Good morning. Yes, still good morning. I'm John Hodulik, the telecom and media analyst here at UBS, and I'm very pleased to have Dave Schaeffer, the founder and CEO of Cogent Communications, with us. Dave, thanks for being here.
Hey, John. Thank you for hosting me. Thank you, UBS, for a great venue, investors for their time. I was joking with John, this is, I think, the first time in 20 years I've been in the big room.
Exactly. You guys have been making a lot of noise. Thought it was time. Before we dig into everything, Dave, why don't we start off with, you know, if you could maybe give us a quick review of 2022, but just talk about what the priorities are for Cogent as you look out into 2023.
Yeah. Cogent, I think, has two significant priorities moving forward, both of which are a result of things that we experienced in 2022. In our classic business, it is the return to office and the re-acceleration of growth in our corporate segment. Just to remind investors, Cogent is roughly 57% of our revenues come from selling to corporate end users who are located in skyscrapers in the central business districts of major North American cities. That business was severely impacted by COVID. It had grown at a consistent 11% year-over-year for 15 years, only to see COVID drive that growth rate down to -8% at the worst. We've returned back to positive growth, but it's sub 1% today. It is beginning to accelerate and in its improvement.
The second segment of our business is our NetCentric business. That had grown at an average of 9% year-over-year. All these numbers are organic. Going into the pandemic, it was actually growing below trend line at 3%. It accelerated all the way up to 25% year-over-year growth as the migration of linear to streaming accelerated, and we are the primary carrier of that streaming traffic. Last quarter, and for the last couple of quarters, we've had about 16% year-over-year growth. The combination of these two factors is Cogent's total growth rate had decelerated from 10% down to 5%, and our margin expansion declined from 200 basis points a year to 100 basis points.
That is beginning to re-accelerate. We think in 2023 and beyond, that will continue to improve and get back to the 10% growth rate for Cogent's classic business and 200 basis points of margin expansion. The second big initiative was announced in August of 2022. That is our acquisition of Sprint Global Markets Group from T-Mobile. This is the Sprint wireline business. We are acquiring that business. At the point we announced the deal, the company had $560 million of revenues. The business was burning about $300 million of EBITDA and declining. We identified a number of products that were gross margin negative that are being end of lifed. We also are being paid $700 million in cash by T-Mobile to acquire this business and turn it around.
A significant focus of 2023 is the closing of that transaction, the continued reduction in burn rate, and maybe most importantly, the repurposing of the Sprint network. Just to remind investors, Sprint's network was the first nationwide fiber optic network built in the mid-eighties. It was built to carry long distance voice with SMF-28 fiber, 19,000 route miles of inner-city fiber connected to 1,300 miles of metropolitan fiber and 1,300 buildings along the way. We are going to repurpose that network to carry Internet traffic. We will migrate Cogent's traffic onto one pair of fibers. We will take an additional pair of fibers and use those to sell wavelength services or optical networking services, and then we will be selling off dark fiber.
The upside for Cogent is to grow the wavelength business and sell off the surplus inventory while being paid by T-Mobile to fix the business we're acquiring. 2023 is shaping up to be a pretty busy year for us, John.
It really is. You've set the stage well for the, for the rest of the presentation going through each of these segments. Why don't we start off with the, with the Sprint acquisition? I mean, I would say this acquisition is sort of the latest in a long line of acquisitions that you've made for very you know, very economical acquisitions, right? Cogent itself is a, is an amalgamation of a bunch of companies that frankly going back 20 years I used to cover. Actually, Sprint is no different in their GMG group. I guess if you could, you know, maybe it's at a very high level, what did you see in? Is it sort of similar to what you've seen in other acquisitions? What was it about GMG that you saw that you thought this is a made such a great opportunity?
Yeah. Cogent was founded to build an all- IP- over- DWDM network, protected at Layer 3 using Ethernet, but focusing on either skyscrapers or data centers, the business that we're executing today. We were lucky enough to have raised $500 million in 1999 before the telecom meltdown, and then we took that cash, and we bought 13 companies. Six were public, seven were private. We paid $60 million. We acquired $115 million of cash, $400 million of preferences, and $500 million of debt. We acquired companies that had raised $14 billion of invested capital and deployed $4 billion in property, plant, and equipment. We dismantled those companies by firing the customers, the employees, and repurposing those assets. We emerged in 2005 with the world's largest IP network and almost no revenue.
