My name is Bora Lee. I'm a Communications Infrastructure Analyst here at RBC Capital Markets. I'd like to thank Dave Schaeffer for joining us, CEO of Cogent Communications. Thank you, Dave.
Well, thank you, Bora, for hosting. I'd like to thank investors for taking time out of their day and RBC for a great forum.
Great. So, I guess we'll just jump in. So there are a lot of moving pieces post the wireline acquisition, and I wanted to dig into that a little bit. I guess, first of all, you benefited from discontinuing negative gross margin SIP during the most recent quarter. Of the 24 products that you've identified to end of life, do you have line of sight into the pacing and magnitude of these other services and how you plan to discontinue them over the next few quarters? I'm particularly interested in any sort of chunkiness or if we should expect something a bit more gradual.
Yeah. So, the most chunky or largest of those products was what was eliminated initially, which is the voice product or Session Initiation Protocol product. It represented about 8,000 units out of a total 19,000 non-core units or products that are sold. There are a total of 24 non-core gross margin negative products. This was the largest of them, and it was the only one that was entirely usage-based with no contractual term associated with it. Now, as part of our merger with Sprint, we required T-Mobile to begin the end-of-life process for all of those products. For the SIP product, that was begun in September of 2022, and we end-of-lifed it on September 30. Even with a one-year notice period, we actually had one customer file a complaint at the FCC, stating they were not given adequate notice.
Ultimately, we prevailed and did turn off that service. The remaining services have all been end-of-lifed, and the customers have been noticed. However, we also have agreed to support each of those products for their fixed contractual term. So as a result, the remaining 23 products have an end date specific by customer. All of the non-core products will be terminated by the end of 2026.
Great. So safe to say that it's a more gradual, sporadic, end of life for the remaining contracts as they term out. Is that fair?
That is correct.
Okay, great. So Wavelength has been noted as a growth driver for the combined business. What have you learned from running the business and going to market with the service for a full quarter?
So I think there have been a few key lessons. One, the demand set is stronger than we initially had expected, and our backlog, I think, is a good indication of that. Two, we are frustrated that we are not able to provision these services as quickly as our customers would like to see them. We have been very clear with customers in terms of the number of locations and the amount of time to provision, but it has been frustrating to have demand and not be able to meet it. Third, I think we are about where we expected to be on the network conversion. And just to remind investors, in taking the Sprint network, which was originally built to carry long-distance voice, we are taking a subset of the switch sites, the telephone central offices, and converting those to data centers.
We are then extending the Sprint network from those endpoints to meet our metropolitan network. So the first phase has to be the physical interconnection of the two networks. The second phase is the reconfiguration of our metropolitan networks. So Cogent operates about 18,000 route mi of metropolitan fiber around the world. Approximately 11,000 of that is in the U.S. and about 110 markets, where we connect to approximately 800 carrier-neutral data centers. Those networks are configured as a series of rings, giving each point on the network two physical paths back to our aggregation. Because we were previously only selling IP services on that network, we co-mingled multi-tenant office buildings and data centers on the same ring. Because the market for wavelengths is exclusively in carrier-neutral data centers, we don't want those wavelengths passing through multi-tenant office buildings.
It's both harder to administer and a potential incremental point of failure. So the second phase is the re-architecting of our metropolitan rings. We will add no more fiber, and we will have no more rings at the end of this, but we will have pure either data center rings or pure multi-tenant office building rings. The third thing that has to happen is we need to deploy a transponder to accept the wavelength in each of those 800 carrier-neutral data centers. This is typically a 4 RU unit that'll go on a pre-existing rack, and it's a shelf that then you plug in pluggable optics that have very low cost to activate a wavelength. So by plugging one in at each of the two locations, the A location and the Z location, you can then build the end-to-end wavelength.
The final thing that has to happen is along the transmission path, we have to optimize the location of the reprogrammable optical add-drop multiplexers, the ROADMs, across the network. We are embarked on all four of those efforts simultaneously. It will be the end of 2024 before we can do that ubiquitously. Today, we have about 50 facilities where we can provision in a roughly 2-month or 60-day window. Another 200 facilities, that's a 3- 4 month window, and then the other 550 out of the 800 are still waiting for reconfiguration. The third lesson is there's a lot of moving parts to get the network to where we want it to be.
Does the reconfiguration of the network architecture so that it's data center specific prevent you from or at least slow down your ability to sell wavelengths in the meantime, until that's fully done, completed?
