Great. Thank you very much. Appreciate everybody being here. My name is Frank Louthan. I'm the senior wireline analyst here at Raymond James. We're very pleased to have Dave Schaeffer, CEO of Cogent, back here with us again at the conference. We'll jump right into it, Dave. You know, maybe... Oops. Have the right sheet of questions, things will go a lot better.
Long list, Frank.
Yeah, yeah, you probably don't want me asking questions about fiber to the home, but.
Well, you can ask me.
All right, so just talk to us a little bit so you made a lot of changes with the business. I think one of the more interesting aspects is what you're doing with the wavelength product. Talk to us about wavelengths and, you know, how, how, you know, what does it take? What are some of the challenges to getting that product available to the market in the near term?
Okay. So first of all, I'd like to thank you for inviting me. I'd like to thank investors for taking time out of their day, and Raymond James for providing us this great venue. So Cogent started 23 years ago with the premise that the internet would be the only network that would matter, and eventually, all traffic would migrate to the internet. That thesis has proven generally correct. However, there is a parallel market that Cogent has historically not participated in, selling optical transport services or wavelength services, as kind of the nickname for those products. So when you buy internet connectivity, a provider gives you a port, and it's that provider's responsibility to get that bit anywhere in the world connected to any other network. That is done typically through a Layer 3 routed network.
The internet is based on the principle that the packet is sent to the next available router without the concept of an end-to-end physical circuit or connectivity. The cost to move a bit over the internet is about two and a half times cheaper than it is using a wavelength, so it's a logical question: Why would anyone want a wavelength product? Wavelength product is deterministic, meaning you know the starting point and the ending point of the service. It has exact latency characteristics. You know exactly how long it takes to get the packet from one end to the other, and there are no packet size restrictions as there are on the public internet. And finally, it's completely isolated from the public internet, and therefore, more secure. The universal buyers for that service is relatively small.
The wavelength market and dollar value is actually greater than the internet transit market. The challenges for Cogent in entering that market are repurposing the Sprint long distance telephone network to convert it into an optical transmission network or wavelength network. So Cogent had built its IP network out of IRUs secured from 315 different suppliers around the world. Typically, one pair of fibers running IP directly over DWDM, protected at Layer Three, and using Ethernet as an interface. That network is comprised of 61,000 route miles of innercity fiber and 18,600 route miles of metropolitan fiber, connected to 1,600 data centers in 55 countries, and 1,860 skyscrapers in the U.S. and Canada, representing over a billion sq ft.
We are the largest carrier of internet traffic in the world, carrying about 25% of global traffic on that network. We had the opportunity to acquire the Sprint Global Markets business from T-Mobile. In the acquisition of that business, there were really two distinct segments. The first being a enterprise sale of managed services and transport services to large enterprise customers. The second being a completely fallow network that was originally constructed to carry voice traffic. We are repurposing that network. The four distinct challenges that we face are: one, physically connecting that network to our metro footprint. So the 19,000 route miles of innercity fiber and 1,200 route miles of metropolitan fiber owned in the Sprint network did not connect to many carrier-neutral data centers, where the demand for wavelength services exist....
The second challenge is reconfiguring our metropolitan networks to optimize those for delivering wavelength services. That means segregating multi-tenant office buildings from the carrier-neutral data centers. When we were selling just IP services, it was perfectly reasonable to commingle the two types of buildings. When we are optimizing the network to sell wavelengths, those will only be provisioned in carrier-neutral data centers. Having those wavelengths go through multi-tenant office buildings has two negatives. One, it injects additional loss into the span, which degrades the quality of service. Two, there's an additional potential point of failure with that multi-tenant office building in the path. So we are reconfiguring our U.S. network inside of metro markets to segregate multi-tenant office buildings from carrier-neutral data centers. This does not require additional route miles of fiber, but just the reconfiguration of the fiber that we have.
The third effort is to deploy reconfigurable optical add-drop multiplexers at the endpoints on the Sprint backbone. So the Sprint backbone oftentimes had routes coming from multiple directions to a tandem switch site 15 or 20 miles from downtown. At that switch site, we put a ROADM there to allow for the correct direction of those wavelengths to all of the carrier-neutral data centers on a given metro footprint. And then finally, in each of 800 carrier-neutral data centers, we need to deploy a transponder shelf to accept the wavelength services. The industry is defined by wavelength provisioning windows of about 90-180 days. The high-cap transit market was similar in 2021 when Cogent began selling, and we introduced a product that we could provision with a guarantee of 17 days and an average installation time of nine days.
