Great. Thank you all for joining us. My name is John Hodulik. I'm the media and telecom analyst here at UBS, and welcome to our presentation with Cogent Communications. We're very pleased to have Dave Schaeffer, the founder and CEO of the company, with us today. I've got a bunch of questions that I'm going to run through. We've got about a half an hour, but if you have any questions, you can either just raise your hand and blurt them out, or I've got the iPad here
So you can use the app to enter any questions that you have. So with that, Dave, you know, this time of the year, it's always good to sort of, I think, reflect on what, what, what's happened in the year and look out to 2024. So if you could give us some perspective on sort of the positioning of the company as we've gone through the year and talk about your priorities for 2024.
So first of all, I'd like to thank investors for taking their time. I'd like to thank UBS and I guess former Credit Suisse, for a great venue, and John for his time and interest in the company. You know, Cogent has had a very consistent business. In many ways, we're a company that hasn't changed in 18 years as a public company. You know, at the beginning of Cogent, we set out with a fundamental belief that the Internet would be the only network that mattered, and we built a network that was optimized for the Internet. Along the way, the telecom crash occurred and the dotcom bust, and we capitalized by buying a total of 13 public and private companies that we dismantled and integrated into Cogent.
And then for the next 18 years, executed on that business plan of running an IP over DWDM network, servicing mid-size businesses and other service providers, selling internet and VPN services. Earlier this year, we did our first acquisition in 18 years. We acquired the former Sprint Global Markets Group business from T-Mobile. This was the country's first nationwide fiber optic network, built in the mid-eighties, and a customer base of approximately 1,400, mostly large enterprise customers, buying a complex set of solutions, generating about $565 million in revenue, but burning $1 million a day. The company had negative $300 million of EBITDA, negative $30 million of CapEx on a revenue run rate of $565. From signing to closing, T-Mobile picked up that burn and implemented a number of structural changes in the business. We're continuing those changes post-closing.
We have shown material progress in reducing the burn in that business and expect within one year of closing, to have that burn rate down to about $80 million a year from $300 million at signing, about $190 million at closing, through streamlining the product set and bringing traffic on-net. 93% of the revenues of the Sprint GMG business never touched the Sprint network. It was completely off-net. That was not an economically viable model. The second thing that we acquired was the physical network. This is a fiber optic network that spans 19,000 route miles of intercity fiber, about 1,200 miles of metro fiber, and 482 owned telephone switch sites. We are converting 45 of those sites into data centers. We are dismantling dead telecom equipment.
We are interconnecting those facilities to our metro footprint and converting the fiber optic network into one to sell wavelength services, optical transport services. We are progressing well on that effort. By the end of 2024, we will have integrated the network. We will have been able to enable 800 U.S. carrier-neutral data centers where we can sell wavelength services. Wavelength services complement the Internet. They are more expensive, but they have three attributes that make them uniquely valuable to their customers. They are deterministic. You know the exact latency from point A to point B. Two, they are suited for very large file transfers, big packets. And third, they are secure because they do not touch the public Internet.
So in the acquisition, we are doing two things: fixing the acquired customer base and turning them from a negative margin to a positive margin business, and we're more than halfway through that exercise. The second is taking what was a fallow asset and repurposing it for optical transport. So it's what happened to us in 2023, and it will be our major focus in 2024, allowing us to capture about 25% of that wavelength market, similar to what we've done in the IP transit market.
So let's talk about the optical or the wavelength market a little bit. So you said the goal is to get to 25% of that market. How big is that market? Who are the competitors, and then what are the growth drivers? Is this a growth business, or is it, you know, think of the future of the internet, and AI, and 5G, and, you know, other big drivers of traffic that are expected?
So the fundamental thesis of Cogent in 1999 is as true today as it was in 1999, which is the internet will eventually consume most other telecom services and products. The internet transit market is one that Cogent dominates. When we entered that market, there were 200 players, and we were literally 0% market share. Fast forward 24 years, Cogent is the largest carrier of internet transit in the world. We connect to 1,600 data centers and carry roughly 25% of global internet traffic. That's about 1.3 exabytes a day of traffic. That market is going to continue to grow. The wavelength market in 1999 was actually dominated by AT&T and MCI. There were 2 companies selling wavelengths to long-distance resellers. That market ended in the early 2000s.