We have organically grown and not done an acquisition for 17 years. When we were presented the opportunity to buy the Sprint asset, we saw it as an asset that was being underutilized and could be repurposed. We saw really two things. One, a large enterprise customer base, very different than the types of customers Cogent sells to. We could stabilize that $450 million of revenue by delivering 10 times the bandwidth to those customers, bringing them onto Cogent's 18,000 route miles of metropolitan fiber, and by modernizing their VPN architecture from legacy MPLS to more modern VPLS architecture. One of the key attributes was a new customer segment. The second and more exciting part of the acquisition was the physical network and the realization that it could be a value add repurposing.
Much like finding an office building that's vacant and converting it into residential, we're taking a network that was built originally for long distance voice and now modernizing it to carry internet traffic. The transaction is being de-risked by the $700 million cash payment from T-Mobile, allowing us the capital necessary to stop the burn and physically interconnect the network and repurpose the asset. We see a lot of potential with a asset that was basically under managed for 20 years.
If you look at all the acquisitions that you've done, where does this fit in terms of, like, your excitement, you know, your potential value add or value creation along with this? 'Cause, like you said, you've done a bunch. Again, a lot, a couple of them done at sort of similar terms, where you're actually getting cash in the door to take on these assets. I mean, just what's the opportunity set here versus what you've seen in the past?
Well, this is by far and away the largest acquisition. you know, we're being paid $700 million, not $115 million. We're acquiring an asset that had deployed over $20 billion in PPA initially, had a market cap at peak of $129 billion, and so the scale is different. I think, it's also an asset that better aligns with Cogent's business today in the sense that we are able to reduce our dependence on third-party fiber. We're able to reduce our maintenance expenses by shifting onto the network. We're also able to add one additional product that's adjacent to our Internet Transit business, sell to the same customers, and do that with little or no marketing costs or risks. Cogent today, prior to the acquisition, is about 1,050 employees.
750 of those are in sales. We can take that sales force and focus on this asset and create some value very quickly. We're pretty excited about the opportunity.
Maybe we'll talk about each of these sort of segments. First the wavelength business. What's the TAM of that business? Who are your competitors? You said it's an adjacency to your current business on the corporate side, I would imagine. Can you just talk about how? Is that sold together? Do they compete with each other or do companies buy both?
The primary market for optical network transport or wavelengths are other service providers. They would be adjacent to the transit market segment that Cogent's NetCentric sales force is selling into. Of our 750 salespeople, 550 are quota-bearing, 225 or roughly half of the sales force, a little less than half, is focused on the NetCentric market. The buyers for wavelength services are two major groups. There are regional access networks that buy optical transport to link those markets together. A cable company that has disparate markets, a telephone company, or a regional ISP. The second major market segment are the hyperscalers. They would use wavelengths to link their data centers together to do data center to data center data replication.
If you were, say, Facebook with a large data center in Iowa and another facility in North Carolina, you want to replicate that data between those two facilities for both better performance and disaster recovery purposes. While you could do that over the public internet, having a dedicated wavelength network is better because the internet, while being the lowest cost per bit, is not well suited for very large data transfers. The bits are split up into smaller packets and sent over different routes, whereas by owning a wavelength, you can dedicate that wavelength entirely to large file transfers between those locations. Today, the market is about $2 billion. It is relatively static. There are third-party studies that are indicating the market is growing, but it is, I think, realistic for us to expect it stays flat at $2 billion. It is dominated by Lumen.
They have nearly 90% market share. Zayo, the second-largest player, about 7% market share. Windstream and Uniti, each about 1%, and about 1% with some small regional carriers. Because we have a negative cost basis in the asset, because we can tie the Sprint intercity network together with the Cogent metropolitan network, we can now access 800 North American carrier-neutral data centers. That's actually more than any other provider. We have, I think, three key advantages that should help us win market share. One, the ability to price at a price lower than the market. Two, over 90% of the routes are physically unique, meaning there's no other carrier sharing that right of way, which has a significant positive attribute for our customers for diversity.