That, coupled with pricing, allowed us to gain market share. We intend to do the same thing with wavelengths. So, our architecture will allow us to provision a wavelength in less than 17 days with a unique route and more endpoints than any other provider. It is our belief that when we accomplish these foundational modifications to the Sprint network, we should be in a position to garner 25% of the wavelength market, just as we have done in the transit market.
All right. As always, a very thorough explanation. So this is a great opportunity for you, but clearly, you've laid out a big logistical challenge on the front end. How long will this take to get all this in place? And then, what could accelerate that, and then, what are some of the bigger challenges in actually, physically getting all that work done? I think oftentimes investors think you can just push a button, and it gets done, but it's not that simple.
It is definitely not that simple. Considering there are four discrete projects, and they impact 800 locations, at minimum, there are 3,200 discrete field efforts that have to occur. So we signed an agreement to purchase the Sprint GMG network from T-Mobile in September of 2022. That deal closed in May of 2023. We began the planning for this reconfiguration at signing. So the original Sprint network was the country's first fiber optic network built to carry long distance. It was built between 1982 and 1989 at a capital cost of $20.5 billion. It was comprised of those fiber route miles that I've described and 482 telephone switch facilities, tandem and local switch facilities. We are repurposing 45 of those facilities as data centers.
We are implementing all of those interconnection and reconfiguration work projects in parallel to the reconfiguration of those endpoints. Cogent previously had 100 individuals in its field service organization. Sprint had 363 field service technicians. Most of those technicians are still at the company. Some have left. The average tenure of those employees was over 22 years. And some were due to retire, some were not necessarily suitable for this new effort. But the integration of the field service teams, in conjunction with third-party contractors, are performing approximately 400 work events a week across the network. Now, many sites require multiple passes or work efforts. You know, most of the equipment was equipment we had. We did have to order some supplemental equipment.
That was baked into the $50 million number that we laid out as a capital cost to do this repurposing exercise, but the majority of the effort is actually just physical field services work. We have a program in place to complete this work by the end of 2024. Today, we are about 40% of the way through that project.
All right, great. So talk to us about the typical profile of a wave customer. So what sort of applications are they putting on a wave? And why are they buying them versus direct access or dark fiber?
Yeah. So let me explain the wave market. 25 years ago, there were two suppliers, AT&T and MCI. The only use of waves was to link switches together to sell long-distance telephone service. That market disappeared about 20 years ago, and a second wave, or a second group, of wave buyers emerged, which were regional ISPs, who were building internet backbones on top of wavelengths. At that time, AT&T and MCI began to exit the market, and the predecessor companies of Lumen dominated. Williams, Global Crossing, Level 3, Broadwing, were the major providers. In the mid-2000s, most of those ISPs disappeared. The wavelength market today is dominated by three types of users. The first group are content publishers who are looking to replicate content at multiple data centers. So these could be hyperscalers, Amazon, Microsoft, Meta, Netflix.
They could be a third-party content distributor, like an Akamai, a Fastly, an EdgeCo, that are looking to distribute hosting companies, like a Rackspace, that we were chatting about a minute ago. The second group of buyers are isolated regional ISPs, who are linking together islands of traffic. So companies like Cox, Charter, or Spectrum, Comcast, MetroNet, Frontier, that have isolated markets that want to build a homogeneous network. Included in that are international carriers that look to extend their networks into the U.S. And the third and smallest group of buyers are large enterprise and governments who are looking to build networks independent of the internet. It is a relatively concentrated buyer universe, with about 200 companies dominating the wavelength market. Globally, that market is about $7 billion.
About $3.5 billion of it is in the US, $3.5 billion is either international or subsea. And then, of the $3.5 billion in the US, roughly $1.5 billion are metro waves. Those are wavelength or transport services staying within a given market, and $2 billion for intercity wavelength services. Cogent intends to focus on the intercity, not intra-city market.
Okay, great. So one of the things you've talked about, you've been pretty clear you don't intend to use pricing to get into this market. Talk to us about how your positioning is and how you're gonna view pricing as you enter the wave market.
