Disclaimer: This interview is for informational purposes only and should not be relied upon as a basis for investment decisions. In Practise is an independent publisher and all opinions expressed by guests are solely their own opinions and do not reflect the opinion of In Practise.

{audio:0:06:46} Your optical network infrastructure, transport -waves, IP transit, dark fiber experience is what piqued my interest, especially in relation to the recent Cogent Sprint deal. Maybe let’s start there. Walk me through what the Windstream products are, on the wholesale side. When you are talking wholesale, are you talking carrier to carrier or are you talking carrier to any large buyer?

{audio:0:07:17} Let’s talk about the demographics that we deal with here. Our customers are wholesale carriers such as the AT&Ts, the Lumens, the regional providers, where you have Bluebird or other regional players throughout the country. You have content providers, like the Amazons, the Googles, the Twitters, the Facebooks; large content providers and hyperscale companies. They are buying pipes upon pipes; millions of dollars each year, just because their business focuses on content. It is pretty simple. We sell to some large data centers because of their customer base. We connect and interconnect data centers for some of the larger data centers, such as Equinix and Digital 365.

{audio:0:08:20} My customers are building networks; that is their job. They’re not just utilizing this as a commodity. They are building networks because they have end users.

{audio:0:08:34} Let me simplify it for myself and let’s use Netflix as an example, because it’s an easy-to-understand service. Why does Netflix need to build a network? Can’t they just link into some carrier neutral data centers and ride the internet through the transit network? Why do they need to build their own network?

{audio:0:09:06} Netflix wouldn’t be an example of a company that needs to build their network. Netflix is a large commodity user. Internet stream is a commodity, kind of like a Shopify. These companies use the internet as the commodity of their business. 

{audio:0:09:28} But maybe Amazon does because of AWS?

{audio:0:09:32} Correct. They have to build infrastructure because they need end of scale access, throughout every region, every city and state in this country and globally. They need to be at every POP in Los Angeles, every POP in San Diego, every POP in the Bay Area and every POP in Virginia.

{audio:0:10:06} Do they actually have to be in every POP? They can’t just consolidate to a handful of really big POPs?

{audio:0:10:14} No, because they want to be at the edge. They want to have access at the edge. They don’t want to have to consolidate. Obviously, this is a continuous process. We sold terabits to Amazon and pipes that were selling them are in various regions. They are always looking for new regions. We have done builds – not so much for Amazon, but for Google – and used them as an anchored tenant because they want new routing.

{audio:0:10:49} Do you mean laying new fiber?

{audio:0:10:53} Physically laying new routing; not just procuring dark fiber or IRUs?

{audio:0:10:57} Laying new routes and putting dark fiber on top of it.

{audio:0:11:04} When you are saying they need to build networks, just so I understand, these hyperscalers – Google, Amazon, Microsoft – really need a presence on the edge. They are chasing something forever on that front, I guess. They always have to expand their network.

{audio:0:11:25} There is always growth. Population is always growing; usage is always growing. Look at it this way; look at the 405 Freeway. You take the 405 south, if Amazon or Google owns the 405, each off ramp is an edge. There are people that live off each off ramp.

{audio:0:12:03} Does the latency matter that much, whether you are a Coresight versus an Equinix, in LA? You have to be in both because the latency difference is measurable enough?

{audio:0:12:18} It does matter, to a certain degree, but it also depends on the congestion. Certain off ramps are a little bit more impacted than others.

{audio:0:12:35} What are these guys buying? Are they buying wavelengths? Are they buying dark fiber? Are they buying IP transit? They are probably buying all of it but how do they decide which route needs to be dark fiber, which route needs to be wavelength? 

{audio:0:13:03} It is led by the carrier and is managed by the carrier. It’s on network. For dark fiber, you can place your own equipment on top of it and you control that fiber.

{audio:0:13:12} Are they, effectively, achieving the same thing, just managed differently? Is one a managed service and the other an unmanaged service?

{audio:0:13:21} It’s not the same thing because, with dark fiber, you don’t have to use transport on top of it. You can even put DIA or transit with it. It’s how you manage it and what you do with it. If you are on the freeway and you have two lanes, that’s wave; you can only do those two things on those two lanes. Dark fiber gives you a five-lane freeway; there is more stuff that you can put on it. 

{audio:0:13:56} And much longer-term contracts.

{audio:0:14:02} Yes; contractual is totally different. That’s not always necessarily the case because you can do a lease on dark fiber. I don’t see a lot of that but you can do a five or 10-year lease. You can do a five-year MRC for waves. But what we’re seeing is that the Amazons and Googles have been purchasing a lot of waves, obviously at the terabit level, but they are also on these new builds that I was telling you about. Google, for example, is buying dark fiber because of these new routes, because of communities and suburbs that are growing.

