Cogent Communications Holdings, Inc. (CCOI) Earnings Call Transcript & Summary

So the Sprint network was comprised of 19,000 route miles of intercity fiber, 1,200 route miles of metropolitan fiber. And this was all owned fiber. In addition to that, there was nearly 4,600 route miles of leased fiber. We are eliminating that leased fiber as it is not necessary for Cogent's network services. We have identified 800 carrier-neutral data centers where we intend to offer wavelength services with any-to-any footprint that we will provision a wavelength on an average of 2 weeks. We are taking a network that was dormant and originally designed for voice traffic and repurposing it. Our initial thought was that we would be able to sell large data center to large data center connectivity. And while we have sold some of that, the vast majority of our backlog and initial orders came from smaller data centers to larger data centers. We can provision wavelength services at the end of Q1 in 420 of those 800 facilities but with an extended provisioning window. We have a number of foundational network upgrades and modifications that need to be completed to hit both the ubiquity of our coverage and the speed of installation that we have outlined. I think investors will see an increase in wavelength installs throughout the year, but it really will be at the end of the year when we will be able to hit full cadence and be able to deliver those services across all 800 locations with a shorter provisioning window. The foundational steps that we have to undertake, first included the connection of the Sprint network to the Cogent metropolitan network in approximately 100 markets. That work is completed at this time. The second thing it required was the installation of optical transponder shelves to accept wavelengths, a 4RU shelf and each of those 800 carrier-neutrals we were at 420 at quarter end. Third, it requires the deployment of reconfigurable add-drop multiplexers at the intersection points of the long-haul and metropolitan network. We're about 50% of the way complete on that effort. And then finally, the most challenging of the efforts is the reconfiguration of approximately 14,000 route miles of metropolitan network that Cogent operates. That network is comprised of about 800 physical rings that supported 2,650 buildings with multi-tenant office buildings and data centers co-mingled on each of those rings. We are going through a process of reconfiguring those rings, segregating the multi-tenant office footprint of nearly 1 billion square feet from the carrier-neutral data center footprint. This is the most labor-intensive portion of this reconfiguration. And we're about 40% of the way through that work effort. That work effort includes touching each and every one of these buildings, typically in a planned maintenance with multiple service impacting outages for customers, so they need to be coordinated. That effort is going according to plan, and we feel comfortable that we will have that reconfiguration work done by year-end, along with the completion of the 2 other steps of transponder deployment and ROADM deployment that would then give Cogent an architectural advantage over our competitors. So in looking at that $2 billion wavelength addressable market, Cogent will have 4 discrete advantages, more endpoints, unique routes, faster provisioning times due to this architectural distinction and configuration that is different than the way other wavelength providers deliver their service. And then maybe most importantly, we have no cost basis in this asset. It's an asset that we acquired for #1. Remember, the acquisition of Sprint from T-Mobile was really to somewhat decoupled acquisitions. The transfer of the network assets to Cogent for a $1. And then the second, the assumption of that enterprise customer base with a cash subsidy from T-Mobile of $700 million paid to Cogent over a 54-month period.

So as we see, you'll have the architectural advantage. How should we think about Cogent, I guess, go to market at that point? And how you're thinking about pricing?

So the wavelength market has four distinct customer bases. Three of those four are serviced by our NetCentric sales force. This is a sales force of 284 quota-bearing individuals that have assisted Cogent in becoming the largest provider of transit services in the world. The four segments are content-generating businesses who are looking to replicate content between data centers. The second segment is access networks, whether global or domestic, that are looking to link islands of traffic. Both of these segments are today serviced by Cogent's NetCentric sales team. The third segment is new and emerging, and that is AI training using data centers that are distributed for their training models and may not be coresident with the databases that are being used for that training. That also is a market segment that is focused on by our NetCentric sales force. And the final and smallest segment would be government and large enterprise, which is covered by our enterprise sales team. As we think about the competitive dynamic, we have these multidimensional advantages that I think will allow Cogent to continue to gain market share. Oftentimes in our NetCentric transit business, customers associate Cogent with low price, but it was also our more expeditious provisioning that allowed us to become the largest transit provider in the world.