The second wave of wavelength customers were those who were regional ISPs trying to build a backbone by renting wavelengths. Level three entered that market and eventually took significant market share, and today, AT&T and Verizon are no longer players in that market. The market is dominated by Lumen and Zayo, are the two major players, and a few more regionalized players like Windstream and Uniti. It is approximately a $2 billion addressable market, and the primary use cases for wavelengths fall into three categories. First, large content-generating businesses looking to replicate that content. So companies like Google, Microsoft, Meta, Amazon, all use wavelengths to move data from data center to data center. The second group of customers are regional access networks. Think of them as islands of internet connectivity that need to be connected together. They can be international carriers or regional domestic carriers.
The final addressable market for wavelengths are very large enterprises and governments who value the security and independence and are not very price sensitive. That market is a $2 billion U.S. addressable market. This is inner-city wavelength services. There's an additional $1.5 billion market in metropolitan wavelengths that is within a city, and then finally, about another $3 billion global market. Cogent intends to focus on that U.S. inter-city, $2 billion market. We will have some significant competitive advantages. One, we will have more endpoints than any of our competitors. Two, our routes are unique. They are a long right of way that is not shared, where most of the other networks were built on public right of way. Ours are on railroad right of way, which is unique. Third, we will have the ability to give very accurate, detailed mapping of our network.
That's extremely important in a deterministic product like a wavelength, irrelevant for the internet. So a customer cares about the exact location of the fiber. It is an unprotected service. All of our competitors are roll-ups. The network we operate was built by a single party. Quite honestly, it sat dormant for 15 years, but the records are still accurate, and we are able to give our customers that detailed, granular information. And then finally, because we're starting with an empty network, we can architect it in a way that we can provision more quickly than our competitors. Today, a wavelength service ordered from one of our competitors will take 90, 180, or more days to install.
Cogent today is not able to serve all locations with rapid provisioning, but by the end of next year, we will be able to serve 800 locations and provision a wavelength in a 17-day guarantee. That is vastly superior to our competitors, and we have a network that has a zero cost basis. We paid $1 for the actual network. So we will be aggressive on pricing. We think because of our brand, our credibility in transit, our customer relationships, our sales force, within a seven year period, we can replicate the success we had in transit and get to 25% market share. The final part of your question, is the market growing? So in unit volume, all telecom services are growing. The Internet is probably the fastest growing of those segments.
I was a bit skeptical about the dollar growth in the wavelength market when we acquired these Sprint assets, and I based our analysis and due diligence on the concept that it would be a flat market. Subsequent to that deal being announced, the demand that we've received has shown us that the market is probably growing, as industry analysts expect, at about 5%-7% per year. So I think we will capture both the existing market share and growth in the expansion of the market.
Are there any synergies between s o we should think of the wavelength or optical business enabled by the Sprint transaction as a different business than your existing transit business. But are there any sort of synergies, whether it's from an infrastructure standpoint or from a client standpoint, that we should think about?
Yeah, there are a number of synergies, and it's a big part of the way we are extracting $220 million of costs out of the combined company. The synergies are, we have one field service organization.
Right.
We can migrate some of our traffic off of IRU fiber onto the now owned fiber. We have one provisioning team, one set of hub sites or aggregation points, one metro network, and maybe most importantly, one sales force focused on this market segment. Cogent today has about 650 salespeople out of our 1,950 employees. Sprint, at the time of acquisition, had nearly nine-- we acquired 950 employees. When we signed the deal, there were about 1,350. When T-Mobile had decided to start to market this business, there were 1,800 employees. We're down by more than 50%, but there were only 51 salespeople in Sprint, and in my opinion, it was a stretch to call them salespeople. They were account managers. They didn't have quotas; they didn't hunt for new business.
Cogent has a 257-person, net-centric sales force that sells transit globally. We also have 400 corporate-facing sales reps that sell to businesses. We have a small enterprise sales team acquired from Sprint that'll focus on those legacy accounts, and our 257 wholesale salespeople are the exact same team that will sell wavelength services. They already have pre-existing customer relationships. So today, we're the primary provider of transit to Amazon or Microsoft or Meta. We just go to those same buyers, and we say, "We also have a wavelength product." Same salesperson, same engineering group, and that familiarity with our value proposition or quality gives us a huge competitive advantage and creates a tremendous synergy in sales and marketing.