{audio:0:14:39} I can’t say the state, because that will give away what they are doing, but in these various regions, the population is growing, communities are growing so there is going to be an increase in demand. We can use the analogy that, if more people are moving towards a certain off ramp or new houses are by a certain off ramp on the freeway, you’re going to have to put an off ramp there so that people can have access to it.

{audio:0:15:10} It’s just another layer of critical infrastructure. You could lay new electrical wire, plumbing, water and all of that, but this is just the modern economy utility infrastructure that needs to exist?

{audio:0:15:32} Definitely. Look at where Moreno Valley, Riverside or Temecula was, years ago. Let’s think about how they were 30 years ago. They didn’t have the population they do now. With that population increasing tremendously, thousands of percent over the last 25 years, there has to be some infrastructure to support all of those end users that are going online, that are using fiber to connect to their house.

{audio:0:16:07} How is it all architected and put together? There is so much fiber, under the ground, owned by so many different parties. It sounds as if some of the hyperscalers are some of the biggest buyers and biggest players in that market; how do they organize their networks and decide which route they need? When they are buying a wavelength, that’s just point to point? That might be LA to Temecula and they are buying that route. Are they buying a direct route or are they hopping through five different POPs to get there and they are buying a wavelength, with routing, through different POPs?

{audio:0:16:52} It depends on the actual route. Let’s just use a larger example. For instance, say you are with Lumen, you are here in Los Angeles and you want to go point to point; you have a corporate office in Chicago. You need a point to point from your LA hub to your corporate office in Chicago because, in Chicago, we can connect you to one of the largest IP transit providers, which is Cogent. 

{audio:0:18:23} Some routes can take you to Las Vegas, around this route, to Salt Lake City, to Denver. Other routes can take you along this string; these are all point to points. There are some routes that go straight through, maybe with a different carrier. I doubt it could go straight through. But the point is, how these people set up their infrastructure is based on a few things. One of those is redundancy; they want one going this way and one going that way. They don’t want it all going the same way because there is then only just one point of failure.

{audio:0:19:03} It’s almost like mapping out the freeway or highway system in the US where you can get from point A to point B and you are, obviously, going to go through cities to get there. To go from LA to Chicago, you’re probably hopping 10 times, across a bunch of different city pairs.

{audio:0:19:40} Cogent’s infrastructure is set up totally differently from Windstream.

{audio:0:19:48} They want route diversity so one of the things that Cogent has said about this deal is that the Sprint network was on the first long distance fiber network deployed and its routes are armored buried fiber and the right of way is completely different from other rights of ways. I think it follows a lot of the railroads. Dave argues that it provides routes that are independent, versus what others are buying in the market. I don’t know how true that is. I’m sure it’s true in some places and not in others. Maybe it’s one of the selling points if you are looking for redundancy and diversity of routes.

{audio:0:20:35} To give you an example of what we were talking about before, Dave owns the 405 but now he owns the 5. They can both take you to San Diego, to a certain degree. 

{audio:0:20:54} I have a question on how these deals are sold. My understanding, just from what Dave has said is, if the market is growing at high single digits – someone estimated it as being a $2 billion market in North America – is the growth just from these extensions of population into these new territories, or is there something else that is driving the growth in that market?

{audio:0:21:27} When you think about technology, think about anything that deals with content and data, and anything that needs some type of internet backbone; even the apps on our phone need internet backbone or the phones won’t function. They are connected to cell towers and cell towers are connected to POPs. Everybody knows that market share is continuously growing. Everything is going to be app based. Look at our homes nowadays; everything is smart. Of course there is growth, it speaks for itself. But how you measure that, they don’t pay me enough for that so I’m not sure what analytics they would use.

{audio:0:22:23} I was just curious as to whether these contracts come with price increases on them? Is that contributing to some of the growth annually? If it is high single digit growth – let’s say it’s $2 billion growing $150 million a year, roughly – is $75 million of that coming from annual contracted price increases and $75 million is coming from new routes being deployed and what you talked about, with Google and Amazon being an anchored tenant on new fiber? I’m just trying to think through what the contract terms are and how to decompose the growth.

{audio:0:23:01} I think it depends on who is giving the numbers, personally.

{audio:0:23:12} How are the contracts in the wavelength market put together? Are they long term, short term? Are there any volume components to them or is it all fixed cost?