Now shifting gears, the IPv4, I think, is interesting in new opportunity that some investors had not necessarily appreciated. On the call, you noted that Cogent's current portfolio of leased Internet addresses currently generates $3.4 million of revenue per month and has been growing at 2% to 3% monthly over the last 1.5 years. IPv4 is an opportunity that, I guess, should be pretty interesting over time, but it raises the question, perhaps why was IPv4 not a bigger focus for Cogent in the past?

Yes. So let's maybe go back in history a little bit to the initial architecture of the Internet. Three basic protocols define what is the Internet today. The first is TCP/IP, the way in which two devices communicate with one another. The second key protocol is BGP, the way in which 2 networks interconnect. To support both of these protocols, it is necessary to have a unique numbering scheme. When the Internet was initially designed, first is the DARPANET, then as the ARPANET, it was decided that there would be 2^32 unique hexadecimal addresses. These are IPv4 addresses. There are 4.3 billion addresses possible. The U.S. government initially controlled all of those addresses. While they retained approximately 800 million addresses for their own needs, they made 3.5 billion addresses available to support the public Internet. . Cogent owns about 1% of that pool or approximately 37.8 million addresses. Addresses were readily available for free from 1991 with the initial commercialization of the Internet till 2011. At that point, addresses began to become rationed. In 2015, Cogent was unique among service providers, and it began leasing out its address space at an average price of about $0.30 per address per month, but only to those customers that purchased bandwidth from us. In midyear '22, Cogent relaxed that restriction, saw a material increase in our leasing rates. And today, we are continuing to grow that business at between 2% and 3% sequentially per month. We also have taken some additional steps to recognize value out of these addresses. One is we securitize a portion of our leased address space and raised $206 million through an asset-backed securitization. The second is we increased the prices for new address leasing starting about 6 weeks ago and continue to see unimpacted demand in terms of new leases. We have leased only about 30% of our total inventory. This is an area that Cogent will focus on more going forward.

How big could this opportunity be as you think about the pricing umbrella relative to AWS, Azure as well as the unutilized inventory?

So the opportunity has two dimensions to it. There's a possibility we could sell some of our unleased addresses. The market price for those addresses is between $50 and $60 per address. The second would be to lease out the remainder of the address pool, the approximately 24 million unleased addresses and to continue to increase pricing on those address leases. We got an additional boost in our leasing business when Amazon and Microsoft began leasing addresses in 2023 at a price that was 12x Cogent's pricing. That pricing umbrella gave us the leeway to raise prices. We are initially increasing prices by 50% on new sales. We will then revisit the installed base and may then implement additional increases on both new and existing customers. While it is unlikely we will ever be able to achieve the average of $3.60 an address and fully lease out our entire inventory, I think there's significant room to grow this business, both in terms of increasing the leased inventory, tripling it and then maybe continuing to increase prices.

That's great. And so shifting to the NetCentric business. This was positively impacted by the pandemic as demand for connectivity rose sharply. However, growth in that segment has started to moderate. Do you still expect this business to grow at 10% per annum long term? And how should we think about pricing and volume dynamics there?

Yes. So in our NetCentric business, we are selling bulk Internet connectivity in those 1,680 data centers around the world. That business for the past 20 years has grown at an average of 9% a year. Heading into the pandemic, we were actually growing our NetCentric business substantially below long-term trend line at only 3% year-over-year. With the onset of the pandemic, that business materially accelerated and reached a peak growth rate in the first year of the pandemic of 26% year-over-year. That surge in growth has moderated, but stabilized at around 10% year-over-year. So just slightly above the long-term average. That business will probably grow faster than that because of the inclusion of wavelength sales in our NetCentric revenue classification, and the fact that 85% of our IPv4 leasing is NetCentric with only 14% of it being corporate and 1% being enterprise.