Third, the Sprint network was hampered by the fact that it terminated in only 23 carrier-neutral data centers nationwide. It made selling wavelengths very difficult because you would have to get from a proprietary Sprint data center to a major aggregation point like an Equinix, and the cost of those metropolitan connections would be prohibitive, both in terms of complexity and cost. By physically interconnecting the Sprint intercity network to the Cogent intracity network, we can immediately offer those wavelength products in 800 data centers. That's actually 200 more than our closest competitor, far more than Lumen. By being able to meet customers' needs in a greater number of locations at a lower price point, we think that our market share expectations are quite reasonable.
Got it. Out of the $2 billion market for wavelength services, what percent comes out of data centers? I mean, is it a heavily data center driven or is it the entire market driven?
The entire market-
The entire market.
comes out of data centers.
What percentage is with the hyperscalers? You said that was a small segment.
It's about equally split between access networks and hyperscalers. The good news is the hyperscalers are a smaller number of customers. There's five or six customers that account for half of the market, on the access side, there's probably a couple of dozen customers that account for the other segment of the market.
Got it.
In terms of customer count, it's a relatively concentrated market.
Got it. What about dark, y ou also mentioned the dark fiber opportunity. Can you give us the sort of, you know, outlines of that market?
We absolutely intend to sell dark fiber. The Sprint network has some really positive attributes. One, the right of ways are unique. Two, the fiber was installed in armor, not in plastic conduit, and buried deeply six feet below grade versus a typical conduit network that's two feet below grade, meaning the network has sustained far fewer cuts over the years. When we look at the span loss budgets of the Sprint network and compare that to, say, the Lumen network that we're today utilizing, even though the Lumen network is 15 years newer, it actually has worse span attributes. And then on the negative side, it's very hard to upgrade or add additional fibers. In looking at the Sprint network, depending on the sitting pairs, the cross-sectional density ranges from 24 to 144 fibers.
We will be utilizing four of those fibers, two to run the Cogent IP network that will also subsume the Sprint lit services network. Today, Cogent carries a little over one exabyte of traffic daily. The Sprint network, both for its MPLS VPN services and its DIA retail business, carries about 10 PB, or about 1% of the traffic that Cogent is carrying. On day one, we will end up migrating that Sprint traffic onto Cogent's existing network. We will take that pair of fibers that was used by Sprint for its IP network and use that to sell wavelength services. We will light a second pair of fibers on the Sprint network and eventually migrate Cogent's IP network onto that. That leaves anywhere from 20 to 140 fibers available for sale. We will sell those on a market-by-market basis.
Unfortunately, there's not great third-party data to size that market, and it's very route specific, but because, again, we have no basis in this asset, we're gonna do whatever it takes to clear the market to generate cash.
Yeah, 100%. You'd, again, imagine that the hyperscalers would be potential customers for that kind of infrastructure.
Yeah. I actually took the opportunity to go to Los Angeles in October for NANOG, North American Network Operators' Group. This is the primary engineering conference. It moves around. It's held four times a year for engineers. It's actually one of the few conferences that don't have investors or bankers at them. It's purely an engineering conference. I had the opportunity to sit down with about 12 of my key customers, their engineering teams, and I was really pleasantly surprised that, you know, whether it be cable operators or hyperscalers, they came to the meetings doing their homework. They had evaluated Sprint network, and they had specific shopping lists, certain routes they wanted wavelength, certain routes they wanted dark fiber. While we were interested in selling both, we could not take orders until the deal receives all of the necessary regulatory approvals.
We were pretty encouraged that there's pretty deep demand for both products.
What is the timing and the cost behind all this thing? You talked about the terms of the deal, but you talked about modernizing the network. Is that covered by the $700 million, or what's the sort of capital outline? Second of all, you when you said sort of day one, we connect the networks, get it all integrated, and we go to market with this strategy laid out. When do you start sort of selling services on this new network?
We have already begun the process of physically tying the networks together. As part of the purchase, we concurrently entered into a series of reciprocal commercial agreements between ourselves and T-Mobile, where either party can buy services or colocation, and we have building entry rights into their facilities. We will spend about $50 million one time to tie the networks together in every major North American market. Most of that expense is for the physical construction of fiber from the closest point in the Cogent network to the closest point in the Sprint network, which is traditionally at a Sprint-owned facility. Going forward, we will spend about $30 million a year on CapEx for the Sprint network. That $30 million expenditure will be to maintain and grow the wavelength business.