So my last question on the, the Sprint transaction. You know, you're talking about 25% of a $2 billion and growing market. What, what should we think of, you know, longer term and obviously some of, I guess, wind down on some of the legacy services that are still being serviced through the GMG infrastructure? What kind of margin should we think about that business longer term? Just to give us a sense for run rate, sort of, EBITDA potential for that, that new business.
So last quarter, we had the highest margins in Cogent's history at EBITDA margins of 47.7%. That is partially as a result of the payment stream from T-Mobile, where they're paying us $700 million over time, which gets added into EBITDA. If you strip that out, our margins had actually declined. So going into the transaction, Cogent's organic business had approximately 39% EBITDA margins. The negative margin that we acquired from Sprint brought the combined company's margins to about 20%, and then you add back the payments from T-Mobile to get to that 47.7%.
Got it.
Those payments will go away after 54 months. They're stairstep down after the first 12 months. Our margins at that point should be in the mid-30s, about 35%. The combined company will be growing at between 5% and 7% a year, a revenue base of about $1.5 billion, and an EBITDA margin or EBITDA of about $500 million. In understanding the growth in the business, you need to really understand three unique businesses combined together on one network. So the enterprise business will be stable at about $450 million of revenue and no growth and 20% margins. While we will see some attrition, we will also sign some new customers, and we think that is a non-growth business perpetually for Cogent.
This is your corporate business?
No, this is the enterprise.
Okay, got it.
This is the acquired Sprint large-
Oh, I see.
enterprise customer base.
Right.
Cogent's organic business has two components. It has a corporate segment, where we sell to end users, which has historically grown at 11% year-over-year. The impact of the pandemic and the increase in vacancy in central business districts hurt that business. It went to -9% growth. It's rebounded today to +1%. It will eventually get back to being a roughly 10-11% growing business as the impact of the pandemic continues to wane. Our NetCentric business, which was in Cogent's classic business, selling bulk internet connectivity and data centers, had always been volatile. It had averaged 9% growth, but going into the pandemic, it was growing below trend line at 3%. It actually shot up to +26%. Today, it's about a 10% growing business.
The combined organic Cogent business of selling internet and VPN services, either to service providers or to mid-sized businesses, will grow at about 10%. Roughly three-quarters of that business will be on-net, and a quarter of that business will be off-net. The enterprise business, because of the global footprint and the locations that these enterprise customers need service, will be about 50% on-net, 50% off-net. Then finally, the Wavelength business, which will be entirely on-net, will carry 95% contribution margins.
Right.
It will grow very rapidly off of a low base. But then eventually, as we plateau in 25% market share, it should be growing at that 5%-7% range. So bundling these three units together, the legacy Cogent, the Wavelength business, and the acquired enterprise business, five years will be at $1.5 billion of revenue, $500 million of EBITDA, roughly 100 basis points a year of margin expansion. When we add a dollar of on-net revenue, whether it be Wavelengths or Transit, it carries $0.95 EBITDA contribution margin. When we add a dollar of off-net, whether it's an enterprise or a corporate customer, we only get $0.45 of incremental EBITDA.
Blending these three growth streams together will give us a sustainable business, generating free cash flow that will allow us to continue to accelerate the return of capital to shareholders. For those of you who've been familiar with Cogent, you know, for 45 consecutive quarters, we've grown our dividend. We today have, I guess, about a 6% dividend yield, and we're growing our dividend at about 4.5% a year. With the full integration of Sprint and the stabilization of these businesses, we will both delever and have excess cash to repatriate to shareholders.
So a number of follow-ups there. First, can you give us sort of an update on the return of the growth on the corporate side post the pandemic, which is a part of the story. Can you just give us sort of an update on where we are there and you're confident that you can get back to those historic growth rates?