So I may differ with your statement slightly, Frank, in that we will use pricing as necessary, but different than our entry into the transit market, where Cogent defined itself around its aggressive pricing strategy. So when we came to market to sell transit, the average price per megabit was $300, and our starting price was $10 a megabit, so we were at 3% of the market price. Over the subsequent 22 years, that average price has fallen to $0.10 for Cogent, and we remain the price leader with a 50% discount to our competitors. In selling transport or wavelength services, there are other qualitative attributes of the service that are more important to the buyers than just price. So when a customer is determining who to buy a wavelength from, they look at the endpoints.
Cogent will have more endpoints with wave services available than any other carrier. Today, the market is dominated by Lumen and Zayo. Each of them have about 350 endpoints for wavelength services, we will have 800. Secondly, the customer looks at the physical path between the two locations, and they value the uniqueness of the right of way. 90% of Cogent's wavelength network is along unique right of way. No other carrier shares it. So the Sprint network was constructed primarily along railroad right of way, with an armored cable buried 6 feet below the track. No other carrier is in that right of way. Most fiber today is deployed along public space right of way, highways, and plastic conduit 2 feet below grade.
So even though the Sprint fiber is, on average, 15 years older than the later generations of networks, it has had roughly 50% less cuts per mile than those networks. Those cuts are the primary reason for degradation in the quality of the fiber. Each time the fiber is spliced, there is some loss injected into the fiber, which degrades the quality of service. So our network, while older, actually has better characteristics. Third, our network was not a roll-up. It was built by one single company. We have extremely accurate GIS information, so we can provide a customer a map with each quote that is accurate within one meter of exactly where the fiber is. That's viewed as a huge advantage over our competitors. And then, fourth, we will provision that service more quickly.
The architecture we're deploying, and that foundational work I described earlier, will allow us to shorten the provisioning times. So in transit, Cogent has defined itself around price, but for the past 11 years, we have been the No. 1 transit provider, as measured by third parties in provisioning times globally. We intend to have that same kind of provisioning excellence in our wavelength product. And then finally, because we are doing this on a network that we have a $1 cost basis in, we bought that network that cost $20 billion to build, was appraised for $1 billion by KPMG, and we paid a $1 for it. We will use price. We won't lead with price, but I can guarantee investors we will not lose on price.
Okay, great. That's really helpful. So, so you've outlined... You've got a great network with extremely valuable rights of way that I think are really misunderstood and underappreciated and so forth. The typical play when we see telecom deals, and we haven't seen nearly as many as we used to, but the typical play there for the competitors is go around, go to all their customers before the deal closes or, you know, before you're really up and running, lock them up with new three- to five-year deals, and then it's really hard to sell into them. And then investors are shocked because the growth rate ramp doesn't appear because you have to wait for all those deals to roll off.
What went on in the background here leading up to the deal, and how do you feel about your competitive positioning, as you're entering the market?
We have been pleasantly surprised by the depth of demand for our network and the wavelength services we have. Historically, Cogent does not disclose sales funnel numbers, but to give investors an indication of this demand, we have, and will for the next couple of quarters, disclose that funnel. Eventually, all investors should care about is GAAP revenue, not sales funnels. But we have seen both inbound inquiries from customers, as well as outbound by our sales organization. So I think there are three points. The first one is, we have existing relationships with three-quarters of the buyers. We call on 100% of the buyers, and we have the largest dedicated sales force for that market, 257 sales professionals, of any carrier.
Secondly, wavelengths are typically purchased in either 1, 2, or 3-year contracts, meaning the average remaining life on those contracts is probably the midpoint to the longest of those three terms, or 18 months. Our deal was announced 14 months ago. Many customers value the diversity that we are offering. So no one buys just a single wavelength. You buy wavelengths typically to construct your own network and to have physical path diversity. That is not required on the internet, because the internet is dynamic and reroutes traffic automatically. A wavelength is deterministic, and if that physical path fails, it's gone. You can't get the traffic from point A to point B. So companies usually try to have 2 diverse paths between any two city pairs. The fact that we're on this unique right of way is extremely interesting to many customers.