{audio:0:23:28} It just depends, because it is point to point. Let’s say Google needs 10 waves. LA to Chicago, LA to New York, Seattle to Portland. They are going to maybe need a one-year MRC or possibly a three or five year. It’s usually a one, three or five. Sometimes companies do two, but it’s between one and five years.

{audio:0:23:58} If it’s a company like the names we are mentioning, obviously, there are pricing discounts and the pricing is a lot more aggressive, just because of the market demand, just like anything else. The bigger your name is, the bigger the discount.

{audio:0:24:20} What are the switching costs? When you are selling against Lumen and you are trying to steal Google away from Lumen on a certain route or 10 waves, let’s say, how do you do that? What is the proposition? How do you win that business?

{audio:0:24:40} This is pretty straightforward. There is no carrier that has monopoly on any company. Google is not going to put 80% of their network with one carrier. It is just not going to happen; it’s like Walmart using one type of product.

{audio:0:25:06} Exactly; why would you handcuff yourself to the one business?

{audio:0:25:10} With the various routes, with the various markets, there are certain routes that might benefit a particular carrier, versus another carrier. Maybe this route is unique and the other carrier does not have it. Perhaps this price point is more aggressive with this route than with another carrier. There are price factors involved, there is redundancy involved and, sometimes, when it is a big RFP, two or three carriers can get the bid. If we are talking about a terabit, sometimes that doesn’t all go to just one carrier.

{audio:0:25:48} If you meet some of the guys that are doing this, if you go to NANOG or any of these trade shows, especially with the content providers, like the Googles and Facebooks, these guys are young. They go to work in shorts and t-shirts every day, and they are making million-dollar business decisions. They’re pretty diversified.

{audio:0:26:14} Let’s say, after one, three or five years, there are some Lumen routes that you think you can win from Google or Amazon, how easy is it to pull that share onto your network? What does it take, operationally, to do that? Is it a big switching cost? Another way of asking is, how much of that revenue do you expect to renew on each renewal cycle?

{audio:0:26:46} That’s interesting. This might not be the best example because we are talking about the giants in the industry, so you tend to take less of a profit margin compared to any other potential deals just because of the long-term relationship and potential growth.

{audio:0:27:07} I get that. You don’t want to piss off Amazon and lose them as a customer so you are going to give them the best price.

{audio:0:27:13} If your business model says you need 20% revenue or 20% net profit, if you have got a large deal with Amazon, you will take 12%. That happens with everybody. 

{audio:0:27:30} Do they ever switch? Do they ever turn them off and switch to another?

{audio:0:27:34} Sure. Zayo lost a lot to Windstream and that is why their pricing, recently, has got aggressive, because we have been able to a big chunk of their market share. We’ve done it, surprisingly, over the last four or five years. Keep in mind, our wholesale division isn’t really that old, especially for content providers. We have been able to really get our foot in the door with some of the largest content providers and carriers. Obviously, the AT&Ts and Verizons are our largest customers; they are buying 70 million a month. The numbers are just ridiculous just because of the metro switches and the interaction on that level. That’s a whole other conversation. That’s not wave or dark fiber.

{audio:0:28:26} At the end of a contract term, what does it take to pull the traffic onto your network, from another? What does the customer have to do, if anything?

{audio:0:28:44} The customer doesn’t have to do anything. If they have 10 waves with Lumen, right now and their contract expires in October, we have a date to install, at the beginning of October. All we are doing is getting our field techs and our technicians out there to do the site survey and plug into the electronics that are already there, to our POP. It is really not a major deal. We can do an install, on that magnitude, within 90 days, easily. 

{audio:0:29:12} We did a big route, in Sacramento, California, on 2,200 gig waves, in 28 days. That was major; we caught a lot of attention on that. It doesn’t really take a lot, with the right people involved, if they are just focused on that install.

{audio:0:29:33} Basically, they can move market share around. If someone is getting really aggressive on price, they are not so locked in technically that it’s hard to do?

{audio:0:29:44} No. It’s really not price driven a lot of times because price is always shifting. That is why they never sign five years. A company like Amazon is not going to sign a five-year MRC because the pricing is shifting. We’re looking at one to three years.

{audio:0:30:06} Is there a contractual step down or step up in pricing, with inflation or anything like that?

{audio:0:30:12} From what I’ve seen and experienced through customers and through deals is that we have the right to increase our pricing. That’s not just Windstream; most carriers have that right. We have a price increase that is taking place at the end of this quarter. As somebody like myself, I am contacting some of my larger customers and saying, hey, renew right now so you don’t get hit with the price increase. There are ways to avoid it.