Conversely, Cogent's corporate business was negatively impacted by the pandemic and has not quite returned to prior growth rates. As you think about the macro backdrop and the pivot to hybrid work models, can the Corporate segment get back to sustainable growth levels over time? What are you hearing currently from your customers?

I think you're being too kind in your question. Our corporate business has disappointed. It has grown historically pre-pandemic at 11% year-over-year. When the pandemic hit, the growth rate in that business decelerated to negative 9% year-over-year. It has reaccelerated to between 3% and 4% annual growth. Many companies have directed employees back to the office. I believe your boss has been particularly vocal on that topic. But my guess is, if I went to 383 Madison, the attendance would not be at pre-pandemic levels. If we look across the entire footprint of multi-tenant office space, if we take security card entry rates at pre-pandemic levels being 100, we're today at about 62% of that. So 38% of the employee days are not in the office. That problem is probably even a little more acute since the office employee base has probably grown 4% in that 4-year period. So we're really looking at about a 40% reduction. That reduction in number of days in the office has caused many companies to rethink their IT infrastructure. Now what we have seen is many companies are now settling on a more permanent infrastructure, and they are now implementing architectural changes that they may have pushed out until they understood their exact real estate requirements. That, I think, has been helpful in increasing our corporate business. The return to office has been uneven across the country. We continue to see weakness in the Pacific Northwest and Northern California. We see particular strength in Florida and Texas. Cities such as Boston and New York are probably more in the middle of that pack. But I think there is a gradual return to office that is underway. There's also companies reimplementing new architectures that they had procrastinated on deploying. For those two reasons, I think we'll continue to see a gradual improvement in our corporate business, but given up on when I'm comfortable in saying we'll return to the double-digit growth that we saw pre-pandemic. What I do know is that the addressable market is still there to support that growth rate.

And so being mindful of time here, just wanted to make sure that we touched on capital returns and the delevering story at Cogent because we have gotten some questions and concerns around the company's leverage profile and the ability to fund the dividend. While historically, Cogent has funded the dividend through debt. We covered a lot today, so there's a lot going on in the business between the Sprint integration, standing up some longer-term opportunities, the Corporate segment you just kind of talked about as well in the path to recovery there. How does the -- how do all those factors or all the different moving pieces within the business factor into your view about the appropriate level of leverage for the business and the company's payout ratio over the next several years?

So we actually have materially delevered in the past year with our net leverage going from 4.7x EBITDA down to 3.17x, and we're going to continue to delever for the next several quarters. Cogent was very fortunate that it built its business without debt. We have used our balance sheet and its underlevered status to return increasing amounts of capital to shareholders. We have implemented a dividend in 2010 and grown that dividend for 47 sequential quarters. We intend to be able to continue to do that going forward. We also had retired approximately 22% of our outstanding shares. Now this was done with a combination of internally generated free cash flow plus the addition of leverage. I think as our payment subsidies from T-Mobile drop off, our leverage will increase in the short term as we are continuing to take costs out of the acquired enterprise business and ramping the high-margin business associated with our wavelength sales. We have said that 5 years post-closing, so that is May of '28, the company will be on a revenue run rate of $500 million of EBITDA with $1.5 billion of aggregate revenue. Our EBITDA today is about $350 million annually, while I know that Q1's rate of 115 was substantially higher, that will step down in the latter part of the year due to the step down in those subsidy payments from T-Mobile. But I think with the combination of the securitization, the cash flow generation of the business, the assets that we have available for sale and our borrowing capabilities, we feel quite comfortable that we'll be continuing to increase the capital that we're returning to shareholders.

Right on time. Dave, I think it's a great place to end it. Thanks for joining us. Thanks, everybody.

Thanks, Sebastiano.