So we're about halfway from where we were to where we went down to, and now rebounding back to pre-pandemic levels. You know, the progress is very geographically uneven. Companies are still struggling with what the new work environment looks like, how much is hybrid, how much is in the office, how much is virtual? I think that's becoming clearer and clearer as time passes. Vacancy rates have stopped increasing and are starting to decrease. Companies are now mandating certain number of days in the office, and most importantly, companies that had postponed architectural decisions around their network are now ready to make those decisions. So we went from negative 9% growth to positive 1%. It should continue to grow going forward. And I've given up trying to predict-
Right.
when we'll get back to that pre-pandemic rate. But I do know that in our footprint we today are still only about 20% penetrated, and the value proposition we offer our corporate customer is compelling. Price parity with our competitors, 30-60 times the throughput, an installation window that's 9 times faster, and a reliability level that's three times higher than our competitors. Now, our corporate on-net footprint only works in the very biggest buildings. We today have approximately 1,860 buildings and about 1 billion sq ft. That represents 11% of North American multi-tenant office space. But we're not going to go to that other 89% of the market, because the return on capital for those construction projects is too low.
Are you in all the big, all the major buildings that you-
Pretty much every building.
So the expansion is, it's just a question of penetrating those buildings?
It is. We do expand the footprint at a modest rate. We used to be adding 60-70 million sq ft a year. In 2023 and in 2024, we'll probably add more like 25 million sq ft. So there is still some expansion, but it's 2% of the base, not 7% of the base.
Got it. Now, you know, Dave, we've been doing this a number of years, and I always not only do I love understanding more about your guys' business, but you, you have a great perspective on the, the internet in general and how traffic is growing. And just because you sort of sit on top of it all, you see all the sources. And so how quickly is internet bandwidth growing right now? And just whether, you know, from all the streaming and, you know, the rise of mobile, and now as we look out, you know, the, the conversation, because I cover the data centers as well, the, the big conversation is AI, especially AI, combined with sort of, you know, true 5G, low latency connectivity pervasively.
Do you expect. Maybe give your sort of lay of the land on what you think the next five years looks like in terms of, you know, and related to your business as well, but the growth in sort of internet traffic driven by these issues. Is there a chance that we see acceleration? Because you usually talk about a sort of a very flat growth rate, but just would love to get your perspective.
Yeah. So I would actually describe Cogent as sitting underneath the internet
Okay
... not above it.
Right.
That we are the plumbers that allow the bits to flow. The internet has grown for over 30 years since the World Wide Web was introduced by Tim Berners-Lee and the internet was commercialized, and that was off of a very small base, but a very high compounded growth rate of about 22%-23% per year. The internet is still growing at about that pace. That is truly amazing, considering how large the internet has gotten. There are three things that drive growth: number of users, number of minutes a day of use, and the bit intensity of those minutes. Every application comes down to answering those three questions. The internet is growing faster outside of the U.S. than in. The internet is growing today primarily because of the accelerated transition from linear to streaming video.
But we will see augmented reality, virtual reality, other applications, gaming, that will drive higher bit intensity and drive growth. We will see people spend more time on the internet. You also mentioned AI. AI is possible because of the large amounts of data that have been collected by the internet. The internet made the cost of data collection very low. That created the foundation for AI. AI, first of all, is misnamed. It is not artificial intelligence, it's inference from existing data sets. So it is not cognitive thinking. That will happen, but we are very far from that. Today, we are in the training phase, where all of the data that's been collected over the past 15 or 20 years is now being run through very CPU-intensive programs to create patterns out of it.
The next phase will be the inference, taking those patterns and projecting them into future activities. It is certain that most of what people do on a daily basis is repetitive. It's patterns that they do every day. No offense to you, John, you cover companies, and you take your learning from one company and project it to the next. AI is great for that type of application, the AI that we're building now. That will be a key driver. It is certain that there will be thousands of new business models, most of which will fail.
The internet succeeded because of its openness. It has very low barrier to entry, and it gives you global reach. Two things that never existed in human history up until 25 years ago. Now, we're layering on top of it the ability to take the collective patterns of the world and project them into best practices. That is going to have a really profound effect on the economy, on every job, and it should ensure another several decades of internet traffic growth.
So you don't think as we move from training to inference, it drives acceleration? You think it just keeps us at that 20-30, 20-25?
I think that's right. The base is getting really large.
Right. Right.
Nothing grows forever, but the internet has grown for such a protracted period
Right
Simply because it keeps adapting. The beauty of the internet is twofold: its openness and its ubiquity. No other technology, not even the printing press, had those two attributes.
Right. Well, Dave, actually, that's all that we have time for. Thank you for coming, and always interesting to spend time with you.
Thanks, John, and I'll see you next week.