunker. Now it's a giant data center. Hyperscalers have pushed, as you mentioned, a lot of innovation of the cable technology itself. What have been the biggest innovations they spurred?

Alan Mauldin, TeleGeography:

As we mentioned before, why they're building cables is they really have the most demand and the fastest growing demand on many routes. So for content providers, from printer point of view, cables are a big cost for their business. So they want to lower the cost as much as possible on a unit basis, which means trying to increase the capacity of the cable as much as you possibly can. So what this is leading to right now is we are seeing is the development of SDM cables, which really have a higher fiber pair count. So cables used to have between four to eight fiber pairs in them. Now we're seeing 12, like the Dunant cable has, I mentioned earlier. Soon, we're going to see 16 fiber pair cables, and there's plans for 20, 24 beyond that. So those are going to allow the operators to have a massive amount of capacity on a single cable, but also to massively reduce the cost per bit of that capacity as well.

Yevgeniy Sverdlik, DCK:

So economies of scale, because it costs about the same to deploy it, right? You need the ships and operating the ship and laying it. All that costs about the same.

Alan Mauldin, TeleGeography:

The cost to maintain a cable also is not based on how high the capacity is, but on how long it is, how many kilometers of cable has to be guarded and protected and have ships ready to go out and fix the cable.

Yevgeniy Sverdlik, DCK:

But is there any kind of regular maintenance that happens along the entire length of the cable throughout the years? Or is it just a matter of like being able to access any part of it when something happens?

Alan Mauldin, TeleGeography:

Maintenance does happen from time to time on cables, but the primary reason that there are a fleet of ships available to go out and fix cables is that cables do tend to break fairly often. Usually you don't hear anything about it at all, but it's very important to have ships ready to go out as fast as possible to fix the cables.

Yevgeniy Sverdlik, DCK:

A lot of the cables currently in the ocean are pretty old. When do they generally reach end of life, and what happens then?

Alan Mauldin, TeleGeography:

This is becoming a bigger issue in the industry right now. So as I mentioned earlier, a lot of cables were built in this telecom boom time, the early part of the 2000s, and these cables were able to expand their capacity drastically over the years, from 10 gigabit per second wavelengths, now 100 gigabit per second wavelengths. It's been great. But there's really been a need in the last few years to go beyond these older cables, which are still very useful and valuable, but go to higher capacity cables to really refresh some of the cables on many of these routes. So right now you're seeing a large number of new cables built really all over the world, and it's part of trying to refresh and replace these older cables.

Alan Mauldin, TeleGeography:

Now, the older cables, cables are designed with a minimum design life of 25 years. That's a rough engineering guide. It's not a magical number. That cable doesn't fall apart when it's 25 years old and stop working. It's designed to last at least that long. So what often happens, though, is a cable can be turned off before that date if it becomes too expensive relative to other tables on that path on a unit cost basis. So a cable has only, let's say a terabyte of capacity. Could be very expensive. It's an older one compared to a new one that's offering maybe 40 terabytes, right? So at some point, it's too expensive to keep it going and you turn it off. Also though, you can keep the cable running well past 25 years. If it's a route that has less demand, fewer cables are on that route, and you can keep adding capacity to an older cable, why not do it?

Yevgeniy Sverdlik, DCK:

So once they do go out of service, they just get pulled out and all the equipment along with it, or what happens?

Alan Mauldin, TeleGeography:

That's right. The cables tend to be salvaged. Ships go out and they will recover the cables and then take them and salvage them. There are cases, however, where cables are moved and reused at different locations. That's fairly rare though, for that to happen.

Yevgeniy Sverdlik, DCK:

One of the big drivers behind new cable construction is providing a redundant paths to maintain connectivity when cables break. What happens when a cable breaks somewhere in the world? I mean, generally most people don't don't notice, but some places maybe don't have such robust infrastructure.

Alan Mauldin, TeleGeography:

Yeah. So most countries are linked to the world with many different paths. So whenever one cable has an outage, generally there's no real noticeable impact for the average user in a country. But when you have multiple cable cut events, which do happen due to weather, anchors, earthquakes, fishing, whatever. There's many things that can happen to cables and cause outages. If you lose a larger number, there can be some impact then for the user experience. So there are plans that exist for how companies plan to use other other cables as backups in case their cable does have an outage, but ultimately the best way to ensure a high level of service is to have a lot of cables. A larger number of cables gives you the best protection against an outage.

Yevgeniy Sverdlik, DCK:

Are there are places around the world that don't have that redundancy at all?

Alan Mauldin, TeleGeography:

Some spots have just I think fewer cables. So one that I think is worth noting is Vietnam, actually. They've got a few number of cables, and the sea there, the South China Sea is very, very shallow, a lot of fishing, so they have a lot of outages of cables recently that connect to Vietnam. So there's plans for more cables, of course, to help to mitigate some of that. Other places, such as Tonga, for example, they have one cable. They got it a few years ago, but that cable had a fault. I think it was two or three years ago. So they had to rely on satellite backup. So even one cable sometimes is a great thing for these islands to have, but getting two would be even better.

Yevgeniy Sverdlik, DCK:

Before I let you go, let's get something straight. There is a myth that seems to be undying about sharks being a threat to submarine cables. We at DCK, I want to be clear here, we have been guilty of perpetuating it unfortunately. Alan, as an expert on submarine cables with two decades of experience, tell us. Do sharks bite cables and cause network outages?

Alan Mauldin, TeleGeography:

Yes. I'm really sorry to tell you this, but it's not sharks. The vast majority cable faults are caused by fishing and anchors. Those are the main causes. Also, you have earthquakes and underwater events like that. But sharks, no. They are not a cause of cable faults. They have caused I think zero faults in cables the last 20 years, so there is no risk for sharks trying to eat any of your delicious bandwidth anytime soon.

Yevgeniy Sverdlik, DCK:

So it seems that the West's addiction to sushi is in conflict with our addiction to the internet.

Alan Mauldin, TeleGeography:

Could be. So I don't have any guidance for you here, but I think the cables are going to be okay.

Yevgeniy Sverdlik, DCK:

Just drawing a weird parallel there.

Alan Mauldin, TeleGeography:

That's okay.

Yevgeniy Sverdlik, DCK:

Okay, Alan, that's all I have. Thank you so much for your time.

Alan Mauldin, TeleGeography:

Thank you for having me. It was great to be here.

Hide comments

Comments

  • Allowed HTML tags: <em> <strong> <blockquote> <br> <p>

Plain text

  • No HTML tags allowed.
  • Web page addresses and e-mail addresses turn into links automatically.
  • Lines and paragraphs break automatically.
Publish