We anticipate that most of the $700 million that we're receiving will actually go for operational restructuring. The day the deal was announced, the business was spending about $30 million in CapEx and - $300 million in EBITDA. Because we were able to identify 24-28 product lines at Sprint as gross margin negative and unprofitable, mainly due to the complexity of the products and the small scale, T-Mobile has agreed to end the life of those products, sent notices to customers. We anticipate about 200 of the 1,400 customers will eventually go away since they derive the majority of their services from these end-of-life products. The remaining 1,200 customers, however, fall into the four products that we'll support going forward. That being dedicated Internet access, VPNs, co-location, and wavelength services.
Most of the burn will be eliminated in the first year. Between signing and closing, we anticipate the negative EBITDA to decline to $180 million in the first year, by the second year to be down to somewhere between $70 million and $80 million, and by year three, down to zero. Concurrent with that reduction in burn will be the ramp-up in the wavelength business, and we anticipate it'll take us between five and seven years to reach 25% market share, or about $500 million. That product is entirely on net and therefore will carry 95% incremental EBITDA margins. After that initial spurt in growth in gaining market share, we think that it's reasonable that we can continue to gain share, but at a more moderate, you know, kind of 5%-7% a year.
For the first five to seven years, we will be rapidly taking share.
Yeah. Maybe let's turn to, you know, the, the sort of traditional Cogent corporate business. Obviously, you saw some major challenges that you outlined during the pandemic. you know, what gives you confidence the business can revert back to the, to the historical levels and, and sort of what's the pathway to get there?
The vacancy rates and the footprint that we serve went from 6% pre-pandemic to a peak of 18%. It has started to have net positive absorption and is down to 17.7%. It will eventually get back to that 6% vacancy rate. What we have seen in previous economic downturns is that office rents fall in Class A buildings and tenants that would traditionally be in Class B and C buildings migrate upmarket. The exact same thing is happening now. It is happening more slowly than we would have expected, but we are seeing an increased level in new lease activity. These new leases tend to be for about 20% less floor space and tend to be shorter in duration. The second leading indicator of improvement in that corporate business are building entries as measured by security badge swipes.
We went from going down to only 2% at the trough of the pandemic to nationwide, we're back just under 50% in the high 40s, but it's very geographically uneven. In the best markets in the South and Southwest, we're back to 90% pre-pandemic levels. In the worst markets, San Francisco, Seattle, we're still sub 30%. New York's pretty much in the middle. We're at about 52%-53% employee days back in the office. We are seeing each month gradual improvement. The second thing we're seeing is companies begin to be willing to re-architect their networks. I think the pandemic had a lot of lessons that companies learned.
One, they're switching their designs from a model where 97% of employee days were in the office, 3% remote, to the new design standard being 60% in office, 40% remote, therefore a hybrid model. Companies are increasingly looking for secondary or tertiary locations for those ad hoc VPN aggregation points in data centers. That's an incremental revenue opportunity for Cogent. I think many companies are now replacing their MPLS networks that they've kind of postponed those replacements for two years during the pandemic. On the negative side, what we are seeing is companies continuing to groom branch locations, eliminating some of those satellite offices. I know, for example, UBS has, you know, its primary offices here downtown, but a big facility in Stamford. My guess is that there's probably a move on to consolidate those.
We see that across all of our customers. You know, when you think about an office in Stamford, it was there, one, to ease people's commutes, and two, lower the office expense. If office rents are coming down CBD, and you're allowing employees to work remote two days a week, the idea of consolidating to one location within an MSA makes a lot more sense.
It does. We're definitely seeing that. Putting it all together, you know, you talk through a lot of different trends there, and you've got some that are definitely more secular, some that are, you know, maybe more macro-focused. How does it all sort of come together in terms of, you know, for the sort of Cogent corporate business? At the end of the day, does it mean sort of, I guess, fewer connections, but, you know, I don't know. I'm thinking about it from a profitability standpoint.
Yeah. Connection sizes have increased materially as companies actually need more bandwidth to support cloud applications and I/O from their primary locations. The corporate business will return to being a low double-digit grower over the next few years. At the same time, our NetCentric business, which is 43% of revenues and has grown at this outpaced rate during the pandemic, will probably slow and return to that kind of 9% average that we've experienced. In totality, Cogent's classic business should be growing at about 10%, and with 75% of sales being on-net, 25% off-net, we can experience about 200 basis points of margin expansion. Within the acquired Sprint business, the large enterprise business is probably not going to grow materially, at best 1%, and it'll be a lower margin business.