That is one of the reasons.
Okay.
Actually, all four of those discrete steps I've defined impact some sites. You need all four of them to be complete to have the wavelength services enabled and have a two-week provision.
Got it. And then you spoke about having a healthy sales pipeline for wavelengths in the most recent quarter. Can you talk about the puts and takes of the competitive nature of these routes, and as well as maybe touch on, there's a competitor in the space that is a bit pressured at the moment, if that has any impact on customers' willingness to look for alternative options and go to you?
Yeah, I'll take those in somewhat reverse order, Bora. You know, while I think investors are extremely concerned about vendors' balance sheets, I'm not sure customers are that concerned. They've been well educated by the market that a bankruptcy does not necessarily imply liquidation, but rather reorganization. So I would be surprised if a bankruptcy would take supply out of the market. Now, I think for us, it's a combination of having more endpoints than any other provider being a competitive advantage. So a customer can get a wavelength from point A to point C without having to piece together two or three different suppliers. I think that is a positive. The second positive is that all of the markets that we serve are already served by someone, so there is demand there, but the routes between those markets are unique.
So most of the current fiber that is deployed is deployed along public right-of-way, typically roadways and plastic conduit buried two feet below grade. The Sprint network was built very differently. It's a long railroad. It's buried in an armored cable six feet below the train tracks. Now, the positive is that this cable is very secure and has not been cut. The negative is it's very hard, if not impossible, to pull incremental fiber. The good news for us is that this fiber, even though it is 15 years older than some of the newer networks, has actually 50% of the frequency of cuts per mile as those newer networks, giving it better performance characteristics.
The underlying fiber itself is SMF-28, single-mode fiber 28, which is an optimal fiber for utilizing coherent optics, which are capable today of going from anywhere from 100 Gb per wave up to 3.2 Tb per wavelength. On the later fibers that were deployed, such as LEAF or TrueWave Classic, those large effective area fibers, which have dispersion compensation, typically are not capable of nearly as much throughput per fiber. The fact that we can give a customer an exact route map, KMZ file with accuracy within 1 m is unique to us. That's, in fact, done with each quote, whereas our competitors oftentimes have rolled up multiple assets and struggle to have all of their information in a coherent way that they can present that to a customer. The ability to provision quickly because of the homogeneous architecture that I've described will give us an advantage.
And then finally, and maybe most importantly, we will have a cost advantage. We have a network that we paid $1 for with no debt. Our competitors have been saddled with billions of dollars in obligations that they need to service, so therefore, don't have the latitude of pricing that we have. So I think all of those five dimensions I've described give us a great deal of confidence we'll capture an increasing amount of market share.
So a lot of this is dependent on gaining market share and increasing your penetration of the addressable market, which relies on your sales force. You're relying primarily on your NetCentric sales force, I believe. Can you talk about their go-to-market strategy and what their experience has been thus far?
So first of all, Cogent has a total of 650 quota-bearing sales reps. Those reps sit in 82 discrete teams around the world, and they sit in 48 offices. 257 of them focus on the NetCentric market. We are the largest provider of Transit services in the world. We service 1,588 carrier-neutral data centers. We have over 7,900 access networks as customers and over 5,000 content-generating Transit customers. These are the exact customer bases that will be purchasing Wavelengths. Roughly 90% of our Wavelengths will come from the NetCentric segment. Three quarters of the potential customers are already buying Transit services from Cogent. So because we have a track record of delivering, we have credibility with those customers, we can leverage that.
We believe we will replicate our 25% market share in the Wavelength market, as we've done in Transit, but we'll do it in about half the time it took us in Transit because of the sales force efficacy.
Would you consider adding additional, offering additional services on top of that, be it something like Network- as- a-S ervice or any other types of services?
So it's funny, our industry has all these buzzwords. I couldn't even describe to you, having been in this industry for over 25 years, what Network- as- a-S ervice means. When you build a physical network, you provision a certain amount of capacity on that network, and then you deliver it. You can absolutely give the customer the ability to throttle up and throttle down that usage, but that's a billing and marketing technique. It has nothing to do with the network. Either the capacity is there or it's not there. Our belief has always been, once you deploy it, you spend the capital and the labor to do it, leave it on all the time, and the customer may use it as they choose. I guess that's our flavor of Network- as- a-S ervice. There is no service. You have the network, and you use it as you want.