The final point on our wavelength market acceptance is the track record that we've had in delivering transit. So these are customers that have trusted us and continue to buy services from us. The other kind of overlay to these three points is the fact that the unit volume continues to grow, the applications continue to evolve, prices on a per-bit basis continue to come down. So we can achieve our market share goals by capturing incremental growth, diversity requirements, and a portion of displacement. We do not anticipate running the current suppliers out of business. We believe that, just as we have done in transit, our value proposition is sufficiently compelling that we can get to a 25% market share. What's different this time is we've got a sales force, and we've got a brand, and we've got relationships.
While it took us 18 years to achieve 25% market share in transit, we project, based on our conversations with customers, that we can get to that 25% market share comparable to where we're at in transit in a 7-year period.
Okay, great. I got a question about the legacy part of the business, and then maybe stay and see if we got some questions from the audience. But talk to us about return to office and how that's impacted your corporate business, and what are your thoughts if return to office trends don't materially accelerate from here? What does that mean for the long-term value of your legacy commercial business?
So we sell dedicated internet access and VPN services in slightly over 1 billion sq ft of multi-tenant office space, concentrated in the central business districts of North American cities. That footprint was probably the most impacted footprint by the pandemic. Our corporate business had grown consistently at approximately 11% per year from 2005 to 2020. When the pandemic hit, that growth rate plunged to -9%, so the business was actually declining. Today, we're back to approximately a +1% growth rate. So we're about halfway back from where we were at the height of the pandemic. Companies are still figuring out what their return-to-office policy is going to be, what their real estate requirements are going to be, and what their IT infrastructure is going to look like. For Cogent, our corporate product is a fixed connection.
It is not a metered service like it is for our NetCentric customers. So a company buys it, and it's all you can eat. It is required for employees in the office. If they're in one day a week, one employee, they still need Internet connectivity. And then secondly, most of our customers also use their Internet connectivity from Cogent to operate their VPNs for their remote employees. Now, we have seen portions of the country return to pre-pandemic activity levels. Southeast pretty much looks like it did before the pandemic. The Pacific Northwest still looks like it's in the middle of the pandemic. So it is geographically uneven. The return-to-office trend is much slower than we expected.
We do anticipate continued improvement, and we will get back to that double-digit growth rate, particularly as companies make those architectural changes that they have delayed for the past several years as they were deciding what their office footprint. Virtually no companies, less than, far less than 1% of companies have decided to go completely virtual. So almost all companies are going to have an office presence, and if that exists, Cogent will sell them services because our corporate product delivers 30-60 times the throughput, it is typically installed 9 times faster, and it's 3 times more reliable, the same price point. The footprint we're in is the most desirable office space in the world. Now, vacancy rates have increased. They went from 6% up to 18%. They're only back to 17%, so we're still running 11% more vacancy than we were pre-pandemic.
But it's typically the B and C offices that bear the brunt of that vacancy. Rents come down in A, and those buildings tend to fill up. The final point is that for new leases signed in our footprint, the average new tenant lease is approximately 20% less square footage than the lease that exists in the building today. That ultimately bodes well for Cogent, because that will allow us to have a 20% larger addressable market for corporate products.
All right, one last question. You were a big proponent of Title II the last time it came around. FCC just recently passed another rulemaking on that. What are your thoughts on this current iteration of Title II?
So we remain absolutely supportive of net neutrality, and we think a first choice would be congressional legislation to clarify this. Absent that, which is unlikely in this political environment, the regulatory approach makes sense. The FCC decided to abdicate its authority over the internet when it changed the classification of internet services from Title I to Title... from Title II to Title I. As a result, 26 states stepped in, led by California. California net neutrality rules are actually more prescriptive and stricter than the federal rules. They today dominate. The federal reassertion of authority under Title II will not change behavior. It may actually result in slightly less restrictive laws, than the current California legislation. So in 2014, this became a major issue. Cogent was the primary carrier of internet traffic for Netflix.
Certain access networks were trying to block Netflix, and they constricted access to the internet, resulting in degradation of video transmissions. Now that 48% of all video is delivered over the internet, I think it would be commercially impossible for any provider to exercise that kind of degradation in traffic. So the takeaway here is, while we support the FCC's efforts, the reality is it's kind of belts and suspenders. It's already been fixed, both by California and the market.
All right, great. Thank you, Dave. Really appreciate it. Thanks for being here.
Hey, thanks, Frank.
All right.
Take care. Thank you, all.