{audio:0:30:53} How do I think about the difference between a buyer of dark fiber and a buyer of wavelength?

{audio:0:31:01} It’s really not a big difference. You can’t really separate opportunities like that.

{audio:0:31:07} What’s the buying decision? Let’s say it’s Amazon or Google or one of these guys that we were talking about. They’re signing one to three year MRCs, for wavelength. My understanding is that dark fiber is usually sold on longer-term contracts?

{audio:0:31:28} Yes; dark fiber is usually five, 10 or 20. 

{audio:0:31:34} Why would Google want a 10-year dark fiber infrastructure instead of a three-year wavelength contract?

{audio:0:31:49} Because the fiber is going to be utilized and have more leverage than just the wavelength. The wavelength speaks for itself. It’s going to be a point to point. The dark fiber gives them the ability to manage their infrastructure differently and to place any type of application that they choose to put on the dark fiber. 

{audio:0:32:11} It’s more application agnostic?

{audio:0:32:13} It’s more architectural implementation.

{audio:0:32:23} I’ve heard that Lumen, AT&T and Verizon have been reluctant to sell dark fiber, recently? I don’t know how true that is. 

{audio:0:32:34} That is the second time I’ve heard that. I haven’t come across that, personally, in my dealings. But I’ve been hearing that so I’m starting to believe it’s true. I haven’t had any opportunities knock on my door and say, hey, AT&T won’t do this. They’re not usually a dark fiber competitor, nor a wave competitor. It’s usually just Lumen or Zayo.

{audio:0:33:06} Zayo got acquired a few years ago and it sounds as if you have been taking some share from them. Have you noticed anything change with them, operationally, since they got acquired?

{audio:0:33:22} Do you think they are missing out on sales they used to get?

{audio:0:33:27} I think they are just throwing a lot of stuff against the wall to see what sticks.

{audio:0:33:39} On this Cogent Sprint, you worked at Cogent and you’ve been in this industry for a long time, what do you think of it?

{audio:0:33:48} I wouldn’t say I know Dave personally, but I’ve had a chance to spend some time with him, when I was hired with him and on some of the first trainings I was involved with, it was him that did a lot of the speaking and explaining. Really going by his track record and how he runs Cogent – which could be good or bad, depending on what side you are on – he doesn’t usually make dumb or unwise decisions. He is very calculating. To be really transparent, I really don’t have the intellect to say, no, that has to be a bad decision. If I was a betting man, I would bet that he made a good decision.

{audio:0:34:43} They’re getting paid $700 million, by T-Mobile, to take it on.

{audio:0:34:53} Everything is not what it sounds like, especially when you hear numbers like that. I don’t know the back end or the commitment of that. I don’t know how any of that is written contractually. That’s media and press.

{audio:0:35:04} That is contractually written in the agreement. Cogent buys for $1 and T-Mobile buys transit for $700 million; they don’t plan to use any of it. That part is solid. The hard part to unpack is, currently, the company is burning through $300 million a year, of cash. The question is, can Dave and his team transform that asset? This is like Dave’s playbook for acquisition. Can he strip it for its parts and reconfigure the network in such a way that you have some additional growth opportunities for wavelength and dark fiber and then can you get some cost efficiencies of the core Cogent business, which sounds as if he can? If you’re curious, the way he has laid it out is, something to the tune of 225 million of network costs between the two companies, just by pulling off net circuits from Sprint, onto the Cogent network and combining the networks internationally to reduce some of the international costs that Sprint has. Then getting off the Lumen long-haul IRU that Cogent has been on for a long time saves another 15 to 20 million of maintenance spending.

{audio:0:36:38} I don’t understand how Sprint or T-Mobile actually benefit.

{audio:0:36:47} I think, for the T-Mobile side, they have a business that is in decline, they don’t want and it’s non-core to them. They turned off the Sprint network and the Sprint network was utilizing that Sprint wireline business. Mathematically, if you look at it, their total commitment is 1.3 to 1.5 billion, if you account for all the cash burn that they would have to incur over the next year plus, until they close the deal. Plus, there is 700 million of IP transit that they are going to buy from Cogent. There is 25 million of future severance that they are going to have to pay to Cogent, as they terminate employees, and there is 100 million of additional indemnification coverage, on top. 