Now, we will migrate the 93% of traffic that's off-net to as much on-net as possible. Probably on a blended basis, the combined company will be about 60% on-net and 40% off-net. We also know that the wavelength business will grow very rapidly, you know, 40%, 50% a year for several years, but that's not sustainable as we gain market share and should settle into being kind of that 5%-7% growth rate. In packaging these three segments together, the combined company within five years will be $1.5 billion in revenues, EBITDA margins in the, you know, low to mid-30s%, down from the 39.4% that Cogent delivered last quarter, but expanding about 100 basis points a year.
In looking at the totality of the business, we will be at about $500 million of EBITDA, or better than $10 a share in EBITDA. Our combined CapEx should be in the order of about $65 million, about $30 million on the acquired Sprint network and about $35 million for Cogent's classic business. If we continue to expand into new markets, that will be incremental capital, but that's been slowing down. Cogent's been spending today about $30 million on new market expansion. We put all these pieces together with the fact that Cogent has not issued equity, in fact, has bought back equity since being public, we should be in a position to be producing better than $9 a share of free cash, up from the $2 a share that we're producing today.
That's a lot of growth. That's a good segue to capital allocation. You know, you're above the high end of your, of your leverage target, and the board recently slowed the pace of dividend increases. How should we think of sort of use of cash over the next several years?
Cogent was very fortunate that it built its network without debt. We realized that we had a latent asset on our balance sheet that was under levered, and as interest rates came down, we added leverage and are actually paying out a dividend greater than free cash flow that's generated. Our dividend right now is $0.915 a share per quarter. We've grown our dividend sequentially for 41 consecutive quarters. The growth rate in the dividend had been around 15%, mirroring the growth rate in our free cash flow. During the pandemic, the growth rate in our cash flow decelerated to about 6% year-over-year. And we had kept the dividend growth rate at $0.025 sequentially, or about 13%.
Last quarter, we made the decision to slow the growth rate and the dividend to a penny a share sequentially or about 4.5% annually. That will allow us to gradually de-lever. Cogent today is about 3.8 times levered. Our stated goal is to be between 2.5x- 3.5x levered. We will naturally get back to that lower leverage rate by growing the dividend more modestly. We'll actually cheat and get there more quickly with the $700 million cash payment from T-Mobile, which is all basically EBITDA. We will artificially, during that period, be below our guidance range. We will be using some of that free cash to reinvest in the network and the sales force to help us sell wavelengths.
Takeaway, we anticipate the entire business stabilizing, reevaluating our dividend growth rate, getting below our targeted range, and then probably re-accelerating our return of capital. We've been agnostic to using both dividends or buybacks. Cogent has also bought back 22% of its outstanding shares in the open market, along with the growth in the dividend. We've also been fortunate that the majority of our dividend for the past five years has been treated as return of capital, so it's been tax advantageous to use the dividend for our shareholders.
Right. Dave, obviously a lot of moving parts to this story, and the Sprint GMG acquisition brings another dimension and potential for a big change in scale. I mean, I guess my last question is this: as you look out, you know, over the next couple of years, what are both the sort of biggest risks to the story that you just outlined, and then where do you think the biggest opportunities are?
On the risk side, I would say it's tactical execution. You know, hiring a sales force, managing them, training them, promoting them, retaining them is a challenge. We've done it. We will continue to do it, but it's not easy. I don't see a lot of network integration risk simply because we've been there and done that so many times in the past. While the scale is larger here, we also have much more experience and much more scale. I think the real opportunity is taking that underutilized fallow asset that's been ignored for 20 years by Sprint and now T-Mobile and trying to figure out how to monetize it. The 1.3 million sq ft of technical space, for example, is data center space that, you know, there's today 150 megawatts of power.
There are 47 facilities of scale that we can quickly convert from switching centers to data centers. There's a lot of incremental opportunity. All in all, I feel more optimistic about Cogent today than I have at least for the past four or five years.
You've had a lot of success with similar acquisitions and doing the same thing with similar assets in the past. We'll see how this plays out. Thanks for being with us, Dave. We appreciate your time.
Hey, thanks, John. Thanks everyone for their time.