Okay. With transport services, you've often taken more of a cost leader type approach for pricing. How are you seeing prices currently? I noticed that there was a bit of a tick down in Wavelength ARPU. I'm not sure if that was more of a one-off due to the transaction, or if that has something to do with your marketing strategy.
So, I think the volatility in Wavelength pricing is for the next few quarters, based on the fact it's a relatively small base. Wavelengths are sold with three dimensions to their pricing grid. The longer the distance, the higher the price, different than the internet. Two, the bigger the port, the higher the price. And third, the longer the contract term. You know, our ARPUs were $1,899 in Q2 and a little below that in Q3. It will be a bit volatile. You know, the lower capacity, 10 Gb services are much less expensive. 100 Gb dominates the market today, and there is an embryonic 400 Gb market developing. I think ARPUs will stabilize around $2,000, maybe a little bit less than that, and the mix of 400 will increase, and the mix of 10 will come down.
In terms of discounting, we have discounted in select cases, but we've been more disciplined here because it is a route and location-specific service and have not taken the market-wide discounting structure that we used in Transit, but rather a more selective, kind of targeted path-by-path reduction.
Okay. So you've talked about a target growth rate of 5%-7%, but that assumes corporate staying where it is and NetCentric decelerating a bit. If we were to think about scenarios of corporate recovering, say, about halfway to pre-pandemic levels, and then secondly, NetCentric staying where it is, what sort of growth rate ranges should we think of?
So within our revenue, there are four discrete drivers of revenue growth. Our corporate business had grown at an average of 11% a year for 15 years pre-pandemic. It went to a -9% growth at the depth of the pandemic, and today is back to about a 1% year-over-year growth rate. So we think that will gradually continue to improve and get back to growth rates pre-pandemic at some point. Our NetCentric revenues actually averaged 9% going into the pandemic, were below average at a 9%, well, below that 9% number at 3%, shot up to 26%, and today have moderated to about a 10% year-over-year growth rate. Our enterprise business, which has non-core products which are bleeding off, but the core enterprise business is about 22% of revenues and expected to be completely stable.
Then finally, our Wavelength business. Our Wavelength base today is very small, so much to minimize. As that Wavelength business grows, we expect it to be a much more meaningful contributor to growth. In order to do that, we have to do that network foundational work I described earlier. When you blend these four drivers together, we're anticipating a long-term growth rate in that 5%-7% range, with approximately 100 basis points a year of margin expansion. You know, right now, we actually have elevated EBITDA margins at 47.7%. By far and away, the best in the company's history, and that's because of the transfer payments from T-Mobile to induce us to take the wireline business. Those payments will step down.
We expect our margins to go down, but kind of be in that mid-30s range as that subsidy winds down, and then because of the mix of on-net and off-net sales, be able to continuously show about 100 basis points a year of EBITDA margin expansion into the mid-40s.
I have one last question. If anyone in the audience has a question, please raise your hand. So, I guess just as a preview, what are your top strategic and operational priorities going into 2024?
You know, 2024 will be a very work intense year for Cogent, as we're executing on those four physical reconfigurations that I described in order to meet our Wavelength requirements. Secondly, we have 1.3 million sq ft and 160 MW of former switch sites that need to be converted to data centers. We need to remove all of that surplus telephone equipment that has not been in service for decades, but still is occupying space. So I would say our number one work effort will be on these physical network reconfigurations. Our top priority is always sales. You know, Cogent, prior to the merger, had roughly 600 salespeople. Quota-bearing, had 750 people in the sales organization and 1,100 total employees. When we acquired Sprint, there were 942 employees, and only 50 of them had quotas.
We went to 650, but we picked up 890 non-quota-bearing people. You know, we have been right-sizing, reorganizing, repurposing people, you know, but our number one challenge is always to sell. You know, you can build the greatest networks, you can have the most ubiquitous footprint, the best technology, but if a salesman doesn't bring in the order, you don't have a business. You know, running a company, a management team has two places to get money: investors, such as the people in this room, and customers. I can tell you, it's much better to be able to get your gross margin from your customers and not from your investors. They eventually don't like giving you money unless you can show a return, and I think that's been part of the frustration.
In telecom over the years, the management teams have sat on stages like this, promised returns and not delivered them. You know, the fact that Cogent has delivered consistently 45 sequential quarters of dividend growth, coupled with shrinking our share float by 20%, kind of, I think, validates our commitment to rational allocation of capital.
Okay. I think that takes us to time. Unless there's a question from the audience, I think we'll end it there.
Thank you very much, Bora.
Thank you, Dave.