{audio:0:37:31} If you are T-Mobile, you look at it like, I am bringing $300 million a year, on this business; I’m paying $1.5 billion to get rid of it, so it looks like a 20% return. And it’s non-core and Cogent can do things that T-Mobile can’t. For instance, T-Mobile bought Sprint and they have a lot of PR and regulatory issues around firing people. Cogent, obviously, won’t have those issues. Dave has been very careful, publicly, about what he is going to do with the headcount but he walked through some examples when they announced the acquisition. They bought PSINet which had about 9,000 employees and they retained 51. Allied Riser had 971 employees and they retained two.

{audio:0:38:23} The difference is, Sprint’s network is physical and it relies on tens of thousands of route miles of physical infrastructure and hundreds of facilities. He is going to have to retain more people than he has done in the past and I think he will want to do that anyway. Let’s say there are 800 people that could be laid off from that business, to restructure the cost, T-Mobile can’t really do that because it is very bad PR and the yield to them is very low, relative to size of the company and the wireless business. 

{audio:0:39:01} Whereas for Cogent, if they lay off 800 people over three years, let’s say, they don’t have a PR nightmare around it; at that point, it is just part of M&A and it is much more material to the Cogent business. It is kind of a sensible deal from both sides. I definitely think that Cogent is getting very good value for it. But key to the whole question is, if they can’t grow in these other new services, like wavelength and dark fiber, then it’s a fine deal, but it’s not a good deal. They are not going to be worse off for it, but it’s going to be a big operational headache to integrate the companies and do all this work, if they are not going to get any of the new revenues, which is why I’m so curious about the prospect of them getting these new revenues.

{audio:0:39:59} The way he describes it, he’s getting access to the wavelength market and the dark fiber market, with negative cost.

{audio:0:40:11} That’s such a Dave move.

{audio:0:40:21} Smart people buy dumb pipes; I’ll undercut anyone by 50%. That’s the mantra of the business. You said it’s not a price competitive market, which is really interesting. If it’s more architecturally based and the customer are building networks and not just buying a commodity, like IP transit, how much does his negative basis matter? That’s one of my questions. Yes, you have a great cost structure; yes, you can offer the lowest price, prices that others don’t want to touch. But maybe your routes suck.

{audio:0:41:04} Keep in mind the kind of customers we are dealing with. We’re talking about the Amazons and the Googles. That is a different mindset compared to a regional wave customer or smaller end customers. That is not the general rule, because price does matter to a certain degree. At the same time, some of the largest content providers are not getting waves for a dirt-cheap price, compared to the market. They are paying $2,300, $2,400, which is very much the market price. It’s not extremely low and it’s not high. 

{audio:0:41:45} What do you mean by $2,300, $2,400? Is that the monthly cost?

{audio:0:41:53} What do they get for incapacity? One of the other questions I had was, with all the technology upgrades on DWDM, it sounds like you can run up to 800 gigabits per second on a given wave, using the latest DWDM; is that accurate?

{audio:0:42:23} I haven’t heard that. I deal mostly with a 10 gig or 100 gig wave. Very seldom one gig. We also have a 400 gig wave. Those are point to point lighted fiber, from an end point, so I’m not quite sure where this 800 comes from.

{audio:0:42:46} Maybe 800 is the future next generation. How are those priced between 10, 100 and 400? 

{audio:0:42:58} It depends one which end to end. If we’re dealing with a POP to POP – I’m sure you’ve heard of LA, Wilshire One and Chicago, 350 Cermak – those are major pops, internationally and domestically. A 10 gig wave will cost you somewhere between 800 to 900. A 100 gig wave can run you anywhere between 2,200 to 3,000. It just depends on who is selling it.

{audio:0:43:34} How much more would a 400 gig wave be?

{audio:0:43:37} You can times that by three. I would say around 6,000 to 6,500.

{audio:0:43:48} They buyer gets some economies of scale, as they go higher?

{audio:0:43:56} Yes, but not everybody’s equipment is prepped for 400 gig; that’s the new thing now. Not everybody’s hardware is ready for that.

{audio:0:44:08} Do you mean that at their POPs they don’t have the interfaces to handle 400?

Cogent's Sprint Fiber Deal

Former Manager at Cogent Comunications

inpractise.com/articles/cogents-sprint-fiber-deal

Executive Bio

Former Manager at Cogent Comunications

Interview Transcript

Related Content

Cogent Communications & Global ISP Competition

Former Director at Cogent Communication

Cogent Communications, Microsoft, & Dark Fiber Networks

Former Director, Global Network Engineering at Microsoft

Weekly Update: Universal Music, Burford Capital, Cogent, Wayfair CastleGate, & ODFL

Cogent Communications: Hyperscalers' Dark Fiber Demand

Network Acquisition Executive at Hyperscaler

Partner Interview